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 "mathed/InsetMathHull.h"
58 #include "support/debug.h"
59 #include "support/docstring_list.h"
60 #include "support/ExceptionMessage.h"
61 #include "support/gettext.h"
62 #include "support/lassert.h"
63 #include "support/lstrings.h"
64 #include "support/textutils.h"
71 using namespace lyx::support;
77 /// Inset identifier (above 0x10ffff, for ucs-4)
78 char_type const META_INSET = 0x200001;
83 /////////////////////////////////////////////////////////////////////
87 /////////////////////////////////////////////////////////////////////
89 class SpellResultRange {
91 SpellResultRange(FontSpan range, SpellChecker::Result result)
92 : range_(range), result_(result)
95 FontSpan const & range() const { return range_; }
97 void range(FontSpan const & r) { range_ = r; }
99 SpellChecker::Result result() const { return result_; }
101 void result(SpellChecker::Result r) { result_ = r; }
103 bool contains(pos_type pos) const { return range_.contains(pos); }
105 bool covered(FontSpan const & r) const
107 // 1. first of new range inside current range or
108 // 2. last of new range inside current range or
109 // 3. first of current range inside new range or
110 // 4. last of current range inside new range
111 //FIXME: is this the same as !range_.intersect(r).empty() ?
112 return range_.contains(r.first) || range_.contains(r.last) ||
113 r.contains(range_.first) || r.contains(range_.last);
116 void shift(pos_type pos, int offset)
118 if (range_.first > pos) {
119 range_.first += offset;
120 range_.last += offset;
121 } else if (range_.last >= pos) {
122 range_.last += offset;
127 SpellChecker::Result result_ ;
131 /////////////////////////////////////////////////////////////////////
135 /////////////////////////////////////////////////////////////////////
137 class SpellCheckerState {
141 needs_refresh_ = true;
142 current_change_number_ = 0;
145 void setRange(FontSpan const & fp, SpellChecker::Result state)
148 RangesIterator et = ranges_.end();
149 RangesIterator it = ranges_.begin();
150 for (; it != et; ++it) {
151 if (!it->covered(fp))
152 result.push_back(SpellResultRange(it->range(), it->result()));
153 else if (state == SpellChecker::WORD_OK) {
154 // trim or split the current misspelled range
155 // store misspelled ranges only
156 FontSpan range = it->range();
157 if (fp.first > range.first) {
158 // misspelled area in front of WORD_OK
159 range.last = fp.first - 1;
160 result.push_back(SpellResultRange(range, it->result()));
163 if (fp.last < range.last) {
164 // misspelled area after WORD_OK range
165 range.first = fp.last + 1;
166 result.push_back(SpellResultRange(range, it->result()));
171 if (state != SpellChecker::WORD_OK)
172 ranges_.push_back(SpellResultRange(fp, state));
175 void increasePosAfterPos(pos_type pos)
177 correctRangesAfterPos(pos, 1);
181 void decreasePosAfterPos(pos_type pos)
183 correctRangesAfterPos(pos, -1);
187 void refreshLast(pos_type pos)
189 if (pos < refresh_.last)
193 SpellChecker::Result getState(pos_type pos) const
195 SpellChecker::Result result = SpellChecker::WORD_OK;
196 RangesIterator et = ranges_.end();
197 RangesIterator it = ranges_.begin();
198 for (; it != et; ++it) {
199 if(it->contains(pos)) {
206 FontSpan const & getRange(pos_type pos) const
208 /// empty span to indicate mismatch
209 static FontSpan empty_;
210 RangesIterator et = ranges_.end();
211 RangesIterator it = ranges_.begin();
212 for (; it != et; ++it) {
213 if(it->contains(pos)) {
220 bool needsRefresh() const
222 return needs_refresh_;
225 SpellChecker::ChangeNumber currentChangeNumber() const
227 return current_change_number_;
230 void refreshRange(pos_type & first, pos_type & last) const
232 first = refresh_.first;
233 last = refresh_.last;
236 void needsRefresh(pos_type pos)
238 if (needs_refresh_ && pos != -1) {
239 if (pos < refresh_.first)
240 refresh_.first = pos;
241 if (pos > refresh_.last)
243 } else if (pos != -1) {
244 // init request check for neighbour positions too
245 refresh_.first = pos > 0 ? pos - 1 : 0;
246 // no need for special end of paragraph check
247 refresh_.last = pos + 1;
249 needs_refresh_ = pos != -1;
252 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
254 needs_refresh_ = true;
257 current_change_number_ = change_number;
260 typedef vector<SpellResultRange> Ranges;
261 typedef Ranges::const_iterator RangesIterator;
263 /// the area of the paragraph with pending spell check
266 /// spell state cache version number
267 SpellChecker::ChangeNumber current_change_number_;
270 void correctRangesAfterPos(pos_type pos, int offset)
272 RangesIterator et = ranges_.end();
273 Ranges::iterator it = ranges_.begin();
274 for (; it != et; ++it) {
275 it->shift(pos, offset);
281 /////////////////////////////////////////////////////////////////////
283 // Paragraph::Private
285 /////////////////////////////////////////////////////////////////////
287 class Paragraph::Private
289 // Enforce our own "copy" constructor
290 Private(Private const &) = delete;
291 Private & operator=(Private const &) = delete;
292 // Unique ID generator
293 static int make_id();
296 Private(Paragraph * owner, Layout const & layout);
297 /// "Copy constructor"
298 Private(Private const &, Paragraph * owner);
299 /// Copy constructor from \p beg to \p end
300 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
303 void insertChar(pos_type pos, char_type c, Change const & change);
305 /// Output the surrogate pair formed by \p c and \p next to \p os.
306 /// \return the number of characters written.
307 int latexSurrogatePair(otexstream & os, char_type c, char_type next,
308 OutputParams const &);
310 /// Output a space in appropriate formatting (or a surrogate pair
311 /// if the next character is a combining character).
312 /// \return whether a surrogate pair was output.
313 bool simpleTeXBlanks(OutputParams const &,
316 unsigned int & column,
318 Layout const & style);
320 /// Output consecutive unicode chars, belonging to the same script as
321 /// specified by the latex macro \p ltx, to \p os starting from \p i.
322 /// \return the number of characters written.
323 int writeScriptChars(otexstream & os, docstring const & ltx,
324 Change const &, Encoding const &, pos_type & i);
326 /// This could go to ParagraphParameters if we want to.
327 int startTeXParParams(BufferParams const &, otexstream &,
328 OutputParams const &) const;
330 /// This could go to ParagraphParameters if we want to.
331 bool endTeXParParams(BufferParams const &, otexstream &,
332 OutputParams const &) const;
335 void latexInset(BufferParams const &,
340 Font const & outerfont,
342 Change & running_change,
343 Layout const & style,
345 unsigned int & column);
348 void latexSpecialChar(
350 BufferParams const & bparams,
351 OutputParams const & runparams,
352 Font const & running_font,
353 Change const & running_change,
354 Layout const & style,
357 unsigned int & column);
364 unsigned int & column);
370 unsigned int & column);
373 void validate(LaTeXFeatures & features) const;
375 /// Checks if the paragraph contains only text and no inset or font change.
376 bool onlyText(Buffer const & buf, Font const & outerfont,
377 pos_type initial) const;
379 /// a vector of speller skip positions
380 typedef vector<FontSpan> SkipPositions;
381 typedef SkipPositions::const_iterator SkipPositionsIterator;
383 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
385 Language * getSpellLanguage(pos_type const from) const;
387 Language * locateSpellRange(pos_type & from, pos_type & to,
388 SkipPositions & skips) const;
390 bool hasSpellerChange() const
392 SpellChecker::ChangeNumber speller_change_number = 0;
393 if (theSpellChecker())
394 speller_change_number = theSpellChecker()->changeNumber();
395 return speller_change_number > speller_state_.currentChangeNumber();
398 bool ignoreWord(docstring const & word) const ;
400 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
402 pos_type textsize = owner_->size();
403 // check for sane arguments
404 if (to <= from || from >= textsize)
406 FontSpan fp = FontSpan(from, to - 1);
407 speller_state_.setRange(fp, state);
410 void requestSpellCheck(pos_type pos)
413 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
415 speller_state_.needsRefresh(pos);
418 void readySpellCheck()
420 speller_state_.needsRefresh(-1);
423 bool needsSpellCheck() const
425 return speller_state_.needsRefresh();
428 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
430 speller_state_.refreshRange(first, last);
432 last = owner_->size();
435 pos_type endpos = last;
436 owner_->locateWord(first, endpos, WHOLE_WORD);
439 owner_->locateWord(last, endpos, WHOLE_WORD);
444 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
448 while (it != et && it->first < start) {
449 int skip = it->last - it->first + 1;
457 void markMisspelledWords(pos_type const & first, pos_type const & last,
458 SpellChecker::Result result,
459 docstring const & word,
460 SkipPositions const & skips);
462 InsetCode ownerCode() const
464 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
467 /// Which Paragraph owns us?
