3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Richard Heck (XHTML output)
9 * \author Jean-Marc Lasgouttes
10 * \author Angus Leeming
12 * \author André Pönitz
14 * \author Jürgen Vigna
16 * Full author contact details are available in file CREDITS.
21 #include "Paragraph.h"
23 #include "LayoutFile.h"
25 #include "BufferParams.h"
28 #include "BufferEncodings.h"
29 #include "InsetList.h"
31 #include "LaTeXFeatures.h"
37 #include "OutputParams.h"
38 #include "output_latex.h"
39 #include "output_xhtml.h"
40 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
43 #include "texstream.h"
44 #include "TextClass.h"
47 #include "WordLangTuple.h"
50 #include "frontends/alert.h"
52 #include "insets/InsetBibitem.h"
53 #include "insets/InsetLabel.h"
54 #include "insets/InsetSpecialChar.h"
56 #include "support/debug.h"
57 #include "support/docstring_list.h"
58 #include "support/ExceptionMessage.h"
59 #include "support/gettext.h"
60 #include "support/lassert.h"
61 #include "support/lstrings.h"
62 #include "support/textutils.h"
68 using namespace lyx::support;
74 /// Inset identifier (above 0x10ffff, for ucs-4)
75 char_type const META_INSET = 0x200001;
80 /////////////////////////////////////////////////////////////////////
84 /////////////////////////////////////////////////////////////////////
86 class SpellResultRange {
88 SpellResultRange(FontSpan range, SpellChecker::Result result)
89 : range_(range), result_(result)
92 FontSpan const & range() const { return range_; }
94 void range(FontSpan const & r) { range_ = r; }
96 SpellChecker::Result result() const { return result_; }
98 void result(SpellChecker::Result r) { result_ = r; }
100 bool contains(pos_type pos) const { return range_.contains(pos); }
102 bool covered(FontSpan const & r) const
104 // 1. first of new range inside current range or
105 // 2. last of new range inside current range or
106 // 3. first of current range inside new range or
107 // 4. last of current range inside new range
108 //FIXME: is this the same as !range_.intersect(r).empty() ?
109 return range_.contains(r.first) || range_.contains(r.last) ||
110 r.contains(range_.first) || r.contains(range_.last);
113 void shift(pos_type pos, int offset)
115 if (range_.first > pos) {
116 range_.first += offset;
117 range_.last += offset;
118 } else if (range_.last >= pos) {
119 range_.last += offset;
124 SpellChecker::Result result_ ;
128 /////////////////////////////////////////////////////////////////////
132 /////////////////////////////////////////////////////////////////////
134 class SpellCheckerState {
138 needs_refresh_ = true;
139 current_change_number_ = 0;
142 void setRange(FontSpan const & fp, SpellChecker::Result state)
145 RangesIterator et = ranges_.end();
146 RangesIterator it = ranges_.begin();
147 for (; it != et; ++it) {
148 if (!it->covered(fp))
149 result.push_back(SpellResultRange(it->range(), it->result()));
150 else if (state == SpellChecker::WORD_OK) {
151 // trim or split the current misspelled range
152 // store misspelled ranges only
153 FontSpan range = it->range();
154 if (fp.first > range.first) {
155 // misspelled area in front of WORD_OK
156 range.last = fp.first - 1;
157 result.push_back(SpellResultRange(range, it->result()));
160 if (fp.last < range.last) {
161 // misspelled area after WORD_OK range
162 range.first = fp.last + 1;
163 result.push_back(SpellResultRange(range, it->result()));
168 if (state != SpellChecker::WORD_OK)
169 ranges_.push_back(SpellResultRange(fp, state));
172 void increasePosAfterPos(pos_type pos)
174 correctRangesAfterPos(pos, 1);
178 void decreasePosAfterPos(pos_type pos)
180 correctRangesAfterPos(pos, -1);
184 void refreshLast(pos_type pos)
186 if (pos < refresh_.last)
190 SpellChecker::Result getState(pos_type pos) const
192 SpellChecker::Result result = SpellChecker::WORD_OK;
193 RangesIterator et = ranges_.end();
194 RangesIterator it = ranges_.begin();
195 for (; it != et; ++it) {
196 if(it->contains(pos)) {
203 FontSpan const & getRange(pos_type pos) const
205 /// empty span to indicate mismatch
206 static FontSpan empty_;
207 RangesIterator et = ranges_.end();
208 RangesIterator it = ranges_.begin();
209 for (; it != et; ++it) {
210 if(it->contains(pos)) {
217 bool needsRefresh() const
219 return needs_refresh_;
222 SpellChecker::ChangeNumber currentChangeNumber() const
224 return current_change_number_;
227 void refreshRange(pos_type & first, pos_type & last) const
229 first = refresh_.first;
230 last = refresh_.last;
233 void needsRefresh(pos_type pos)
235 if (needs_refresh_ && pos != -1) {
236 if (pos < refresh_.first)
237 refresh_.first = pos;
238 if (pos > refresh_.last)
240 } else if (pos != -1) {
241 // init request check for neighbour positions too
242 refresh_.first = pos > 0 ? pos - 1 : 0;
243 // no need for special end of paragraph check
244 refresh_.last = pos + 1;
246 needs_refresh_ = pos != -1;
249 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
251 needs_refresh_ = true;
254 current_change_number_ = change_number;
257 typedef vector<SpellResultRange> Ranges;
258 typedef Ranges::const_iterator RangesIterator;
260 /// the area of the paragraph with pending spell check
263 /// spell state cache version number
264 SpellChecker::ChangeNumber current_change_number_;
267 void correctRangesAfterPos(pos_type pos, int offset)
269 RangesIterator et = ranges_.end();
270 Ranges::iterator it = ranges_.begin();
271 for (; it != et; ++it) {
272 it->shift(pos, offset);
278 /////////////////////////////////////////////////////////////////////
280 // Paragraph::Private
282 /////////////////////////////////////////////////////////////////////
284 class Paragraph::Private
286 // Enforce our own "copy" constructor by declaring the standard one and
287 // the assignment operator private without implementing them.
288 Private(Private const &);
289 Private & operator=(Private const &);
292 Private(Paragraph * owner, Layout const & layout);
293 /// "Copy constructor"
294 Private(Private const &, Paragraph * owner);
295 /// Copy constructor from \p beg to \p end
296 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
299 void insertChar(pos_type pos, char_type c, Change const & change);
301 /// Output the surrogate pair formed by \p c and \p next to \p os.
302 /// \return the number of characters written.
303 int latexSurrogatePair(otexstream & os, char_type c, char_type next,
304 OutputParams const &);
306 /// Output a space in appropriate formatting (or a surrogate pair
307 /// if the next character is a combining character).
308 /// \return whether a surrogate pair was output.
309 bool simpleTeXBlanks(OutputParams const &,
312 unsigned int & column,
314 Layout const & style);
316 /// Output consecutive unicode chars, belonging to the same script as
317 /// specified by the latex macro \p ltx, to \p os starting from \p i.
318 /// \return the number of characters written.
319 int writeScriptChars(otexstream & os, docstring const & ltx,
320 Change const &, Encoding const &, pos_type & i);
322 /// This could go to ParagraphParameters if we want to.
323 int startTeXParParams(BufferParams const &, otexstream &,
324 OutputParams const &) const;
326 /// This could go to ParagraphParameters if we want to.