471 Inset const * inset_owner_;
480 ParagraphParameters params_;
482 /// for recording and looking up changes
486 InsetList insetlist_;
489 pos_type begin_of_body_;
491 typedef docstring TextContainer;
495 typedef set<docstring> Words;
496 typedef map<string, Words> LangWordsMap;
500 Layout const * layout_;
502 SpellCheckerState speller_state_;
506 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
507 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
514 int Paragraph::Private::make_id()
516 // The id is unique per session across buffers because it is used in
517 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
520 static atomic_uint next_id(0);
525 Paragraph::Private::Private(Private const & p, Paragraph * owner)
526 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
528 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
529 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
532 requestSpellCheck(p.text_.size());
536 Paragraph::Private::Private(Private const & p, Paragraph * owner,
537 pos_type beg, pos_type end)
538 : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()),
539 params_(p.params_), changes_(p.changes_),
540 insetlist_(p.insetlist_, beg, end),
541 begin_of_body_(p.begin_of_body_), words_(p.words_),
544 if (beg >= pos_type(p.text_.size()))
546 text_ = p.text_.substr(beg, end - beg);
548 FontList::const_iterator fcit = fontlist_.begin();
549 FontList::const_iterator fend = fontlist_.end();
550 for (; fcit != fend; ++fcit) {
551 if (fcit->pos() < beg)
553 if (fcit->pos() >= end) {
554 // Add last entry in the fontlist_.
555 fontlist_.set(text_.size() - 1, fcit->font());
558 // Add a new entry in the fontlist_.
559 fontlist_.set(fcit->pos() - beg, fcit->font());
561 requestSpellCheck(p.text_.size());
565 void Paragraph::addChangesToToc(DocIterator const & cdit,
566 Buffer const & buf, bool output_active) const
568 d->changes_.addToToc(cdit, buf, output_active);
572 void Paragraph::addChangesToBuffer(Buffer const & buf) const
574 d->changes_.updateBuffer(buf);
578 bool Paragraph::isChangeUpdateRequired() const
580 return d->changes_.isUpdateRequired();
584 bool Paragraph::isDeleted(pos_type start, pos_type end) const
586 LASSERT(start >= 0 && start <= size(), return false);
587 LASSERT(end > start && end <= size() + 1, return false);
589 return d->changes_.isDeleted(start, end);
593 bool Paragraph::isChanged(pos_type start, pos_type end) const
595 LASSERT(start >= 0 && start <= size(), return false);
596 LASSERT(end > start && end <= size() + 1, return false);
598 return d->changes_.isChanged(start, end);
602 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
604 // keep the logic here in sync with the logic of eraseChars()
608 Change const & change = d->changes_.lookup(size());
609 return change.inserted() && change.currentAuthor();
613 void Paragraph::setChange(Change const & change)
615 // beware of the imaginary end-of-par character!
616 d->changes_.set(change, 0, size() + 1);
619 * Propagate the change recursively - but not in case of DELETED!
621 * Imagine that your co-author makes changes in an existing inset. He
622 * sends your document to you and you come to the conclusion that the
623 * inset should go completely. If you erase it, LyX must not delete all
624 * text within the inset. Otherwise, the change tracked insertions of
625 * your co-author get lost and there is no way to restore them later.
627 * Conclusion: An inset's content should remain untouched if you delete it
630 if (!change.deleted()) {
631 for (pos_type pos = 0; pos < size(); ++pos) {
632 if (Inset * inset = getInset(pos))
633 inset->setChange(change);
639 void Paragraph::setChange(pos_type pos, Change const & change)
641 LASSERT(pos >= 0 && pos <= size(), return);
642 d->changes_.set(change, pos);
644 // see comment in setChange(Change const &) above
645 if (!change.deleted() && pos < size())
646 if (Inset * inset = getInset(pos))
647 inset->setChange(change);
651 Change const & Paragraph::lookupChange(pos_type pos) const
653 LBUFERR(pos >= 0 && pos <= size());
654 return d->changes_.lookup(pos);
658 void Paragraph::acceptChanges(pos_type start, pos_type end)
660 LASSERT(start >= 0 && start <= size(), return);
661 LASSERT(end > start && end <= size() + 1, return);
663 for (pos_type pos = start; pos < end; ++pos) {
664 switch (lookupChange(pos).type) {
665 case Change::UNCHANGED:
666 // accept changes in nested inset
667 if (Inset * inset = getInset(pos))
668 inset->acceptChanges();
671 case Change::INSERTED:
672 d->changes_.set(Change(Change::UNCHANGED), pos);
673 // also accept changes in nested inset
674 if (Inset * inset = getInset(pos))
675 inset->acceptChanges();
678 case Change::DELETED:
679 // Suppress access to non-existent
680 // "end-of-paragraph char"
682 eraseChar(pos, false);
693 void Paragraph::rejectChanges(pos_type start, pos_type end)
695 LASSERT(start >= 0 && start <= size(), return);
696 LASSERT(end > start && end <= size() + 1, return);
698 for (pos_type pos = start; pos < end; ++pos) {
699 switch (lookupChange(pos).type) {
700 case Change::UNCHANGED:
701 // reject changes in nested inset
702 if (Inset * inset = getInset(pos))
703 inset->rejectChanges();
706 case Change::INSERTED:
707 // Suppress access to non-existent
708 // "end-of-paragraph char"
710 eraseChar(pos, false);
716 case Change::DELETED:
717 d->changes_.set(Change(Change::UNCHANGED), pos);
719 // Do NOT reject changes within a deleted inset!
720 // There may be insertions of a co-author inside of it!
728 void Paragraph::Private::insertChar(pos_type pos, char_type c,
729 Change const & change)
731 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
734 changes_.insert(change, pos);
736 // This is actually very common when parsing buffers (and
737 // maybe inserting ascii text)
738 if (pos == pos_type(text_.size())) {
739 // when appending characters, no need to update tables
741 // but we want spell checking
742 requestSpellCheck(pos);
746 text_.insert(text_.begin() + pos, c);
748 // Update the font table.
749 fontlist_.increasePosAfterPos(pos);
752 insetlist_.increasePosAfterPos(pos);
754 // Update list of misspelled positions
755 speller_state_.increasePosAfterPos(pos);
759 bool Paragraph::insertInset(pos_type pos, Inset * inset,
760 Font const & font, Change const & change)
762 LASSERT(inset, return false);
763 LASSERT(pos >= 0 && pos <= size(), return false);
765 // Paragraph::insertInset() can be used in cut/copy/paste operation where
766 // d->inset_owner_ is not set yet.
767 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
770 d->insertChar(pos, META_INSET, change);
771 LASSERT(d->text_[pos] == META_INSET, return false);
773 // Add a new entry in the insetlist_.
774 d->insetlist_.insert(inset, pos);
776 // Some insets require run of spell checker
777 requestSpellCheck(pos);
783 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
785 LASSERT(pos >= 0 && pos <= size(), return false);
787 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
790 Change change = d->changes_.lookup(pos);
792 // set the character to DELETED if
793 // a) it was previously unchanged or
794 // b) it was inserted by a co-author
796 if (!change.changed() ||
797 (change.inserted() && !change.currentAuthor())) {
798 setChange(pos, Change(Change::DELETED));
799 // request run of spell checker
800 requestSpellCheck(pos);
804 if (change.deleted())
808 // Don't physically access the imaginary end-of-paragraph character.
809 // eraseChar() can only mark it as DELETED. A physical deletion of
810 // end-of-par must be handled externally.
816 d->changes_.erase(pos);
818 // if it is an inset, delete the inset entry
819 if (d->text_[pos] == META_INSET)
820 d->insetlist_.erase(pos);
822 d->text_.erase(d->text_.begin() + pos);
824 // Update the fontlist_
825 d->fontlist_.erase(pos);
827 // Update the insetlist_
828 d->insetlist_.decreasePosAfterPos(pos);
830 // Update list of misspelled positions
831 d->speller_state_.decreasePosAfterPos(pos);
832 d->speller_state_.refreshLast(size());
838 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
840 LASSERT(start >= 0 && start <= size(), return 0);
841 LASSERT(end >= start && end <= size() + 1, return 0);
844 for (pos_type count = end - start; count; --count) {
845 if (!eraseChar(i, trackChanges))
852 int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
853 char_type next, OutputParams const & runparams)
855 // Writing next here may circumvent a possible font change between
856 // c and next. Since next is only output if it forms a surrogate pair
857 // with c we can ignore this:
858 // A font change inside a surrogate pair does not make sense and is
859 // hopefully impossible to input.
860 // FIXME: change tracking
861 // Is this correct WRT change tracking?
862 Encoding const & encoding = *(runparams.encoding);
863 docstring latex1 = encoding.latexChar(next).first;
864 if (runparams.inIPA) {
865 string const tipashortcut = Encodings::TIPAShortcut(next);
866 if (!tipashortcut.empty()) {
867 latex1 = from_ascii(tipashortcut);
870 docstring const latex2 = encoding.latexChar(c).first;
871 if (docstring(1, next) == latex1) {
872 // the encoding supports the combination
873 os << latex2 << latex1;
874 return latex1.length() + latex2.length();
875 } else if (runparams.local_font &&
876 runparams.local_font->language()->lang() == "polutonikogreek") {
877 // polutonikogreek only works without the brackets
878 os << latex1 << latex2;
879 return latex1.length() + latex2.length();
881 os << latex1 << '{' << latex2 << '}';
882 return latex1.length() + latex2.length() + 2;
886 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
889 unsigned int & column,
891 Layout const & style)
893 if (style.pass_thru || runparams.pass_thru)
896 if (i + 1 < int(text_.size())) {
897 char_type next = text_[i + 1];
898 if (Encodings::isCombiningChar(next)) {
899 // This space has an accent, so we must always output it.
900 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
905 if (runparams.linelen > 0
906 && column > runparams.linelen
908 && text_[i - 1] != ' '
909 && (i + 1 < int(text_.size()))
910 // same in FreeSpacing mode
911 && !owner_->isFreeSpacing()
912 // In typewriter mode, we want to avoid
913 // ! . ? : at the end of a line
914 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
915 && (text_[i - 1] == '.'
916 || text_[i - 1] == '?'
917 || text_[i - 1] == ':'
918 || text_[i - 1] == '!'))) {
920 os.texrow().start(owner_->id(), i + 1);
922 } else if (style.free_spacing) {
931 int Paragraph::Private::writeScriptChars(otexstream & os,
932 docstring const & ltx,
933 Change const & runningChange,
934 Encoding const & encoding,
937 // FIXME: modifying i here is not very nice...
939 // We only arrive here when character text_[i] could not be translated
940 // into the current latex encoding (or its latex translation has been forced,)
941 // and it belongs to a known script.
942 // TODO: We need \textcyr and \textgreek wrappers also for characters
943 // that can be encoded in the "LaTeX encoding" but not in the
944 // current *font encoding*.
945 // (See #9681 for details and test)
946 // Parameter ltx contains the latex translation of text_[i] as specified
947 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
948 // The latex macro name "textXXX" specifies the script to which text_[i]
949 // belongs and we use it in order to check whether characters from the
950 // same script immediately follow, such that we can collect them in a
951 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
952 // for the first char but only "<spec>" for all subsequent chars.
953 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
954 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
955 string script = to_ascii(ltx.substr(1, brace1 - 1));
958 bool closing_brace = true;
959 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
960 // Correct encoding is being used, so we can avoid \textgreek.
961 // TODO: wrong test: we need to check the *font encoding*
962 // (i.e. the active language and its FontEncoding tag)
963 // instead of the LaTeX *input encoding*!
964 // See #9637 for details and test-cases.
967 closing_brace = false;
969 os << ltx.substr(pos, length);
970 int size = text_.size();
971 while (i + 1 < size) {
972 char_type const next = text_[i + 1];
973 // Stop here if next character belongs to another script
974 // or there is a change in change tracking status.
975 if (!Encodings::isKnownScriptChar(next, script) ||
976 runningChange != owner_->lookupChange(i + 1))
980 FontList::const_iterator cit = fontlist_.begin();
981 FontList::const_iterator end = fontlist_.end();
982 for (; cit != end; ++cit) {
983 if (cit->pos() >= i && !found) {
984 prev_font = cit->font();
987 if (cit->pos() >= i + 1)
990 // Stop here if there is a font attribute or encoding change.
991 if (found && cit != end && prev_font != cit->font())
993 docstring const latex = encoding.latexChar(next).first;
994 docstring::size_type const b1 =
995 latex.find_first_of(from_ascii("{"));
996 docstring::size_type const b2 =
997 latex.find_last_of(from_ascii("}"));
998 int const len = b2 - b1 - 1;
999 os << latex.substr(b1 + 1, len);
1003 if (closing_brace) {
1011 void Paragraph::Private::latexInset(BufferParams const & bparams,
1013 OutputParams & runparams,
1014 Font & running_font,
1016 Font const & outerfont,
1018 Change & running_change,
1019 Layout const & style,
1021 unsigned int & column)
1023 Inset * inset = owner_->getInset(i);
1026 if (style.pass_thru) {
1027 odocstringstream ods;
1028 inset->plaintext(ods, runparams);
1033 // FIXME: move this to InsetNewline::latex
1034 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1035 // newlines are handled differently here than
1036 // the default in simpleTeXSpecialChars().
1037 if (!style.newline_allowed) {
1041 column += running_font.latexWriteEndChanges(
1042 os, bparams, runparams,
1043 basefont, basefont);
1047 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1050 basefont = owner_->getLayoutFont(bparams, outerfont);
1051 running_font = basefont;
1053 if (runparams.moving_arg)
1057 os.texrow().start(owner_->id(), i + 1);
1061 if (owner_->isDeleted(i)) {
1062 if( ++runparams.inDeletedInset == 1)
1063 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1066 if (inset->canTrackChanges()) {
1067 column += Changes::latexMarkChange(os, bparams, running_change,
1068 Change(Change::UNCHANGED), runparams);
1069 running_change = Change(Change::UNCHANGED);
1073 odocstream::pos_type const len = os.os().tellp();
1075 if (inset->forceLTR()
1076 && !runparams.use_polyglossia
1077 && running_font.isRightToLeft()
1078 // ERT is an exception, it should be output with no
1079 // decorations at all
1080 && inset->lyxCode() != ERT_CODE) {
1081 if (running_font.language()->lang() == "farsi")
1082 os << "\\beginL" << termcmd;
1088 // FIXME: Bug: we can have an empty font change here!
1089 // if there has just been a font change, we are going to close it
1090 // right now, which means stupid latex code like \textsf{}. AFAIK,
1091 // this does not harm dvi output. A minor bug, thus (JMarc)
1093 // Some insets cannot be inside a font change command.
1094 // However, even such insets *can* be placed in \L or \R
1095 // or their equivalents (for RTL language switches), so we don't
1096 // close the language in those cases.
1097 // ArabTeX, though, cannot handle this special behavior, it seems.
1098 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1099 || running_font.language()->lang() == "arabic_arabtex";
1100 if (open_font && !inset->inheritFont()) {
1101 bool closeLanguage = arabtex
1102 || basefont.isRightToLeft() == running_font.isRightToLeft();
1103 unsigned int count = running_font.latexWriteEndChanges(os,
1104 bparams, runparams, basefont, basefont, closeLanguage);
1106 // if any font properties were closed, update the running_font,
1107 // making sure, however, to leave the language as it was
1109 // FIXME: probably a better way to keep track of the old
1110 // language, than copying the entire font?
1111 Font const copy_font(running_font);
1112 basefont = owner_->getLayoutFont(bparams, outerfont);
1113 running_font = basefont;
1115 running_font.setLanguage(copy_font.language());
1116 // leave font open if language is still open
1117 open_font = (running_font.language() == basefont.language());
1119 runparams.local_font = &basefont;
1123 size_t const previous_row_count = os.texrow().rows();
1126 runparams.lastid = id_;
1127 runparams.lastpos = i;
1128 inset->latex(os, runparams);
1129 } catch (EncodingException & e) {
1130 // add location information and throw again.
1137 if (running_font.language()->lang() == "farsi")
1138 os << "\\endL" << termcmd;
1143 if (os.texrow().rows() > previous_row_count) {
1144 os.texrow().start(owner_->id(), i + 1);
1147 column += (unsigned int)(os.os().tellp() - len);
1150 if (owner_->isDeleted(i))
1151 --runparams.inDeletedInset;
1155 void Paragraph::Private::latexSpecialChar(otexstream & os,
1156 BufferParams const & bparams,
1157 OutputParams const & runparams,
1158 Font const & running_font,
1159 Change const & running_change,
1160 Layout const & style,
1163 unsigned int & column)
1165 // With polyglossia, brackets and stuff need not be reversed
1166 // in RTL scripts (see bug #8251)
1167 char_type const c = (runparams.use_polyglossia) ?
1168 owner_->getUChar(bparams, i) : text_[i];
1170 if (style.pass_thru || runparams.pass_thru
1171 || contains(style.pass_thru_chars, c)
1172 || contains(runparams.pass_thru_chars, c)) {
1174 Encoding const * const enc = runparams.encoding;
1175 if (enc && !enc->encodable(c))
1176 throw EncodingException(c);
1182 // TIPA uses its own T3 encoding
1183 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1185 // If T1 font encoding is used, use the special
1186 // characters it provides.
1187 // NOTE: Some languages reset the font encoding internally to a
1188 // non-standard font encoding. If we are using such a language,
1189 // we do not output special T1 chars.
1190 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1191 && bparams.font_encoding() == "T1" && latexSpecialT1(c, os, i, column))
1194 // Otherwise, we use what LaTeX provides us.
1197 os << "\\textbackslash" << termcmd;
1201 os << "\\textless" << termcmd;
1205 os << "\\textgreater" << termcmd;
1209 os << "\\textbar" << termcmd;
1214 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1215 (end_pos == -1 || i + 1 < end_pos) &&
1216 text_[i+1] == '-') {
1217 // Prevent "--" becoming an endash and "---" becoming
1219 // Within \ttfamily, "--" is merged to "-" (no endash)
1220 // so we avoid this rather irritating ligature as well
1226 os << "\\char`\\\"" << termcmd;
1231 case '%': case '#': case '{':
1239 os << "\\textasciitilde" << termcmd;
1244 os << "\\textasciicircum" << termcmd;
1251 // avoid being mistaken for optional arguments
1259 // Blanks are printed before font switching.
1260 // Sure? I am not! (try nice-latex)
1261 // I am sure it's correct. LyX might be smarter
1262 // in the future, but for now, nothing wrong is
1270 Encoding const & encoding = *(runparams.encoding);
1271 char_type next = '\0';
1272 if (i + 1 < int(text_.size())) {
1273 next = text_[i + 1];
1274 if (Encodings::isCombiningChar(next)) {
1275 column += latexSurrogatePair(os, c, next, runparams) - 1;
1281 pair<docstring, bool> latex = encoding.latexChar(c);
1282 docstring nextlatex;
1283 bool nexttipas = false;
1284 string nexttipashortcut;
1285 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1286 nextlatex = encoding.latexChar(next).first;
1287 if (runparams.inIPA) {
1288 nexttipashortcut = Encodings::TIPAShortcut(next);
1289 nexttipas = !nexttipashortcut.empty();
1293 if (runparams.inIPA) {
1294 string const tipashortcut = Encodings::TIPAShortcut(c);
1295 if (!tipashortcut.empty()) {
1296 latex.first = from_ascii(tipashortcut);
1297 latex.second = false;
1301 if (Encodings::isKnownScriptChar(c, script)
1302 && prefixIs(latex.first, from_ascii("\\" + script)))
1303 column += writeScriptChars(os, latex.first,
1304 running_change, encoding, i) - 1;
1305 else if (latex.second
1306 && ((!prefixIs(nextlatex, '\\')
1307 && !prefixIs(nextlatex, '{')
1308 && !prefixIs(nextlatex, '}'))
1310 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1312 // Prevent eating of a following
1313 // space or command corruption by
1314 // following characters
1315 if (next == ' ' || next == '\0') {
1316 column += latex.first.length() + 1;
1317 os << latex.first << "{}";
1319 column += latex.first.length();
1320 os << latex.first << " ";
1323 column += latex.first.length() - 1;
1331 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1332 pos_type i, unsigned int & column)
1338 // In T1 encoding, these characters exist
1339 // but we should avoid ligatures
1340 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1342 os << "\\textcompwordmark" << termcmd;
1349 // soul.sty breaks with \char`\"
1350 os << "\\textquotedbl" << termcmd;
1359 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1360 pos_type /*i*/, unsigned int & column)
1370 os << "\\textvertline" << termcmd;
1379 void Paragraph::Private::validate(LaTeXFeatures & features) const
1381 if (layout_->inpreamble && inset_owner_) {
1382 bool const is_command = layout_->latextype == LATEX_COMMAND;
1383 Buffer const & buf = inset_owner_->buffer();
1384 BufferParams const & bp = features.runparams().is_child
1385 ? buf.masterParams() : buf.params();
1387 // Using a string stream here circumvents the encoding
1388 // switching machinery of odocstream. Therefore the
1389 // output is wrong if this paragraph contains content
1390 // that needs to switch encoding.