327 bool endTeXParParams(BufferParams const &, otexstream &,
328 OutputParams const &) const;
331 void latexInset(BufferParams const &,
336 Font const & outerfont,
338 Change & running_change,
339 Layout const & style,
341 unsigned int & column);
344 void latexSpecialChar(
346 BufferParams const & bparams,
347 OutputParams const & runparams,
348 Font const & running_font,
349 Change const & running_change,
350 Layout const & style,
353 unsigned int & column);
360 unsigned int & column);
366 unsigned int & column);
369 void validate(LaTeXFeatures & features) const;
371 /// Checks if the paragraph contains only text and no inset or font change.
372 bool onlyText(Buffer const & buf, Font const & outerfont,
373 pos_type initial) const;
375 /// a vector of speller skip positions
376 typedef vector<FontSpan> SkipPositions;
377 typedef SkipPositions::const_iterator SkipPositionsIterator;
379 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
381 Language * getSpellLanguage(pos_type const from) const;
383 Language * locateSpellRange(pos_type & from, pos_type & to,
384 SkipPositions & skips) const;
386 bool hasSpellerChange() const
388 SpellChecker::ChangeNumber speller_change_number = 0;
389 if (theSpellChecker())
390 speller_change_number = theSpellChecker()->changeNumber();
391 return speller_change_number > speller_state_.currentChangeNumber();
394 bool ignoreWord(docstring const & word) const ;
396 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
398 pos_type textsize = owner_->size();
399 // check for sane arguments
400 if (to <= from || from >= textsize)
402 FontSpan fp = FontSpan(from, to - 1);
403 speller_state_.setRange(fp, state);
406 void requestSpellCheck(pos_type pos)
409 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
411 speller_state_.needsRefresh(pos);
414 void readySpellCheck()
416 speller_state_.needsRefresh(-1);
419 bool needsSpellCheck() const
421 return speller_state_.needsRefresh();
424 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
426 speller_state_.refreshRange(first, last);
428 last = owner_->size();
431 pos_type endpos = last;
432 owner_->locateWord(first, endpos, WHOLE_WORD);
435 owner_->locateWord(last, endpos, WHOLE_WORD);
440 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
444 while (it != et && it->first < start) {
445 int skip = it->last - it->first + 1;
453 void markMisspelledWords(pos_type const & first, pos_type const & last,
454 SpellChecker::Result result,
455 docstring const & word,
456 SkipPositions const & skips);
458 InsetCode ownerCode() const
460 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
463 /// Which Paragraph owns us?
467 Inset const * inset_owner_;
476 ParagraphParameters params_;
478 /// for recording and looking up changes
482 InsetList insetlist_;
485 pos_type begin_of_body_;
487 typedef docstring TextContainer;
491 typedef set<docstring> Words;
492 typedef map<string, Words> LangWordsMap;
496 Layout const * layout_;
498 SpellCheckerState speller_state_;
502 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
503 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
509 // Initialization of the counter for the paragraph id's,
511 // FIXME: There should be a more intelligent way to generate and use the
512 // paragraph ids per buffer instead a global static counter for all InsetText
513 // in the running program.
514 // However, this per-session id is used in LFUN_PARAGRAPH_GOTO to
515 // switch to a different buffer, as used in the outliner for instance.
516 static int paragraph_id = -1;
518 Paragraph::Private::Private(Private const & p, Paragraph * owner)
519 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
520 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
521 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
524 id_ = ++paragraph_id;
525 requestSpellCheck(p.text_.size());
529 Paragraph::Private::Private(Private const & p, Paragraph * owner,
530 pos_type beg, pos_type end)
531 : owner_(owner), inset_owner_(p.inset_owner_),
532 params_(p.params_), changes_(p.changes_),
533 insetlist_(p.insetlist_, beg, end),
534 begin_of_body_(p.begin_of_body_), words_(p.words_),
537 id_ = ++paragraph_id;
538 if (beg >= pos_type(p.text_.size()))
540 text_ = p.text_.substr(beg, end - beg);
542 FontList::const_iterator fcit = fontlist_.begin();
543 FontList::const_iterator fend = fontlist_.end();
544 for (; fcit != fend; ++fcit) {
545 if (fcit->pos() < beg)
547 if (fcit->pos() >= end) {
548 // Add last entry in the fontlist_.
549 fontlist_.set(text_.size() - 1, fcit->font());
552 // Add a new entry in the fontlist_.
553 fontlist_.set(fcit->pos() - beg, fcit->font());
555 requestSpellCheck(p.text_.size());
559 void Paragraph::addChangesToToc(DocIterator const & cdit,
560 Buffer const & buf, bool output_active) const
562 d->changes_.addToToc(cdit, buf, output_active);
566 void Paragraph::addChangesToBuffer(Buffer const & buf) const
568 d->changes_.updateBuffer(buf);
572 bool Paragraph::isChangeUpdateRequired() const
574 return d->changes_.isUpdateRequired();
578 bool Paragraph::isDeleted(pos_type start, pos_type end) const
580 LASSERT(start >= 0 && start <= size(), return false);
581 LASSERT(end > start && end <= size() + 1, return false);
583 return d->changes_.isDeleted(start, end);
587 bool Paragraph::isChanged(pos_type start, pos_type end) const
589 LASSERT(start >= 0 && start <= size(), return false);
590 LASSERT(end > start && end <= size() + 1, return false);
592 return d->changes_.isChanged(start, end);
596 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
598 // keep the logic here in sync with the logic of eraseChars()
602 Change const & change = d->changes_.lookup(size());
603 return change.inserted() && change.currentAuthor();
607 void Paragraph::setChange(Change const & change)
609 // beware of the imaginary end-of-par character!
610 d->changes_.set(change, 0, size() + 1);
613 * Propagate the change recursively - but not in case of DELETED!
615 * Imagine that your co-author makes changes in an existing inset. He
616 * sends your document to you and you come to the conclusion that the
617 * inset should go completely. If you erase it, LyX must not delete all
618 * text within the inset. Otherwise, the change tracked insertions of
619 * your co-author get lost and there is no way to restore them later.