1391 otexstringstream os;
1393 os << '\\' << from_ascii(layout_->latexname());
1394 // we have to provide all the optional arguments here, even though
1395 // the last one is the only one we care about.
1396 // Separate handling of optional argument inset.
1397 if (!layout_->latexargs().empty()) {
1398 OutputParams rp = features.runparams();
1399 rp.local_font = &owner_->getFirstFontSettings(bp);
1400 latexArgInsets(*owner_, os, rp, layout_->latexargs());
1402 os << from_ascii(layout_->latexparam());
1404 size_t const length = os.length();
1405 // this will output "{" at the beginning, but not at the end
1406 owner_->latex(bp, f, os, features.runparams(), 0, -1, true);
1407 if (os.length() > length) {
1410 if (!layout_->postcommandargs().empty()) {
1411 OutputParams rp = features.runparams();
1412 rp.local_font = &owner_->getFirstFontSettings(bp);
1413 latexArgInsets(*owner_, os, rp, layout_->postcommandargs(), "post:");
1416 features.addPreambleSnippet(os.release(), true);
1420 if (features.runparams().flavor == OutputParams::HTML
1421 && layout_->htmltitle()) {
1422 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1425 // check the params.
1426 if (!params_.spacing().isDefault())
1427 features.require("setspace");
1430 features.useLayout(layout_->name());
1433 fontlist_.validate(features);
1435 // then the indentation
1436 if (!params_.leftIndent().zero())
1437 features.require("ParagraphLeftIndent");
1440 InsetList::const_iterator icit = insetlist_.begin();
1441 InsetList::const_iterator iend = insetlist_.end();
1442 for (; icit != iend; ++icit) {
1444 features.inDeletedInset(owner_->isDeleted(icit->pos));
1445 icit->inset->validate(features);
1446 features.inDeletedInset(false);
1447 if (layout_->needprotect &&
1448 icit->inset->lyxCode() == FOOT_CODE)
1449 features.require("NeedLyXFootnoteCode");
1453 // then the contents
1454 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1455 BufferEncodings::validate(text_[i], features);
1459 /////////////////////////////////////////////////////////////////////
1463 /////////////////////////////////////////////////////////////////////
1466 Layout const emptyParagraphLayout;
1469 Paragraph::Paragraph()
1470 : d(new Paragraph::Private(this, emptyParagraphLayout))
1477 Paragraph::Paragraph(Paragraph const & par)
1478 : itemdepth(par.itemdepth),
1479 d(new Paragraph::Private(*par.d, this))
1485 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1486 : itemdepth(par.itemdepth),
1487 d(new Paragraph::Private(*par.d, this, beg, end))
1493 Paragraph & Paragraph::operator=(Paragraph const & par)
1495 // needed as we will destroy the private part before copying it
1497 itemdepth = par.itemdepth;
1501 d = new Private(*par.d, this);
1508 Paragraph::~Paragraph()
1517 // this shall be called just before every "os << ..." action.
1518 void flushString(ostream & os, docstring & s)
1527 void Paragraph::write(ostream & os, BufferParams const & bparams,
1528 depth_type & dth) const
1530 // The beginning or end of a deeper (i.e. nested) area?
1531 if (dth != d->params_.depth()) {
1532 if (d->params_.depth() > dth) {
1533 while (d->params_.depth() > dth) {
1534 os << "\n\\begin_deeper";
1538 while (d->params_.depth() < dth) {
1539 os << "\n\\end_deeper";
1545 // First write the layout
1546 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1548 d->params_.write(os);
1550 Font font1(inherit_font, bparams.language);
1552 Change running_change = Change(Change::UNCHANGED);
1554 // this string is used as a buffer to avoid repetitive calls
1555 // to to_utf8(), which turn out to be expensive (JMarc)
1556 docstring write_buffer;
1559 for (pos_type i = 0; i <= size(); ++i) {
1561 Change const & change = lookupChange(i);
1562 if (change != running_change)
1563 flushString(os, write_buffer);
1564 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1565 running_change = change;
1570 // Write font changes
1571 Font font2 = getFontSettings(bparams, i);
1572 if (font2 != font1) {
1573 flushString(os, write_buffer);
1574 font2.lyxWriteChanges(font1, os);
1579 char_type const c = d->text_[i];
1582 if (Inset const * inset = getInset(i)) {
1583 flushString(os, write_buffer);
1584 if (inset->directWrite()) {
1585 // international char, let it write
1586 // code directly so it's shorter in
1592 os << "\\begin_inset ";
1594 os << "\n\\end_inset\n\n";
1597 // FIXME This can be removed again once the mystery
1598 // crash has been resolved.
1603 flushString(os, write_buffer);
1604 os << "\n\\backslash\n";
1608 flushString(os, write_buffer);
1609 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1616 if ((column > 70 && c == ' ')
1618 flushString(os, write_buffer);
1622 // this check is to amend a bug. LyX sometimes
1623 // inserts '\0' this could cause problems.
1625 write_buffer.push_back(c);
1627 LYXERR0("NUL char in structure.");
1633 flushString(os, write_buffer);
1634 os << "\n\\end_layout\n";
1635 // FIXME This can be removed again once the mystery
1636 // crash has been resolved.
1641 void Paragraph::validate(LaTeXFeatures & features) const
1643 d->validate(features);
1647 void Paragraph::insert(pos_type start, docstring const & str,
1648 Font const & font, Change const & change)
1650 for (size_t i = 0, n = str.size(); i != n ; ++i)
1651 insertChar(start + i, str[i], font, change);
1655 void Paragraph::appendChar(char_type c, Font const & font,
1656 Change const & change)
1659 d->changes_.insert(change, d->text_.size());
1660 // when appending characters, no need to update tables
1661 d->text_.push_back(c);
1662 setFont(d->text_.size() - 1, font);
1663 d->requestSpellCheck(d->text_.size() - 1);
1667 void Paragraph::appendString(docstring const & s, Font const & font,
1668 Change const & change)
1670 pos_type end = s.size();
1671 size_t oldsize = d->text_.size();
1672 size_t newsize = oldsize + end;
1673 size_t capacity = d->text_.capacity();
1674 if (newsize >= capacity)
1675 d->text_.reserve(max(capacity + 100, newsize));
1677 // when appending characters, no need to update tables
1680 // FIXME: Optimize this!
1681 for (size_t i = oldsize; i != newsize; ++i) {
1683 d->changes_.insert(change, i);
1684 d->requestSpellCheck(i);
1686 d->fontlist_.set(oldsize, font);
1687 d->fontlist_.set(newsize - 1, font);
1691 void Paragraph::insertChar(pos_type pos, char_type c,
1694 d->insertChar(pos, c, Change(trackChanges ?
1695 Change::INSERTED : Change::UNCHANGED));
1699 void Paragraph::insertChar(pos_type pos, char_type c,
1700 Font const & font, bool trackChanges)
1702 d->insertChar(pos, c, Change(trackChanges ?
1703 Change::INSERTED : Change::UNCHANGED));
1708 void Paragraph::insertChar(pos_type pos, char_type c,
1709 Font const & font, Change const & change)
1711 d->insertChar(pos, c, change);
1716 void Paragraph::resetFonts(Font const & font)
1718 d->fontlist_.clear();
1719 d->fontlist_.set(0, font);
1720 d->fontlist_.set(d->text_.size() - 1, font);
1723 // Gets uninstantiated font setting at position.
1724 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1728 LYXERR0("pos: " << pos << " size: " << size());
1732 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1733 if (cit != d->fontlist_.end())
1736 if (pos == size() && !empty())
1737 return getFontSettings(bparams, pos - 1);
1739 // Optimisation: avoid a full font instantiation if there is no
1740 // language change from previous call.
1741 static Font previous_font;
1742 static Language const * previous_lang = 0;
1743 Language const * lang = getParLanguage(bparams);
1744 if (lang != previous_lang) {
1745 previous_lang = lang;
1746 previous_font = Font(inherit_font, lang);
1748 return previous_font;
1752 FontSpan Paragraph::fontSpan(pos_type pos) const
1754 LBUFERR(pos <= size());
1757 return FontSpan(pos, pos);
1760 FontList::const_iterator cit = d->fontlist_.begin();
1761 FontList::const_iterator end = d->fontlist_.end();
1762 for (; cit != end; ++cit) {
1763 if (cit->pos() >= pos) {
1764 if (pos >= beginOfBody())
1765 return FontSpan(max(start, beginOfBody()),
1768 return FontSpan(start,
1769 min(beginOfBody() - 1,
1772 start = cit->pos() + 1;
1775 // This should not happen, but if so, we take no chances.
1776 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1777 LASSERT(false, return FontSpan(pos, pos));
1781 // Gets uninstantiated font setting at position 0
1782 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1784 if (!empty() && !d->fontlist_.empty())
1785 return d->fontlist_.begin()->font();
1787 // Optimisation: avoid a full font instantiation if there is no
1788 // language change from previous call.
1789 static Font previous_font;
1790 static Language const * previous_lang = 0;
1791 if (bparams.language != previous_lang) {
1792 previous_lang = bparams.language;
1793 previous_font = Font(inherit_font, bparams.language);
1796 return previous_font;
1800 // Gets the fully instantiated font at a given position in a paragraph
1801 // This is basically the same function as Text::GetFont() in text2.cpp.
1802 // The difference is that this one is used for generating the LaTeX file,
1803 // and thus cosmetic "improvements" are disallowed: This has to deliver
1804 // the true picture of the buffer. (Asger)
1805 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1806 Font const & outerfont) const
1810 Font font = getFontSettings(bparams, pos);
1812 pos_type const body_pos = beginOfBody();
1813 FontInfo & fi = font.fontInfo();
1815 fi.realize(d->layout_->labelfont);
1817 fi.realize(d->layout_->font);
1819 fi.realize(outerfont.fontInfo());
1820 fi.realize(bparams.getFont().fontInfo());
1826 Font const Paragraph::getLabelFont
1827 (BufferParams const & bparams, Font const & outerfont) const
1829 FontInfo tmpfont = d->layout_->labelfont;
1830 tmpfont.realize(outerfont.fontInfo());
1831 tmpfont.realize(bparams.getFont().fontInfo());
1832 return Font(tmpfont, getParLanguage(bparams));
1836 Font const Paragraph::getLayoutFont
1837 (BufferParams const & bparams, Font const & outerfont) const
1839 FontInfo tmpfont = d->layout_->font;
1840 tmpfont.realize(outerfont.fontInfo());
1841 tmpfont.realize(bparams.getFont().fontInfo());
1842 return Font(tmpfont, getParLanguage(bparams));
1846 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1848 char_type c = d->text_[pos];
1849 if (!getFontSettings(bparams, pos).isRightToLeft())
1852 // FIXME: The arabic special casing is due to the difference of arabic
1853 // round brackets input introduced in r18599. Check if this should be
1854 // unified with Hebrew or at least if all bracket types should be
1855 // handled the same (file format change in either case).
1856 string const & lang = getFontSettings(bparams, pos).language()->lang();
1857 bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
1862 uc = arabic ? c : ')';
1865 uc = arabic ? c : '(';
1891 void Paragraph::setFont(pos_type pos, Font const & font)
1893 LASSERT(pos <= size(), return);
1895 // First, reduce font against layout/label font
1896 // Update: The setCharFont() routine in text2.cpp already
1897 // reduces font, so we don't need to do that here. (Asger)
1899 d->fontlist_.set(pos, font);
1903 void Paragraph::makeSameLayout(Paragraph const & par)
1905 d->layout_ = par.d->layout_;
1906 d->params_ = par.d->params_;
1910 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1912 if (isFreeSpacing())
1918 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1919 if (eraseChar(pos, trackChanges))
1925 return count > 0 || pos > 0;
1929 bool Paragraph::hasSameLayout(Paragraph const & par) const
1931 return par.d->layout_ == d->layout_
1932 && d->params_.sameLayout(par.d->params_);
1936 depth_type Paragraph::getDepth() const
1938 return d->params_.depth();
1942 depth_type Paragraph::getMaxDepthAfter() const
1944 if (d->layout_->isEnvironment())
1945 return d->params_.depth() + 1;
1947 return d->params_.depth();
1951 char Paragraph::getAlign() const
1953 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1954 return d->layout_->align;
1956 return d->params_.align();
1960 docstring const & Paragraph::labelString() const
1962 return d->params_.labelString();
1966 // the next two functions are for the manual labels
1967 docstring const Paragraph::getLabelWidthString() const
1969 if (d->layout_->margintype == MARGIN_MANUAL
1970 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
1971 return d->params_.labelWidthString();
1973 return _("Senseless with this layout!");
1977 void Paragraph::setLabelWidthString(docstring const & s)
1979 d->params_.labelWidthString(s);
1983 docstring Paragraph::expandLabel(Layout const & layout,
1984 BufferParams const & bparams) const
1986 return expandParagraphLabel(layout, bparams, true);
1990 docstring Paragraph::expandDocBookLabel(Layout const & layout,
1991 BufferParams const & bparams) const
1993 return expandParagraphLabel(layout, bparams, false);
1997 docstring Paragraph::expandParagraphLabel(Layout const & layout,
1998 BufferParams const & bparams, bool process_appendix) const
2000 DocumentClass const & tclass = bparams.documentClass();
2001 string const & lang = getParLanguage(bparams)->code();
2002 bool const in_appendix = process_appendix && d->params_.appendix();
2003 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2005 if (fmt.empty() && !layout.counter.empty())
2006 return tclass.counters().theCounter(layout.counter, lang);
2008 // handle 'inherited level parts' in 'fmt',
2009 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2010 size_t const i = fmt.find('@', 0);
2011 if (i != docstring::npos) {
2012 size_t const j = fmt.find('@', i + 1);
2013 if (j != docstring::npos) {
2014 docstring parent(fmt, i + 1, j - i - 1);
2015 docstring label = from_ascii("??");
2016 if (tclass.hasLayout(parent))
2017 label = expandParagraphLabel(tclass[parent], bparams,
2019 fmt = docstring(fmt, 0, i) + label
2020 + docstring(fmt, j + 1, docstring::npos);
2024 return tclass.counters().counterLabel(fmt, lang);
2028 void Paragraph::applyLayout(Layout const & new_layout)
2030 d->layout_ = &new_layout;
2031 LyXAlignment const oldAlign = d->params_.align();
2033 if (!(oldAlign & d->layout_->alignpossible)) {
2034 frontend::Alert::warning(_("Alignment not permitted"),
2035 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2036 d->params_.align(LYX_ALIGN_LAYOUT);
2041 pos_type Paragraph::beginOfBody() const
2043 return d->begin_of_body_;
2047 void Paragraph::setBeginOfBody()
2049 if (d->layout_->labeltype != LABEL_MANUAL) {
2050 d->begin_of_body_ = 0;
2054 // Unroll the first two cycles of the loop
2055 // and remember the previous character to
2056 // remove unnecessary getChar() calls
2058 pos_type end = size();
2059 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2062 char_type previous_char = d->text_[i];
2063 if (!(isNewline(i) || isEnvSeparator(i))) {
2065 while (i < end && previous_char != ' ') {
2066 char_type temp = d->text_[i];
2067 if (isNewline(i) || isEnvSeparator(i))
2070 previous_char = temp;
2076 d->begin_of_body_ = i;
2080 bool Paragraph::allowParagraphCustomization() const
2082 return inInset().allowParagraphCustomization();
2086 bool Paragraph::usePlainLayout() const
2088 return inInset().usePlainLayout();
2092 bool Paragraph::isPassThru() const
2094 return inInset().isPassThru() || d->layout_->pass_thru;
2099 // paragraphs inside floats need different alignment tags to avoid
2102 bool noTrivlistCentering(InsetCode code)
2104 return code == FLOAT_CODE
2105 || code == WRAP_CODE
2106 || code == CELL_CODE;
2110 string correction(string const & orig)
2112 if (orig == "flushleft")
2113 return "raggedright";
2114 if (orig == "flushright")
2115 return "raggedleft";
2116 if (orig == "center")
2122 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2123 InsetCode code, bool const lastpar, int & col)
2125 string macro = suffix + "{";
2126 if (noTrivlistCentering(code)) {
2128 // the last paragraph in non-trivlist-aligned
2129 // context is special (to avoid unwanted whitespace)
2130 if (suffix == "\\begin") {
2131 macro = "\\" + correction(env) + "{}";
2132 os << from_ascii(macro);
2133 col += macro.size();
2138 macro += correction(env);
2142 if (suffix == "\\par\\end") {
2146 os << from_ascii(macro);
2147 col += macro.size();
2148 if (suffix == "\\begin") {
2158 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2159 otexstream & os, OutputParams const & runparams) const
2164 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2165 (layout_->toggle_indent != ITOGGLE_NEVER) :
2166 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2168 if (canindent && params_.noindent() && !layout_->pass_thru) {
2169 os << "\\noindent ";
2173 LyXAlignment const curAlign = params_.align();
2175 if (curAlign == layout_->align)
2179 case LYX_ALIGN_NONE:
2180 case LYX_ALIGN_BLOCK:
2181 case LYX_ALIGN_LAYOUT:
2182 case LYX_ALIGN_SPECIAL:
2183 case LYX_ALIGN_DECIMAL:
2185 case LYX_ALIGN_LEFT:
2186 case LYX_ALIGN_RIGHT:
2187 case LYX_ALIGN_CENTER:
2188 if (runparams.moving_arg) {
2195 string const begin_tag = "\\begin";
2196 InsetCode code = ownerCode();
2197 bool const lastpar = runparams.isLastPar;
2200 case LYX_ALIGN_NONE:
2201 case LYX_ALIGN_BLOCK:
2202 case LYX_ALIGN_LAYOUT:
2203 case LYX_ALIGN_SPECIAL:
2204 case LYX_ALIGN_DECIMAL:
2206 case LYX_ALIGN_LEFT: {
2207 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2208 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2210 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2212 } case LYX_ALIGN_RIGHT: {
2213 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2214 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2216 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2218 } case LYX_ALIGN_CENTER: {
2219 corrected_env(os, begin_tag, "center", code, lastpar, column);
2228 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2229 otexstream & os, OutputParams const & runparams) const
2231 LyXAlignment const curAlign = params_.align();
2233 if (curAlign == layout_->align)
2237 case LYX_ALIGN_NONE:
2238 case LYX_ALIGN_BLOCK:
2239 case LYX_ALIGN_LAYOUT:
2240 case LYX_ALIGN_SPECIAL:
2241 case LYX_ALIGN_DECIMAL:
2243 case LYX_ALIGN_LEFT:
2244 case LYX_ALIGN_RIGHT:
2245 case LYX_ALIGN_CENTER:
2246 if (runparams.moving_arg)
2251 bool output = false;
2253 string const end_tag = "\\par\\end";
2254 InsetCode code = ownerCode();
2255 bool const lastpar = runparams.isLastPar;
2258 case LYX_ALIGN_NONE:
2259 case LYX_ALIGN_BLOCK:
2260 case LYX_ALIGN_LAYOUT:
2261 case LYX_ALIGN_SPECIAL:
2262 case LYX_ALIGN_DECIMAL:
2264 case LYX_ALIGN_LEFT: {
2265 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2266 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2268 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2270 } case LYX_ALIGN_RIGHT: {
2271 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2272 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2274 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2276 } case LYX_ALIGN_CENTER: {
2277 corrected_env(os, end_tag, "center", code, lastpar, col);
2282 return output || lastpar;
2286 // This one spits out the text of the paragraph
2287 void Paragraph::latex(BufferParams const & bparams,
2288 Font const & outerfont,
2290 OutputParams const & runparams,
2291 int start_pos, int end_pos, bool force) const
2293 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2295 // FIXME This check should not be needed. Perhaps issue an
2296 // error if it triggers.