621 * Conclusion: An inset's content should remain untouched if you delete it
624 if (!change.deleted()) {
625 for (pos_type pos = 0; pos < size(); ++pos) {
626 if (Inset * inset = getInset(pos))
627 inset->setChange(change);
633 void Paragraph::setChange(pos_type pos, Change const & change)
635 LASSERT(pos >= 0 && pos <= size(), return);
636 d->changes_.set(change, pos);
638 // see comment in setChange(Change const &) above
639 if (!change.deleted() && pos < size())
640 if (Inset * inset = getInset(pos))
641 inset->setChange(change);
645 Change const & Paragraph::lookupChange(pos_type pos) const
647 LBUFERR(pos >= 0 && pos <= size());
648 return d->changes_.lookup(pos);
652 void Paragraph::acceptChanges(pos_type start, pos_type end)
654 LASSERT(start >= 0 && start <= size(), return);
655 LASSERT(end > start && end <= size() + 1, return);
657 for (pos_type pos = start; pos < end; ++pos) {
658 switch (lookupChange(pos).type) {
659 case Change::UNCHANGED:
660 // accept changes in nested inset
661 if (Inset * inset = getInset(pos))
662 inset->acceptChanges();
665 case Change::INSERTED:
666 d->changes_.set(Change(Change::UNCHANGED), pos);
667 // also accept changes in nested inset
668 if (Inset * inset = getInset(pos))
669 inset->acceptChanges();
672 case Change::DELETED:
673 // Suppress access to non-existent
674 // "end-of-paragraph char"
676 eraseChar(pos, false);
687 void Paragraph::rejectChanges(pos_type start, pos_type end)
689 LASSERT(start >= 0 && start <= size(), return);
690 LASSERT(end > start && end <= size() + 1, return);
692 for (pos_type pos = start; pos < end; ++pos) {
693 switch (lookupChange(pos).type) {
694 case Change::UNCHANGED:
695 // reject changes in nested inset
696 if (Inset * inset = getInset(pos))
697 inset->rejectChanges();
700 case Change::INSERTED:
701 // Suppress access to non-existent
702 // "end-of-paragraph char"
704 eraseChar(pos, false);
710 case Change::DELETED:
711 d->changes_.set(Change(Change::UNCHANGED), pos);
713 // Do NOT reject changes within a deleted inset!
714 // There may be insertions of a co-author inside of it!
722 void Paragraph::Private::insertChar(pos_type pos, char_type c,
723 Change const & change)
725 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
728 changes_.insert(change, pos);
730 // This is actually very common when parsing buffers (and
731 // maybe inserting ascii text)
732 if (pos == pos_type(text_.size())) {
733 // when appending characters, no need to update tables
735 // but we want spell checking
736 requestSpellCheck(pos);
740 text_.insert(text_.begin() + pos, c);
742 // Update the font table.
743 fontlist_.increasePosAfterPos(pos);
746 insetlist_.increasePosAfterPos(pos);
748 // Update list of misspelled positions
749 speller_state_.increasePosAfterPos(pos);
753 bool Paragraph::insertInset(pos_type pos, Inset * inset,
754 Font const & font, Change const & change)
756 LASSERT(inset, return false);
757 LASSERT(pos >= 0 && pos <= size(), return false);
759 // Paragraph::insertInset() can be used in cut/copy/paste operation where
760 // d->inset_owner_ is not set yet.
761 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
764 d->insertChar(pos, META_INSET, change);
765 LASSERT(d->text_[pos] == META_INSET, return false);
767 // Add a new entry in the insetlist_.
768 d->insetlist_.insert(inset, pos);
770 // Some insets require run of spell checker
771 requestSpellCheck(pos);
777 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
779 LASSERT(pos >= 0 && pos <= size(), return false);
781 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
784 Change change = d->changes_.lookup(pos);
786 // set the character to DELETED if
787 // a) it was previously unchanged or
788 // b) it was inserted by a co-author
790 if (!change.changed() ||
791 (change.inserted() && !change.currentAuthor())) {
792 setChange(pos, Change(Change::DELETED));
793 // request run of spell checker
794 requestSpellCheck(pos);
798 if (change.deleted())
802 // Don't physically access the imaginary end-of-paragraph character.
803 // eraseChar() can only mark it as DELETED. A physical deletion of
804 // end-of-par must be handled externally.
810 d->changes_.erase(pos);
812 // if it is an inset, delete the inset entry
813 if (d->text_[pos] == META_INSET)
814 d->insetlist_.erase(pos);
816 d->text_.erase(d->text_.begin() + pos);
818 // Update the fontlist_
819 d->fontlist_.erase(pos);
821 // Update the insetlist_
822 d->insetlist_.decreasePosAfterPos(pos);
824 // Update list of misspelled positions
825 d->speller_state_.decreasePosAfterPos(pos);
826 d->speller_state_.refreshLast(size());
832 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
834 LASSERT(start >= 0 && start <= size(), return 0);
835 LASSERT(end >= start && end <= size() + 1, return 0);
838 for (pos_type count = end - start; count; --count) {
839 if (!eraseChar(i, trackChanges))
846 int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
847 char_type next, OutputParams const & runparams)
849 // Writing next here may circumvent a possible font change between
850 // c and next. Since next is only output if it forms a surrogate pair
851 // with c we can ignore this:
852 // A font change inside a surrogate pair does not make sense and is
853 // hopefully impossible to input.
854 // FIXME: change tracking
855 // Is this correct WRT change tracking?
856 Encoding const & encoding = *(runparams.encoding);
857 docstring latex1 = encoding.latexChar(next).first;
858 if (runparams.inIPA) {
859 string const tipashortcut = Encodings::TIPAShortcut(next);
860 if (!tipashortcut.empty()) {
861 latex1 = from_ascii(tipashortcut);
864 docstring const latex2 = encoding.latexChar(c).first;
865 if (docstring(1, next) == latex1) {
866 // the encoding supports the combination
867 os << latex2 << latex1;
868 return latex1.length() + latex2.length();
869 } else if (runparams.local_font &&
870 runparams.local_font->language()->lang() == "polutonikogreek") {
871 // polutonikogreek only works without the brackets
872 os << latex1 << latex2;
873 return latex1.length() + latex2.length();
875 os << latex1 << '{' << latex2 << '}';
876 return latex1.length() + latex2.length() + 2;
880 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
883 unsigned int & column,
885 Layout const & style)
887 if (style.pass_thru || runparams.pass_thru)
890 if (i + 1 < int(text_.size())) {
891 char_type next = text_[i + 1];
892 if (Encodings::isCombiningChar(next)) {
893 // This space has an accent, so we must always output it.
894 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
899 if (runparams.linelen > 0
900 && column > runparams.linelen
902 && text_[i - 1] != ' '
903 && (i + 1 < int(text_.size()))
904 // same in FreeSpacing mode
905 && !owner_->isFreeSpacing()
906 // In typewriter mode, we want to avoid
907 // ! . ? : at the end of a line
908 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
909 && (text_[i - 1] == '.'
910 || text_[i - 1] == '?'
911 || text_[i - 1] == ':'
912 || text_[i - 1] == '!'))) {
914 os.texrow().start(owner_->id(), i + 1);
916 } else if (style.free_spacing) {
925 int Paragraph::Private::writeScriptChars(otexstream & os,
926 docstring const & ltx,
927 Change const & runningChange,
928 Encoding const & encoding,
931 // FIXME: modifying i here is not very nice...
933 // We only arrive here when character text_[i] could not be translated
934 // into the current latex encoding (or its latex translation has been forced,)
935 // and it belongs to a known script.
936 // TODO: We need \textcyr and \textgreek wrappers also for characters
937 // that can be encoded in the "LaTeX encoding" but not in the
938 // current *font encoding*.
939 // (See #9681 for details and test)
940 // Parameter ltx contains the latex translation of text_[i] as specified
941 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
942 // The latex macro name "textXXX" specifies the script to which text_[i]
943 // belongs and we use it in order to check whether characters from the
944 // same script immediately follow, such that we can collect them in a
945 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
946 // for the first char but only "<spec>" for all subsequent chars.
947 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
948 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
949 string script = to_ascii(ltx.substr(1, brace1 - 1));
952 bool closing_brace = true;
953 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
954 // Correct encoding is being used, so we can avoid \textgreek.
955 // TODO: wrong test: we need to check the *font encoding*
956 // (i.e. the active language and its FontEncoding tag)
957 // instead of the LaTeX *input encoding*!
958 // See #9637 for details and test-cases.
961 closing_brace = false;
963 os << ltx.substr(pos, length);
964 int size = text_.size();
965 while (i + 1 < size) {
966 char_type const next = text_[i + 1];
967 // Stop here if next character belongs to another script
968 // or there is a change in change tracking status.
969 if (!Encodings::isKnownScriptChar(next, script) ||
970 runningChange != owner_->lookupChange(i + 1))
974 FontList::const_iterator cit = fontlist_.begin();
975 FontList::const_iterator end = fontlist_.end();
976 for (; cit != end; ++cit) {
977 if (cit->pos() >= i && !found) {
978 prev_font = cit->font();
981 if (cit->pos() >= i + 1)
984 // Stop here if there is a font attribute or encoding change.
985 if (found && cit != end && prev_font != cit->font())
987 docstring const latex = encoding.latexChar(next).first;
988 docstring::size_type const b1 =
989 latex.find_first_of(from_ascii("{"));
990 docstring::size_type const b2 =
991 latex.find_last_of(from_ascii("}"));
992 int const len = b2 - b1 - 1;
993 os << latex.substr(b1 + 1, len);
1005 void Paragraph::Private::latexInset(BufferParams const & bparams,
1007 OutputParams & runparams,
1008 Font & running_font,
1010 Font const & outerfont,
1012 Change & running_change,
1013 Layout const & style,
1015 unsigned int & column)
1017 Inset * inset = owner_->getInset(i);
1020 if (style.pass_thru) {
1021 odocstringstream ods;
1022 inset->plaintext(ods, runparams);
1027 // FIXME: move this to InsetNewline::latex
1028 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1029 // newlines are handled differently here than
1030 // the default in simpleTeXSpecialChars().
1031 if (!style.newline_allowed) {
1035 column += running_font.latexWriteEndChanges(
1036 os, bparams, runparams,
1037 basefont, basefont);
1041 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1044 basefont = owner_->getLayoutFont(bparams, outerfont);
1045 running_font = basefont;
1047 if (runparams.moving_arg)
1051 os.texrow().start(owner_->id(), i + 1);
1055 if (owner_->isDeleted(i)) {
1056 if( ++runparams.inDeletedInset == 1)
1057 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1060 if (inset->canTrackChanges()) {
1061 column += Changes::latexMarkChange(os, bparams, running_change,
1062 Change(Change::UNCHANGED), runparams);
1063 running_change = Change(Change::UNCHANGED);
1067 odocstream::pos_type const len = os.os().tellp();
1069 if (inset->forceLTR()
1070 && !runparams.use_polyglossia
1071 && running_font.isRightToLeft()
1072 // ERT is an exception, it should be output with no
1073 // decorations at all
1074 && inset->lyxCode() != ERT_CODE) {
1075 if (running_font.language()->lang() == "farsi")
1082 // FIXME: Bug: we can have an empty font change here!
1083 // if there has just been a font change, we are going to close it
1084 // right now, which means stupid latex code like \textsf{}. AFAIK,
1085 // this does not harm dvi output. A minor bug, thus (JMarc)
1087 // Some insets cannot be inside a font change command.
1088 // However, even such insets *can* be placed in \L or \R
1089 // or their equivalents (for RTL language switches), so we don't
1090 // close the language in those cases.
1091 // ArabTeX, though, cannot handle this special behavior, it seems.
1092 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1093 || running_font.language()->lang() == "arabic_arabtex";
1094 if (open_font && !inset->inheritFont()) {
1095 bool closeLanguage = arabtex
1096 || basefont.isRightToLeft() == running_font.isRightToLeft();
1097 unsigned int count = running_font.latexWriteEndChanges(os,
1098 bparams, runparams, basefont, basefont, closeLanguage);
1100 // if any font properties were closed, update the running_font,
1101 // making sure, however, to leave the language as it was
1103 // FIXME: probably a better way to keep track of the old
1104 // language, than copying the entire font?
1105 Font const copy_font(running_font);
1106 basefont = owner_->getLayoutFont(bparams, outerfont);
1107 running_font = basefont;
1109 running_font.setLanguage(copy_font.language());
1110 // leave font open if language is still open
1111 open_font = (running_font.language() == basefont.language());
1113 runparams.local_font = &basefont;
1117 int prev_rows = os.texrow().rows();
1120 runparams.lastid = id_;
1121 runparams.lastpos = i;
1122 inset->latex(os, runparams);
1123 } catch (EncodingException & e) {
1124 // add location information and throw again.
1131 if (running_font.language()->lang() == "farsi")
1137 if (os.texrow().rows() > prev_rows) {
1138 os.texrow().start(owner_->id(), i + 1);
1141 column += (unsigned int)(os.os().tellp() - len);
1144 if (owner_->isDeleted(i))
1145 --runparams.inDeletedInset;
1149 void Paragraph::Private::latexSpecialChar(otexstream & os,
1150 BufferParams const & bparams,
1151 OutputParams const & runparams,
1152 Font const & running_font,
1153 Change const & running_change,
1154 Layout const & style,
1157 unsigned int & column)
1159 // With polyglossia, brackets and stuff need not be reversed
1160 // in RTL scripts (see bug #8251)
1161 char_type const c = (runparams.use_polyglossia) ?
1162 owner_->getUChar(bparams, i) : text_[i];
1164 if (style.pass_thru || runparams.pass_thru
1165 || contains(style.pass_thru_chars, c)
1166 || contains(runparams.pass_thru_chars, c)) {
1168 Encoding const * const enc = runparams.encoding;
1169 if (enc && !enc->encodable(c))
1170 throw EncodingException(c);
1176 // TIPA uses its own T3 encoding
1177 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1179 // If T1 font encoding is used, use the special
1180 // characters it provides.
1181 // NOTE: Some languages reset the font encoding internally to a
1182 // non-standard font encoding. If we are using such a language,
1183 // we do not output special T1 chars.
1184 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1185 && bparams.font_encoding() == "T1" && latexSpecialT1(c, os, i, column))
1188 // Otherwise, we use what LaTeX provides us.
1191 os << "\\textbackslash{}";
1195 os << "\\textless{}";
1199 os << "\\textgreater{}";
1203 os << "\\textbar{}";
1208 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1209 (end_pos == -1 || i + 1 < end_pos) &&
1210 text_[i+1] == '-') {
1211 // Prevent "--" becoming an endash and "---" becoming
1213 // Within \ttfamily, "--" is merged to "-" (no endash)
1214 // so we avoid this rather irritating ligature as well
1220 os << "\\char`\\\"{}";
1225 case '%': case '#': case '{':
1233 os << "\\textasciitilde{}";
1238 os << "\\textasciicircum{}";
1245 // avoid being mistaken for optional arguments
1253 // Blanks are printed before font switching.