2297 Layout const & style = inInset().forcePlainLayout() ?
2298 bparams.documentClass().plainLayout() : *d->layout_;
2300 if (!force && style.inpreamble)
2303 bool const allowcust = allowParagraphCustomization();
2305 // Current base font for all inherited font changes, without any
2306 // change caused by an individual character, except for the language:
2307 // It is set to the language of the first character.
2308 // As long as we are in the label, this font is the base font of the
2309 // label. Before the first body character it is set to the base font
2313 // Maybe we have to create a optional argument.
2314 pos_type body_pos = beginOfBody();
2315 unsigned int column = 0;
2318 // the optional argument is kept in curly brackets in
2319 // case it contains a ']'
2320 // This is not strictly needed, but if this is changed it
2321 // would be a file format change, and tex2lyx would need
2322 // to be adjusted, since it unconditionally removes the
2323 // braces when it parses \item.
2326 basefont = getLabelFont(bparams, outerfont);
2328 basefont = getLayoutFont(bparams, outerfont);
2331 // Which font is currently active?
2332 Font running_font(basefont);
2333 // Do we have an open font change?
2334 bool open_font = false;
2336 Change runningChange = Change(Change::UNCHANGED);
2338 Encoding const * const prev_encoding = runparams.encoding;
2340 os.texrow().start(id(), 0);
2342 // if the paragraph is empty, the loop will not be entered at all
2344 if (style.isCommand()) {
2348 if (!style.leftdelim().empty()) {
2349 os << style.leftdelim();
2350 column += style.leftdelim().size();
2353 column += d->startTeXParParams(bparams, os, runparams);
2356 for (pos_type i = 0; i < size(); ++i) {
2357 // First char in paragraph or after label?
2358 if (i == body_pos) {
2361 column += running_font.latexWriteEndChanges(
2362 os, bparams, runparams,
2363 basefont, basefont);
2366 basefont = getLayoutFont(bparams, outerfont);
2367 running_font = basefont;
2369 column += Changes::latexMarkChange(os, bparams,
2370 runningChange, Change(Change::UNCHANGED),
2372 runningChange = Change(Change::UNCHANGED);
2377 if (style.isCommand()) {
2382 if (!style.leftdelim().empty()) {
2383 os << style.leftdelim();
2384 column += style.leftdelim().size();
2388 column += d->startTeXParParams(bparams, os,
2392 runparams.wasDisplayMath = runparams.inDisplayMath;
2393 runparams.inDisplayMath = false;
2394 bool deleted_display_math = false;
2396 // Check whether a display math inset follows
2397 if (d->text_[i] == META_INSET
2398 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2399 InsetMath const * im = getInset(i)->asInsetMath();
2400 if (im && im->asHullInset()
2401 && im->asHullInset()->outerDisplay()) {
2402 runparams.inDisplayMath = true;
2403 // runparams.inDeletedInset will be set by
2404 // latexInset later, but we need this info
2405 // before it is called. On the other hand, we
2406 // cannot set it here because it is a counter.
2407 deleted_display_math = isDeleted(i);
2411 Change const & change = runparams.inDeletedInset
2412 ? runparams.changeOfDeletedInset : lookupChange(i);
2414 if (bparams.output_changes && runningChange != change) {
2416 column += running_font.latexWriteEndChanges(
2417 os, bparams, runparams, basefont, basefont);
2420 basefont = getLayoutFont(bparams, outerfont);
2421 running_font = basefont;
2422 column += Changes::latexMarkChange(os, bparams, runningChange,
2424 runningChange = change;
2427 // do not output text which is marked deleted
2428 // if change tracking output is disabled
2429 if (!bparams.output_changes && change.deleted()) {
2435 // Fully instantiated font
2436 Font const current_font = getFont(bparams, i, outerfont);
2438 Font const last_font = running_font;
2440 // Do we need to close the previous font?
2442 (current_font != running_font ||
2443 current_font.language() != running_font.language()))
2445 column += running_font.latexWriteEndChanges(
2446 os, bparams, runparams, basefont,
2447 (i == body_pos-1) ? basefont : current_font);
2448 running_font = basefont;
2452 string const running_lang = runparams.use_polyglossia ?
2453 running_font.language()->polyglossia() : running_font.language()->babel();
2454 // close babel's font environment before opening CJK.
2455 string const lang_end_command = runparams.use_polyglossia ?
2456 "\\end{$$lang}" : lyxrc.language_command_end;
2457 if (!running_lang.empty() &&
2458 current_font.language()->encoding()->package() == Encoding::CJK) {
2459 string end_tag = subst(lang_end_command,
2462 os << from_ascii(end_tag);
2463 column += end_tag.length();
2464 if (runparams.use_polyglossia)
2465 popPolyglossiaLang();
2468 // Switch file encoding if necessary (and allowed)
2469 if (!runparams.pass_thru && !style.pass_thru &&
2470 runparams.encoding->package() != Encoding::none &&
2471 current_font.language()->encoding()->package() != Encoding::none) {
2472 pair<bool, int> const enc_switch =
2473 switchEncoding(os.os(), bparams, runparams,
2474 *(current_font.language()->encoding()));
2475 if (enc_switch.first) {
2476 column += enc_switch.second;
2477 runparams.encoding = current_font.language()->encoding();
2481 char_type const c = d->text_[i];
2483 // A display math inset inside an ulem command will be output
2484 // as a box of width \columnwidth, so we have to either disable
2485 // indentation if the inset starts a paragraph, or start a new
2486 // line to accommodate such box. This has to be done before
2487 // writing any font changing commands.
2488 if (runparams.inDisplayMath && !deleted_display_math
2489 && runparams.inulemcmd) {
2490 if (os.afterParbreak())
2496 // Do we need to change font?
2497 if ((current_font != running_font ||
2498 current_font.language() != running_font.language()) &&
2501 odocstringstream ods;
2502 column += current_font.latexWriteStartChanges(ods, bparams,
2503 runparams, basefont,
2505 // Check again for display math in ulem commands as a
2506 // font change may also occur just before a math inset.
2507 if (runparams.inDisplayMath && !deleted_display_math
2508 && runparams.inulemcmd) {
2509 if (os.afterParbreak())
2514 running_font = current_font;
2516 docstring fontchange = ods.str();
2517 // check whether the fontchange ends with a \\textcolor
2518 // modifier and the text starts with a space (bug 4473)
2519 docstring const last_modifier = rsplit(fontchange, '\\');
2520 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2521 os << fontchange << from_ascii("{}");
2522 // check if the fontchange ends with a trailing blank
2523 // (like "\small " (see bug 3382)
2524 else if (suffixIs(fontchange, ' ') && c == ' ')
2525 os << fontchange.substr(0, fontchange.size() - 1)
2526 << from_ascii("{}");
2531 // FIXME: think about end_pos implementation...
2532 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2533 // FIXME: integrate this case in latexSpecialChar
2534 // Do not print the separation of the optional argument
2535 // if style.pass_thru is false. This works because
2536 // latexSpecialChar ignores spaces if
2537 // style.pass_thru is false.
2538 if (i != body_pos - 1) {
2539 if (d->simpleTeXBlanks(runparams, os,
2540 i, column, current_font, style)) {
2541 // A surrogate pair was output. We
2542 // must not call latexSpecialChar
2543 // in this iteration, since it would output
2544 // the combining character again.
2551 OutputParams rp = runparams;
2552 rp.free_spacing = style.free_spacing;
2553 rp.local_font = ¤t_font;
2554 rp.intitle = style.intitle;
2556 // Two major modes: LaTeX or plain
2557 // Handle here those cases common to both modes
2558 // and then split to handle the two modes separately.