1254 // Sure? I am not! (try nice-latex)
1255 // I am sure it's correct. LyX might be smarter
1256 // in the future, but for now, nothing wrong is
1264 Encoding const & encoding = *(runparams.encoding);
1265 char_type next = '\0';
1266 if (i + 1 < int(text_.size())) {
1267 next = text_[i + 1];
1268 if (Encodings::isCombiningChar(next)) {
1269 column += latexSurrogatePair(os, c, next, runparams) - 1;
1275 pair<docstring, bool> latex = encoding.latexChar(c);
1276 docstring nextlatex;
1277 bool nexttipas = false;
1278 string nexttipashortcut;
1279 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1280 nextlatex = encoding.latexChar(next).first;
1281 if (runparams.inIPA) {
1282 nexttipashortcut = Encodings::TIPAShortcut(next);
1283 nexttipas = !nexttipashortcut.empty();
1287 if (runparams.inIPA) {
1288 string const tipashortcut = Encodings::TIPAShortcut(c);
1289 if (!tipashortcut.empty()) {
1290 latex.first = from_ascii(tipashortcut);
1291 latex.second = false;
1295 if (Encodings::isKnownScriptChar(c, script)
1296 && prefixIs(latex.first, from_ascii("\\" + script)))
1297 column += writeScriptChars(os, latex.first,
1298 running_change, encoding, i) - 1;
1299 else if (latex.second
1300 && ((!prefixIs(nextlatex, '\\')
1301 && !prefixIs(nextlatex, '{')
1302 && !prefixIs(nextlatex, '}'))
1304 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1306 // Prevent eating of a following
1307 // space or command corruption by
1308 // following characters
1309 if (next == ' ' || next == '\0') {
1310 column += latex.first.length() + 1;
1311 os << latex.first << "{}";
1313 column += latex.first.length();
1314 os << latex.first << " ";
1317 column += latex.first.length() - 1;
1325 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1326 pos_type i, unsigned int & column)
1332 // In T1 encoding, these characters exist
1333 // but we should avoid ligatures
1334 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1336 os << "\\textcompwordmark{}";
1343 // soul.sty breaks with \char`\"
1344 os << "\\textquotedbl{}";
1353 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1354 pos_type /*i*/, unsigned int & column)
1364 os << "\\textvertline{}";
1373 void Paragraph::Private::validate(LaTeXFeatures & features) const
1375 if (layout_->inpreamble && inset_owner_) {
1376 bool const is_command = layout_->latextype == LATEX_COMMAND;
1377 Buffer const & buf = inset_owner_->buffer();
1378 BufferParams const & bp = features.runparams().is_child
1379 ? 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, false);
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 bool start_paragraph, bool close_paragraph,
2757 pos_type initial) const
2761 // track whether we have opened these tags
2762 bool emph_flag = false;
2763 bool bold_flag = false;
2764 bool noun_flag = false;
2765 bool ubar_flag = false;
2766 bool dbar_flag = false;
2767 bool sout_flag = false;
2768 bool wave_flag = false;
2770 bool shap_flag = false;
2772 bool faml_flag = false;
2774 bool size_flag = false;
2776 Layout const & style = *d->layout_;
2778 if (start_paragraph)
2779 xs.startDivision(allowEmpty());
2782 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2784 FontShape curr_fs = INHERIT_SHAPE;
2785 FontFamily curr_fam = INHERIT_FAMILY;
2786 FontSize curr_size = FONT_SIZE_INHERIT;
2788 string const default_family =
2789 buf.masterBuffer()->params().fonts_default_family;
2791 vector<html::FontTag> tagsToOpen;
2792 vector<html::EndFontTag> tagsToClose;
2794 // parsing main loop
2795 for (pos_type i = initial; i < size(); ++i) {
2796 // let's not show deleted material in the output
2800 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
2803 FontState curstate = font.fontInfo().emph();
2804 if (font_old.emph() != curstate)
2805 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
2808 curstate = font.fontInfo().noun();
2809 if (font_old.noun() != curstate)
2810 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
2813 curstate = font.fontInfo().underbar();
2814 if (font_old.underbar() != curstate)
2815 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
2818 curstate = font.fontInfo().strikeout();
2819 if (font_old.strikeout() != curstate)
2820 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
2823 curstate = font.fontInfo().uuline();
2824 if (font_old.uuline() != curstate)
2825 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
2828 curstate = font.fontInfo().uwave();
2829 if (font_old.uwave() != curstate)
2830 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
2833 // a little hackish, but allows us to reuse what we have.
2834 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
2835 if (font_old.series() != font.fontInfo().series())
2836 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
2839 curr_fs = font.fontInfo().shape();
2840 FontShape old_fs = font_old.shape();
2841 if (old_fs != curr_fs) {
2845 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
2848 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
2850 case SMALLCAPS_SHAPE:
2851 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
2857 // the other tags are for internal use
2865 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
2869 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
2872 case SMALLCAPS_SHAPE:
2873 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
2880 // the other tags are for internal use
2887 curr_fam = font.fontInfo().family();
2888 FontFamily old_fam = font_old.family();
2889 if (old_fam != curr_fam) {
2893 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
2896 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
2898 case TYPEWRITER_FAMILY:
2899 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
2901 case INHERIT_FAMILY:
2904 // the other tags are for internal use
2912 // we will treat a "default" font family as roman, since we have
2913 // no other idea what to do.
2914 if (default_family != "rmdefault" && default_family != "default") {
2915 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
2920 if (default_family != "sfdefault") {
2921 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
2925 case TYPEWRITER_FAMILY:
2926 if (default_family != "ttdefault") {
2927 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
2931 case INHERIT_FAMILY:
2934 // the other tags are for internal use
2941 curr_size = font.fontInfo().size();
2942 FontSize old_size = font_old.size();
2943 if (old_size != curr_size) {
2946 case FONT_SIZE_TINY:
2947 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
2949 case FONT_SIZE_SCRIPT:
2950 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
2952 case FONT_SIZE_FOOTNOTE:
2953 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
2955 case FONT_SIZE_SMALL:
2956 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
2958 case FONT_SIZE_LARGE:
2959 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
2961 case FONT_SIZE_LARGER:
2962 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
2964 case FONT_SIZE_LARGEST:
2965 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
2967 case FONT_SIZE_HUGE:
2968 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
2970 case FONT_SIZE_HUGER:
2971 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
2973 case FONT_SIZE_INCREASE:
2974 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
2976 case FONT_SIZE_DECREASE:
2977 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
2979 case FONT_SIZE_INHERIT:
2980 case FONT_SIZE_NORMAL:
2983 // the other tags are for internal use
2989 switch (curr_size) {
2990 case FONT_SIZE_TINY:
2991 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
2994 case FONT_SIZE_SCRIPT:
2995 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
2998 case FONT_SIZE_FOOTNOTE:
2999 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3002 case FONT_SIZE_SMALL:
3003 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3006 case FONT_SIZE_LARGE:
3007 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3010 case FONT_SIZE_LARGER:
3011 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3014 case FONT_SIZE_LARGEST:
3015 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3018 case FONT_SIZE_HUGE:
3019 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3022 case FONT_SIZE_HUGER:
3023 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3026 case FONT_SIZE_INCREASE:
3027 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3030 case FONT_SIZE_DECREASE:
3031 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3034 case FONT_SIZE_NORMAL:
3035 case FONT_SIZE_INHERIT:
3038 // the other tags are for internal use
3045 // Other such tags? What about the other text ranges?
3047 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3048 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3049 for (; cit != cen; ++cit)
3052 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3053 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3054 for (; sit != sen; ++sit)
3057 tagsToClose.clear();
3060 Inset const * inset = getInset(i);
3062 if (!runparams.for_toc || inset->isInToc()) {
3063 OutputParams np = runparams;
3064 np.local_font = &font;
3065 // If the paragraph has size 1, then we are in the "special
3066 // case" where we do not output the containing paragraph info
3067 if (!inset->getLayout().htmlisblock() && size() != 1)
3068 np.html_in_par = true;
3069 retval += inset->xhtml(xs, np);
3072 char_type c = getUChar(buf.masterBuffer()->params(), i);
3075 font_old = font.fontInfo();
3079 // I'm worried about what happens if a branch, say, is itself
3080 // wrapped in some font stuff. I think that will not work.
3082 if (close_paragraph)
3089 bool Paragraph::isHfill(pos_type pos) const
3091 Inset const * inset = getInset(pos);
3092 return inset && inset->isHfill();
3096 bool Paragraph::isNewline(pos_type pos) const
3098 Inset const * inset = getInset(pos);
3099 return inset && inset->lyxCode() == NEWLINE_CODE;
3103 bool Paragraph::isEnvSeparator(pos_type pos) const
3105 Inset const * inset = getInset(pos);
3106 return inset && inset->lyxCode() == SEPARATOR_CODE;
3110 bool Paragraph::isLineSeparator(pos_type pos) const
3112 char_type const c = d->text_[pos];
3113 if (isLineSeparatorChar(c))
3115 Inset const * inset = getInset(pos);
3116 return inset && inset->isLineSeparator();
3120 bool Paragraph::isWordSeparator(pos_type pos) const
3124 if (Inset const * inset = getInset(pos))
3125 return !inset->isLetter();
3126 // if we have a hard hyphen (no en- or emdash) or apostrophe
3127 // we pass this to the spell checker
3128 // FIXME: this method is subject to change, visit
3129 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3130 // to get an impression how complex this is.
3131 if (isHardHyphenOrApostrophe(pos))
3133 char_type const c = d->text_[pos];
3134 // We want to pass the escape chars to the spellchecker
3135 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3136 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3140 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3142 pos_type const psize = size();
3145 char_type const c = d->text_[pos];
3146 if (c != '-' && c != '\'')
3148 int nextpos = pos + 1;
3149 int prevpos = pos > 0 ? pos - 1 : 0;
3150 if ((nextpos == psize || isSpace(nextpos))
3151 && (pos == 0 || isSpace(prevpos)))
3157 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3159 return d->speller_state_.getRange(pos);
3163 bool Paragraph::isChar(pos_type pos) const
3165 if (Inset const * inset = getInset(pos))
3166 return inset->isChar();
3167 char_type const c = d->text_[pos];
3168 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3172 bool Paragraph::isSpace(pos_type pos) const
3174 if (Inset const * inset = getInset(pos))
3175 return inset->isSpace();
3176 char_type const c = d->text_[pos];
3177 return lyx::isSpace(c);
3182 Paragraph::getParLanguage(BufferParams const & bparams) const
3185 return getFirstFontSettings(bparams).language();
3186 // FIXME: we should check the prev par as well (Lgb)
3187 return bparams.language;
3191 bool Paragraph::isRTL(BufferParams const & bparams) const
3193 return getParLanguage(bparams)->rightToLeft()
3194 && !inInset().getLayout().forceLTR();
3198 void Paragraph::changeLanguage(BufferParams const & bparams,
3199 Language const * from, Language const * to)
3201 // change language including dummy font change at the end
3202 for (pos_type i = 0; i <= size(); ++i) {
3203 Font font = getFontSettings(bparams, i);
3204 if (font.language() == from) {
3205 font.setLanguage(to);
3207 d->requestSpellCheck(i);
3213 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3215 Language const * doc_language = bparams.language;
3216 FontList::const_iterator cit = d->fontlist_.begin();
3217 FontList::const_iterator end = d->fontlist_.end();
3219 for (; cit != end; ++cit)
3220 if (cit->font().language() != ignore_language &&
3221 cit->font().language() != latex_language &&
3222 cit->font().language() != doc_language)
3228 void Paragraph::getLanguages(std::set<Language const *> & languages) const
3230 FontList::const_iterator cit = d->fontlist_.begin();
3231 FontList::const_iterator end = d->fontlist_.end();
3233 for (; cit != end; ++cit) {
3234 Language const * lang = cit->font().language();
3235 if (lang != ignore_language &&
3236 lang != latex_language)
3237 languages.insert(lang);
3242 docstring Paragraph::asString(int options) const
3244 return asString(0, size(), options);
3248 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3250 odocstringstream os;
3253 && options & AS_STR_LABEL
3254 && !d->params_.labelString().empty())
3255 os << d->params_.labelString() << ' ';
3257 for (pos_type i = beg; i < end; ++i) {
3258 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3260 char_type const c = d->text_[i];
3261 if (isPrintable(c) || c == '\t'
3262 || (c == '\n' && (options & AS_STR_NEWLINES)))
3264 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3265 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3266 LASSERT(runparams != 0, return docstring());
3267 getInset(i)->plaintext(os, *runparams);
3269 getInset(i)->toString(os);
3278 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3279 bool const shorten) const
3281 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3282 if (!d->params_.labelString().empty())
3283 os += d->params_.labelString() + ' ';
3284 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3287 char_type const c = d->text_[i];
3290 else if (c == META_INSET)
3291 getInset(i)->forOutliner(os, tmplen, false);
3294 Text::shortenForOutliner(os, maxlen);
3298 void Paragraph::setInsetOwner(Inset const * inset)
3300 d->inset_owner_ = inset;
3304 int Paragraph::id() const
3310 void Paragraph::setId(int id)
3316 Layout const & Paragraph::layout() const
3322 void Paragraph::setLayout(Layout const & layout)
3324 d->layout_ = &layout;
3328 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3330 setLayout(tc.defaultLayout());
3334 void Paragraph::setPlainLayout(DocumentClass const & tc)
3336 setLayout(tc.plainLayout());
3340 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3342 if (usePlainLayout())
3343 setPlainLayout(tclass);
3345 setDefaultLayout(tclass);
3349 Inset const & Paragraph::inInset() const
3351 LBUFERR(d->inset_owner_);
3352 return *d->inset_owner_;
3356 ParagraphParameters & Paragraph::params()
3362 ParagraphParameters const & Paragraph::params() const
3368 bool Paragraph::isFreeSpacing() const
3370 if (d->layout_->free_spacing)
3372 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3376 bool Paragraph::allowEmpty() const
3378 if (d->layout_->keepempty)
3380 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3384 bool Paragraph::brokenBiblio() const
3386 // there is a problem if there is no bibitem at position 0 or
3387 // if there is another bibitem in the paragraph.
3388 return d->layout_->labeltype == LABEL_BIBLIO
3389 && (d->insetlist_.find(BIBITEM_CODE) != 0
3390 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3394 int Paragraph::fixBiblio(Buffer const & buffer)
3396 // FIXME: What about the case where paragraph is not BIBLIO
3397 // but there is an InsetBibitem?
3398 // FIXME: when there was already an inset at 0, the return value is 1,
3399 // which does not tell whether another inset has been remove; the
3400 // cursor cannot be correctly updated.