2559 if (c == META_INSET) {
2560 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2561 d->latexInset(bparams, os, rp, running_font,
2562 basefont, outerfont, open_font,
2563 runningChange, style, i, column);
2565 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2567 d->latexSpecialChar(os, bparams, rp,
2568 running_font, runningChange,
2569 style, i, end_pos, column);
2570 } catch (EncodingException & e) {
2571 if (runparams.dryrun) {
2572 os << "<" << _("LyX Warning: ")
2573 << _("uncodable character") << " '";
2577 // add location information and throw again.
2585 // Set the encoding to that returned from latexSpecialChar (see
2586 // comment for encoding member in OutputParams.h)
2587 runparams.encoding = rp.encoding;
2589 // Also carry on the info on a closed ulem command for insets
2590 // such as Note that do not produce any output, so that no
2591 // command is ever executed but its opening was recorded.
2592 runparams.inulemcmd = rp.inulemcmd;
2595 // If we have an open font definition, we have to close it
2597 #ifdef FIXED_LANGUAGE_END_DETECTION
2599 running_font.latexWriteEndChanges(os, bparams,
2600 runparams, basefont,
2601 next_->getFont(bparams, 0, outerfont));
2603 running_font.latexWriteEndChanges(os, bparams,
2604 runparams, basefont, basefont);
2607 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2608 //FIXME: there as we start another \selectlanguage with the next paragraph if
2609 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2610 running_font.latexWriteEndChanges(os, bparams, runparams,
2611 basefont, basefont);
2615 column += Changes::latexMarkChange(os, bparams, runningChange,
2616 Change(Change::UNCHANGED), runparams);
2618 // Needed if there is an optional argument but no contents.
2619 if (body_pos > 0 && body_pos == size()) {
2623 if (!style.rightdelim().empty()) {
2624 os << style.rightdelim();
2625 column += style.rightdelim().size();
2628 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2629 && runparams.encoding != prev_encoding) {
2630 runparams.encoding = prev_encoding;
2631 os << setEncoding(prev_encoding->iconvName());
2634 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2638 bool Paragraph::emptyTag() const
2640 for (pos_type i = 0; i < size(); ++i) {
2641 if (Inset const * inset = getInset(i)) {
2642 InsetCode lyx_code = inset->lyxCode();
2643 // FIXME testing like that is wrong. What is
2645 if (lyx_code != TOC_CODE &&
2646 lyx_code != INCLUDE_CODE &&
2647 lyx_code != GRAPHICS_CODE &&
2648 lyx_code != ERT_CODE &&
2649 lyx_code != LISTINGS_CODE &&
2650 lyx_code != FLOAT_CODE &&
2651 lyx_code != TABULAR_CODE) {
2655 char_type c = d->text_[i];
2656 if (c != ' ' && c != '\t')
2664 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2667 for (pos_type i = 0; i < size(); ++i) {
2668 if (Inset const * inset = getInset(i)) {
2669 InsetCode lyx_code = inset->lyxCode();
2670 if (lyx_code == LABEL_CODE) {
2671 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2672 docstring const & id = il->getParam("name");
2673 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2681 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2685 for (i = 0; i < size(); ++i) {
2686 if (Inset const * inset = getInset(i)) {
2687 inset->docbook(os, runparams);
2689 char_type c = d->text_[i];
2692 os << sgml::escapeChar(c);
2699 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2703 for (i = 0; i < size(); ++i) {
2704 if (Inset const * inset = getInset(i)) {
2705 inset->xhtml(xs, runparams);
2707 char_type c = d->text_[i];
2717 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2720 pos_type size = text_.size();
2721 for (pos_type i = initial; i < size; ++i) {
2722 Font font = owner_->getFont(buf.params(), i, outerfont);
2723 if (text_[i] == META_INSET)
2725 if (i != initial && font != font_old)
2734 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2736 OutputParams const & runparams,
2737 Font const & outerfont,
2738 pos_type initial) const
2740 bool emph_flag = false;
2742 Layout const & style = *d->layout_;
2744 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2746 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2749 // parsing main loop
2750 for (pos_type i = initial; i < size(); ++i) {
2751 Font font = getFont(buf.params(), i, outerfont);
2753 // handle <emphasis> tag
2754 if (font_old.emph() != font.fontInfo().emph()) {
2755 if (font.fontInfo().emph() == FONT_ON) {
2758 } else if (i != initial) {
2759 os << "</emphasis>";
2764 if (Inset const * inset = getInset(i)) {
2765 inset->docbook(os, runparams);
2767 char_type c = d->text_[i];
2769 if (style.pass_thru)
2772 os << sgml::escapeChar(c);
2774 font_old = font.fontInfo();
2778 os << "</emphasis>";
2781 if (style.free_spacing)
2783 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2789 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2790 vector<html::EndFontTag> & tagsToClose,
2791 bool & flag, FontState curstate, html::FontTypes type)
2793 if (curstate == FONT_ON) {
2794 tagsToOpen.push_back(html::FontTag(type));
2797 tagsToClose.push_back(html::EndFontTag(type));
2804 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2806 OutputParams const & runparams,
2807 Font const & outerfont,
2808 bool start_paragraph, bool close_paragraph,
2809 pos_type initial) const
2813 // track whether we have opened these tags
2814 bool emph_flag = false;
2815 bool bold_flag = false;
2816 bool noun_flag = false;
2817 bool ubar_flag = false;
2818 bool dbar_flag = false;
2819 bool sout_flag = false;
2820 bool wave_flag = false;
2822 bool shap_flag = false;
2824 bool faml_flag = false;
2826 bool size_flag = false;
2828 Layout const & style = *d->layout_;
2830 if (start_paragraph)
2831 xs.startDivision(allowEmpty());
2834 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2836 FontShape curr_fs = INHERIT_SHAPE;
2837 FontFamily curr_fam = INHERIT_FAMILY;
2838 FontSize curr_size = FONT_SIZE_INHERIT;
2840 string const default_family =
2841 buf.masterBuffer()->params().fonts_default_family;
2843 vector<html::FontTag> tagsToOpen;
2844 vector<html::EndFontTag> tagsToClose;
2846 // parsing main loop
2847 for (pos_type i = initial; i < size(); ++i) {
2848 // let's not show deleted material in the output
2852 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
2855 FontState curstate = font.fontInfo().emph();
2856 if (font_old.emph() != curstate)
2857 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
2860 curstate = font.fontInfo().noun();
2861 if (font_old.noun() != curstate)
2862 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
2865 curstate = font.fontInfo().underbar();
2866 if (font_old.underbar() != curstate)
2867 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
2870 curstate = font.fontInfo().strikeout();
2871 if (font_old.strikeout() != curstate)
2872 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
2875 curstate = font.fontInfo().uuline();
2876 if (font_old.uuline() != curstate)
2877 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
2880 curstate = font.fontInfo().uwave();
2881 if (font_old.uwave() != curstate)
2882 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
2885 // a little hackish, but allows us to reuse what we have.
2886 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
2887 if (font_old.series() != font.fontInfo().series())
2888 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
2891 curr_fs = font.fontInfo().shape();
2892 FontShape old_fs = font_old.shape();
2893 if (old_fs != curr_fs) {
2897 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
2900 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
2902 case SMALLCAPS_SHAPE:
2903 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
2909 // the other tags are for internal use
2917 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
2921 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
2924 case SMALLCAPS_SHAPE:
2925 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
2932 // the other tags are for internal use
2939 curr_fam = font.fontInfo().family();
2940 FontFamily old_fam = font_old.family();
2941 if (old_fam != curr_fam) {
2945 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
2948 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
2950 case TYPEWRITER_FAMILY:
2951 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
2953 case INHERIT_FAMILY:
2956 // the other tags are for internal use
2964 // we will treat a "default" font family as roman, since we have
2965 // no other idea what to do.
2966 if (default_family != "rmdefault" && default_family != "default") {
2967 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
2972 if (default_family != "sfdefault") {
2973 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
2977 case TYPEWRITER_FAMILY:
2978 if (default_family != "ttdefault") {
2979 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
2983 case INHERIT_FAMILY:
2986 // the other tags are for internal use
2993 curr_size = font.fontInfo().size();
2994 FontSize old_size = font_old.size();
2995 if (old_size != curr_size) {
2998 case FONT_SIZE_TINY:
2999 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3001 case FONT_SIZE_SCRIPT:
3002 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3004 case FONT_SIZE_FOOTNOTE:
3005 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3007 case FONT_SIZE_SMALL:
3008 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3010 case FONT_SIZE_LARGE:
3011 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3013 case FONT_SIZE_LARGER:
3014 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3016 case FONT_SIZE_LARGEST:
3017 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3019 case FONT_SIZE_HUGE:
3020 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3022 case FONT_SIZE_HUGER:
3023 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3025 case FONT_SIZE_INCREASE:
3026 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3028 case FONT_SIZE_DECREASE:
3029 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3031 case FONT_SIZE_INHERIT:
3032 case FONT_SIZE_NORMAL:
3035 // the other tags are for internal use
3041 switch (curr_size) {
3042 case FONT_SIZE_TINY:
3043 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3046 case FONT_SIZE_SCRIPT:
3047 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3050 case FONT_SIZE_FOOTNOTE:
3051 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3054 case FONT_SIZE_SMALL:
3055 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3058 case FONT_SIZE_LARGE:
3059 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3062 case FONT_SIZE_LARGER:
3063 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3066 case FONT_SIZE_LARGEST:
3067 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3070 case FONT_SIZE_HUGE:
3071 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3074 case FONT_SIZE_HUGER:
3075 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3078 case FONT_SIZE_INCREASE:
3079 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3082 case FONT_SIZE_DECREASE:
3083 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3086 case FONT_SIZE_NORMAL:
3087 case FONT_SIZE_INHERIT:
3090 // the other tags are for internal use
3097 // Other such tags? What about the other text ranges?