3402 if (d->layout_->labeltype != LABEL_BIBLIO)
3405 bool const track_changes = buffer.params().track_changes;
3406 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3407 bool const hasbibitem0 = bibitem_pos == 0;
3410 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3411 // There was an InsetBibitem at pos 0, and no other one => OK
3412 if (bibitem_pos == -1)
3414 // there is a bibitem at the 0 position, but since
3415 // there is a second one, we copy the second on the
3416 // first. We're assuming there are at most two of
3417 // these, which there should be.
3418 // FIXME: why does it make sense to do that rather
3419 // than keep the first? (JMarc)
3420 Inset * inset = releaseInset(bibitem_pos);
3421 d->insetlist_.begin()->inset = inset;
3422 return -bibitem_pos;
3425 // We need to create an inset at the beginning
3427 if (bibitem_pos > 0) {
3428 // there was one somewhere in the paragraph, let's move it
3429 inset = d->insetlist_.release(bibitem_pos);
3430 eraseChar(bibitem_pos, track_changes);
3433 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3434 InsetCommandParams(BIBITEM_CODE));
3436 Font font(inherit_font, buffer.params().language);
3437 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3438 : Change::UNCHANGED));
3444 void Paragraph::checkAuthors(AuthorList const & authorList)
3446 d->changes_.checkAuthors(authorList);
3450 bool Paragraph::isChanged(pos_type pos) const
3452 return lookupChange(pos).changed();
3456 bool Paragraph::isInserted(pos_type pos) const
3458 return lookupChange(pos).inserted();
3462 bool Paragraph::isDeleted(pos_type pos) const
3464 return lookupChange(pos).deleted();
3468 InsetList const & Paragraph::insetList() const
3470 return d->insetlist_;
3474 void Paragraph::setBuffer(Buffer & b)
3476 d->insetlist_.setBuffer(b);
3480 Inset * Paragraph::releaseInset(pos_type pos)
3482 Inset * inset = d->insetlist_.release(pos);
3483 /// does not honour change tracking!
3484 eraseChar(pos, false);
3489 Inset * Paragraph::getInset(pos_type pos)
3491 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3492 ? d->insetlist_.get(pos) : 0;
3496 Inset const * Paragraph::getInset(pos_type pos) const
3498 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3499 ? d->insetlist_.get(pos) : 0;
3503 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3504 pos_type & right, TextCase action)
3506 // process sequences of modified characters; in change
3507 // tracking mode, this approach results in much better
3508 // usability than changing case on a char-by-char basis
3509 // We also need to track the current font, since font
3510 // changes within sequences can occur.
3511 vector<pair<char_type, Font> > changes;
3513 bool const trackChanges = bparams.track_changes;
3515 bool capitalize = true;
3517 for (; pos < right; ++pos) {
3518 char_type oldChar = d->text_[pos];
3519 char_type newChar = oldChar;
3521 // ignore insets and don't play with deleted text!
3522 if (oldChar != META_INSET && !isDeleted(pos)) {
3524 case text_lowercase:
3525 newChar = lowercase(oldChar);
3527 case text_capitalization:
3529 newChar = uppercase(oldChar);
3533 case text_uppercase:
3534 newChar = uppercase(oldChar);
3539 if (isWordSeparator(pos) || isDeleted(pos)) {
3540 // permit capitalization again
3544 if (oldChar != newChar) {
3545 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3546 if (pos != right - 1)
3548 // step behind the changing area
3552 int erasePos = pos - changes.size();
3553 for (size_t i = 0; i < changes.size(); i++) {
3554 insertChar(pos, changes[i].first,
3557 if (!eraseChar(erasePos, trackChanges)) {
3560 ++right; // expand selection
3568 int Paragraph::find(docstring const & str, bool cs, bool mw,
3569 pos_type start_pos, bool del) const
3571 pos_type pos = start_pos;
3572 int const strsize = str.length();
3574 pos_type const parsize = d->text_.size();
3575 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3576 // Ignore "invisible" letters such as ligature breaks
3577 // and hyphenation chars while searching
3578 while (pos < parsize - 1 && isInset(pos)) {
3579 odocstringstream os;
3580 getInset(pos)->toString(os);
3581 if (!getInset(pos)->isLetter() || !os.str().empty())
3585 if (cs && str[i] != d->text_[pos])
3587 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3589 if (!del && isDeleted(pos))
3596 // if necessary, check whether string matches word
3598 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3601 && !isWordSeparator(pos))
3605 return pos - start_pos;
3609 char_type Paragraph::getChar(pos_type pos) const
3611 return d->text_[pos];
3615 pos_type Paragraph::size() const
3617 return d->text_.size();
3621 bool Paragraph::empty() const
3623 return d->text_.empty();
3627 bool Paragraph::isInset(pos_type pos) const
3629 return d->text_[pos] == META_INSET;
3633 bool Paragraph::isSeparator(pos_type pos) const
3635 //FIXME: Are we sure this can be the only separator?
3636 return d->text_[pos] == ' ';
3640 void Paragraph::deregisterWords()
3642 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3643 Private::LangWordsMap::const_iterator ite = d->words_.end();
3644 for (; itl != ite; ++itl) {
3645 WordList * wl = theWordList(itl->first);
3646 Private::Words::const_iterator it = (itl->second).begin();
3647 Private::Words::const_iterator et = (itl->second).end();
3648 for (; it != et; ++it)
3655 void Paragraph::locateWord(pos_type & from, pos_type & to,
3656 word_location const loc) const
3659 case WHOLE_WORD_STRICT:
3660 if (from == 0 || from == size()
3661 || isWordSeparator(from)
3662 || isWordSeparator(from - 1)) {
3666 // no break here, we go to the next
3669 // If we are already at the beginning of a word, do nothing
3670 if (!from || isWordSeparator(from - 1))
3672 // no break here, we go to the next
3675 // always move the cursor to the beginning of previous word
3676 while (from && !isWordSeparator(from - 1))
3680 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3683 // no need to move the 'from' cursor
3687 while (to < size() && !isWordSeparator(to))
3692 void Paragraph::collectWords()
3694 for (pos_type pos = 0; pos < size(); ++pos) {
3695 if (isWordSeparator(pos))
3697 pos_type from = pos;
3698 locateWord(from, pos, WHOLE_WORD);
3699 // Work around MSVC warning: The statement
3700 // if (pos < from + lyxrc.completion_minlength)
3701 // triggers a signed vs. unsigned warning.
3702 // I don't know why this happens, it could be a MSVC bug, or
3703 // related to LLP64 (windows) vs. LP64 (unix) programming
3704 // model, or the C++ standard might be ambigous in the section
3705 // defining the "usual arithmetic conversions". However, using
3706 // a temporary variable is safe and works on all compilers.
3707 pos_type const endpos = from + lyxrc.completion_minlength;
3710 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
3711 if (cit == d->fontlist_.end())
3713 Language const * lang = cit->font().language();
3714 docstring const word = asString(from, pos, AS_STR_NONE);
3715 d->words_[lang->lang()].insert(word);
3720 void Paragraph::registerWords()
3722 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3723 Private::LangWordsMap::const_iterator ite = d->words_.end();
3724 for (; itl != ite; ++itl) {
3725 WordList * wl = theWordList(itl->first);
3726 Private::Words::const_iterator it = (itl->second).begin();
3727 Private::Words::const_iterator et = (itl->second).end();
3728 for (; it != et; ++it)
3734 void Paragraph::updateWords()
3742 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3744 SkipPositionsIterator begin = skips.begin();
3745 SkipPositions::iterator end = skips.end();
3746 if (pos > 0 && begin < end) {
3748 if (end->last == pos - 1) {
3753 skips.insert(end, FontSpan(pos, pos));
3757 Language * Paragraph::Private::locateSpellRange(
3758 pos_type & from, pos_type & to,
3759 SkipPositions & skips) const
3761 // skip leading white space
3762 while (from < to && owner_->isWordSeparator(from))
3764 // don't check empty range
3767 // get current language
3768 Language * lang = getSpellLanguage(from);
3769 pos_type last = from;
3770 bool samelang = true;
3771 bool sameinset = true;
3772 while (last < to && samelang && sameinset) {
3773 // hop to end of word
3774 while (last < to && !owner_->isWordSeparator(last)) {
3775 if (owner_->getInset(last)) {
3776 appendSkipPosition(skips, last);
3777 } else if (owner_->isDeleted(last)) {
3778 appendSkipPosition(skips, last);
3782 // hop to next word while checking for insets
3783 while (sameinset && last < to && owner_->isWordSeparator(last)) {
3784 if (Inset const * inset = owner_->getInset(last))
3785 sameinset = inset->isChar() && inset->isLetter();
3786 if (sameinset && owner_->isDeleted(last)) {
3787 appendSkipPosition(skips, last);
3792 if (sameinset && last < to) {
3793 // now check for language change
3794 samelang = lang == getSpellLanguage(last);
3797 // if language change detected backstep is needed
3805 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
3808 const_cast<Language *>(owner_->getFontSettings(
3809 inset_owner_->buffer().params(), from).language());
3810 if (lang == inset_owner_->buffer().params().language
3811 && !lyxrc.spellchecker_alt_lang.empty()) {
3813 string const lang_variety =
3814 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3815 lang->setCode(lang_code);
3816 lang->setVariety(lang_variety);
3822 void Paragraph::requestSpellCheck(pos_type pos)
3824 d->requestSpellCheck(pos);
3828 bool Paragraph::needsSpellCheck() const
3830 SpellChecker::ChangeNumber speller_change_number = 0;
3831 if (theSpellChecker())
3832 speller_change_number = theSpellChecker()->changeNumber();
3833 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
3834 d->speller_state_.needsCompleteRefresh(speller_change_number);
3836 return d->needsSpellCheck();
3840 bool Paragraph::Private::ignoreWord(docstring const & word) const
3842 // Ignore words with digits
3843 // FIXME: make this customizable
3844 // (note that some checkers ignore words with digits by default)
3845 docstring::const_iterator cit = word.begin();
3846 docstring::const_iterator const end = word.end();
3847 for (; cit != end; ++cit) {
3848 if (isNumber((*cit)))
3855 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
3856 WordLangTuple & wl, docstring_list & suggestions,
3857 bool do_suggestion, bool check_learned) const
3859 SpellChecker::Result result = SpellChecker::WORD_OK;
3860 SpellChecker * speller = theSpellChecker();
3864 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
3867 locateWord(from, to, WHOLE_WORD);
3868 if (from == to || from >= size())
3871 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3872 Language * lang = d->getSpellLanguage(from);
3874 wl = WordLangTuple(word, lang);
3879 if (needsSpellCheck() || check_learned) {
3881 if (!d->ignoreWord(word)) {
3882 bool const trailing_dot = to < size() && d->text_[to] == '.';
3883 result = speller->check(wl);
3884 if (SpellChecker::misspelled(result) && trailing_dot) {
3885 wl = WordLangTuple(word.append(from_ascii(".")), lang);
3886 result = speller->check(wl);
3887 if (!SpellChecker::misspelled(result)) {
3888 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
3890 from << ".." << to << "]");
3892 // spell check with dot appended failed too
3893 // restore original word/lang value
3894 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3895 wl = WordLangTuple(word, lang);
3899 if (!SpellChecker::misspelled(result)) {
3900 // area up to the begin of the next word is not misspelled
3901 while (end < size() && isWordSeparator(end))
3904 d->setMisspelled(from, end, result);
3906 result = d->speller_state_.getState(from);
3910 suggestions.clear();
3912 if (SpellChecker::misspelled(result)) {
3913 LYXERR(Debug::GUI, "misspelled word: \"" <<
3915 from << ".." << to << "]");
3917 speller->suggest(wl, suggestions);
3923 void Paragraph::Private::markMisspelledWords(
3924 pos_type const & first, pos_type const & last,
3925 SpellChecker::Result result,
3926 docstring const & word,
3927 SkipPositions const & skips)
3929 if (!SpellChecker::misspelled(result)) {
3930 setMisspelled(first, last, SpellChecker::WORD_OK);
3934 SpellChecker * speller = theSpellChecker();
3935 // locate and enumerate the error positions
3936 int nerrors = speller->numMisspelledWords();
3938 SkipPositionsIterator it = skips.begin();
3939 SkipPositionsIterator et = skips.end();
3940 for (int index = 0; index < nerrors; ++index) {
3943 speller->misspelledWord(index, wstart, wlen);
3944 /// should not happen if speller supports range checks
3945 if (!wlen) continue;
3946 docstring const misspelled = word.substr(wstart, wlen);
3947 wstart += first + numskipped;
3948 if (snext < wstart) {
3949 /// mark the range of correct spelling
3950 numskipped += countSkips(it, et, wstart);
3951 setMisspelled(snext,
3952 wstart - 1, SpellChecker::WORD_OK);
3954 snext = wstart + wlen;
3955 numskipped += countSkips(it, et, snext);
3956 /// mark the range of misspelling
3957 setMisspelled(wstart, snext, result);
3958 LYXERR(Debug::GUI, "misspelled word: \"" <<
3959 misspelled << "\" [" <<
3960 wstart << ".." << (snext-1) << "]");
3963 if (snext <= last) {
3964 /// mark the range of correct spelling at end
3965 setMisspelled(snext, last, SpellChecker::WORD_OK);
3970 void Paragraph::spellCheck() const
3972 SpellChecker * speller = theSpellChecker();
3973 if (!speller || empty() ||!needsSpellCheck())
3977 d->rangeOfSpellCheck(start, endpos);
3978 if (speller->canCheckParagraph()) {
3979 // loop until we leave the range
3980 for (pos_type first = start; first < endpos; ) {
3981 pos_type last = endpos;
3982 Private::SkipPositions skips;
3983 Language * lang = d->locateSpellRange(first, last, skips);
3984 if (first >= endpos)
3986 // start the spell checker on the unit of meaning
3987 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
3988 WordLangTuple wl = WordLangTuple(word, lang);
3989 SpellChecker::Result result = word.size() ?
3990 speller->check(wl) : SpellChecker::WORD_OK;
3991 d->markMisspelledWords(first, last, result, word, skips);
3995 static docstring_list suggestions;
3996 pos_type to = endpos;
3997 while (start < endpos) {
3999 spellCheck(start, to, wl, suggestions, false);
4003 d->readySpellCheck();
4007 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4009 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4010 if (result || pos <= 0 || pos > size())
4012 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4013 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4018 string Paragraph::magicLabel() const
4021 ss << "magicparlabel-" << id();