3099 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3100 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3101 for (; cit != cen; ++cit)
3104 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3105 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3106 for (; sit != sen; ++sit)
3109 tagsToClose.clear();
3112 Inset const * inset = getInset(i);
3114 if (!runparams.for_toc || inset->isInToc()) {
3115 OutputParams np = runparams;
3116 np.local_font = &font;
3117 // If the paragraph has size 1, then we are in the "special
3118 // case" where we do not output the containing paragraph info
3119 if (!inset->getLayout().htmlisblock() && size() != 1)
3120 np.html_in_par = true;
3121 retval += inset->xhtml(xs, np);
3124 char_type c = getUChar(buf.masterBuffer()->params(), i);
3127 font_old = font.fontInfo();
3131 // I'm worried about what happens if a branch, say, is itself
3132 // wrapped in some font stuff. I think that will not work.
3134 if (close_paragraph)
3141 bool Paragraph::isHfill(pos_type pos) const
3143 Inset const * inset = getInset(pos);
3144 return inset && inset->isHfill();
3148 bool Paragraph::isNewline(pos_type pos) const
3150 // U+2028 LINE SEPARATOR
3151 // U+2029 PARAGRAPH SEPARATOR
3152 char_type const c = d->text_[pos];
3153 if (c == 0x2028 || c == 0x2029)
3155 Inset const * inset = getInset(pos);
3156 return inset && inset->lyxCode() == NEWLINE_CODE;
3160 bool Paragraph::isEnvSeparator(pos_type pos) const
3162 Inset const * inset = getInset(pos);
3163 return inset && inset->lyxCode() == SEPARATOR_CODE;
3167 bool Paragraph::isLineSeparator(pos_type pos) const
3169 char_type const c = d->text_[pos];
3170 if (isLineSeparatorChar(c))
3172 Inset const * inset = getInset(pos);
3173 return inset && inset->isLineSeparator();
3177 bool Paragraph::isWordSeparator(pos_type pos) const
3181 if (Inset const * inset = getInset(pos))
3182 return !inset->isLetter();
3183 // if we have a hard hyphen (no en- or emdash) or apostrophe
3184 // we pass this to the spell checker
3185 // FIXME: this method is subject to change, visit
3186 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3187 // to get an impression how complex this is.
3188 if (isHardHyphenOrApostrophe(pos))
3190 char_type const c = d->text_[pos];
3191 // We want to pass the escape chars to the spellchecker
3192 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3193 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3197 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3199 pos_type const psize = size();
3202 char_type const c = d->text_[pos];
3203 if (c != '-' && c != '\'')
3205 int nextpos = pos + 1;
3206 int prevpos = pos > 0 ? pos - 1 : 0;
3207 if ((nextpos == psize || isSpace(nextpos))
3208 && (pos == 0 || isSpace(prevpos)))
3214 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3216 return d->speller_state_.getRange(pos);
3220 bool Paragraph::isChar(pos_type pos) const
3222 if (Inset const * inset = getInset(pos))
3223 return inset->isChar();
3224 char_type const c = d->text_[pos];
3225 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3229 bool Paragraph::isSpace(pos_type pos) const
3231 if (Inset const * inset = getInset(pos))
3232 return inset->isSpace();
3233 char_type const c = d->text_[pos];
3234 return lyx::isSpace(c);
3239 Paragraph::getParLanguage(BufferParams const & bparams) const
3242 return getFirstFontSettings(bparams).language();
3243 // FIXME: we should check the prev par as well (Lgb)
3244 return bparams.language;
3248 bool Paragraph::isRTL(BufferParams const & bparams) const
3250 return getParLanguage(bparams)->rightToLeft()
3251 && !inInset().getLayout().forceLTR();
3255 void Paragraph::changeLanguage(BufferParams const & bparams,
3256 Language const * from, Language const * to)
3258 // change language including dummy font change at the end
3259 for (pos_type i = 0; i <= size(); ++i) {
3260 Font font = getFontSettings(bparams, i);
3261 if (font.language() == from) {
3262 font.setLanguage(to);
3264 d->requestSpellCheck(i);
3270 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3272 Language const * doc_language = bparams.language;
3273 FontList::const_iterator cit = d->fontlist_.begin();
3274 FontList::const_iterator end = d->fontlist_.end();
3276 for (; cit != end; ++cit)
3277 if (cit->font().language() != ignore_language &&
3278 cit->font().language() != latex_language &&
3279 cit->font().language() != doc_language)
3285 void Paragraph::getLanguages(std::set<Language const *> & languages) const
3287 FontList::const_iterator cit = d->fontlist_.begin();
3288 FontList::const_iterator end = d->fontlist_.end();
3290 for (; cit != end; ++cit) {
3291 Language const * lang = cit->font().language();
3292 if (lang != ignore_language &&
3293 lang != latex_language)
3294 languages.insert(lang);
3299 docstring Paragraph::asString(int options) const
3301 return asString(0, size(), options);
3305 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3307 odocstringstream os;
3310 && options & AS_STR_LABEL
3311 && !d->params_.labelString().empty())
3312 os << d->params_.labelString() << ' ';
3314 for (pos_type i = beg; i < end; ++i) {
3315 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3317 char_type const c = d->text_[i];
3318 if (isPrintable(c) || c == '\t'
3319 || (c == '\n' && (options & AS_STR_NEWLINES)))
3321 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3322 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3323 LASSERT(runparams != 0, return docstring());
3324 getInset(i)->plaintext(os, *runparams);
3326 getInset(i)->toString(os);
3335 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3336 bool const shorten) const
3338 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3339 if (!d->params_.labelString().empty())
3340 os += d->params_.labelString() + ' ';
3341 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3344 char_type const c = d->text_[i];
3347 else if (c == META_INSET)
3348 getInset(i)->forOutliner(os, tmplen, false);
3351 Text::shortenForOutliner(os, maxlen);
3355 void Paragraph::setInsetOwner(Inset const * inset)
3357 d->inset_owner_ = inset;
3361 int Paragraph::id() const
3367 void Paragraph::setId(int id)
3373 Layout const & Paragraph::layout() const
3379 void Paragraph::setLayout(Layout const & layout)
3381 d->layout_ = &layout;
3385 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3387 setLayout(tc.defaultLayout());
3391 void Paragraph::setPlainLayout(DocumentClass const & tc)
3393 setLayout(tc.plainLayout());
3397 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3399 if (usePlainLayout())
3400 setPlainLayout(tclass);
3402 setDefaultLayout(tclass);
3406 Inset const & Paragraph::inInset() const
3408 LBUFERR(d->inset_owner_);
3409 return *d->inset_owner_;
3413 ParagraphParameters & Paragraph::params()
3419 ParagraphParameters const & Paragraph::params() const
3425 bool Paragraph::isFreeSpacing() const
3427 if (d->layout_->free_spacing)
3429 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3433 bool Paragraph::allowEmpty() const
3435 if (d->layout_->keepempty)
3437 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3441 bool Paragraph::brokenBiblio() const
3443 // there is a problem if there is no bibitem at position 0 or
3444 // if there is another bibitem in the paragraph.
3445 return d->layout_->labeltype == LABEL_BIBLIO
3446 && (d->insetlist_.find(BIBITEM_CODE) != 0
3447 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3451 int Paragraph::fixBiblio(Buffer const & buffer)
3453 // FIXME: What about the case where paragraph is not BIBLIO
3454 // but there is an InsetBibitem?
3455 // FIXME: when there was already an inset at 0, the return value is 1,
3456 // which does not tell whether another inset has been remove; the
3457 // cursor cannot be correctly updated.
3459 if (d->layout_->labeltype != LABEL_BIBLIO)
3462 bool const track_changes = buffer.params().track_changes;
3463 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3464 bool const hasbibitem0 = bibitem_pos == 0;
3467 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3468 // There was an InsetBibitem at pos 0, and no other one => OK
3469 if (bibitem_pos == -1)
3471 // there is a bibitem at the 0 position, but since
3472 // there is a second one, we copy the second on the
3473 // first. We're assuming there are at most two of
3474 // these, which there should be.
3475 // FIXME: why does it make sense to do that rather
3476 // than keep the first? (JMarc)
3477 Inset * inset = releaseInset(bibitem_pos);
3478 d->insetlist_.begin()->inset = inset;
3479 return -bibitem_pos;
3482 // We need to create an inset at the beginning
3484 if (bibitem_pos > 0) {
3485 // there was one somewhere in the paragraph, let's move it
3486 inset = d->insetlist_.release(bibitem_pos);
3487 eraseChar(bibitem_pos, track_changes);
3490 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3491 InsetCommandParams(BIBITEM_CODE));
3493 Font font(inherit_font, buffer.params().language);
3494 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3495 : Change::UNCHANGED));
3501 void Paragraph::checkAuthors(AuthorList const & authorList)
3503 d->changes_.checkAuthors(authorList);
3507 bool Paragraph::isChanged(pos_type pos) const
3509 return lookupChange(pos).changed();
3513 bool Paragraph::isInserted(pos_type pos) const
3515 return lookupChange(pos).inserted();
3519 bool Paragraph::isDeleted(pos_type pos) const
3521 return lookupChange(pos).deleted();
3525 InsetList const & Paragraph::insetList() const
3527 return d->insetlist_;
3531 void Paragraph::setBuffer(Buffer & b)
3533 d->insetlist_.setBuffer(b);
3537 void Paragraph::resetBuffer()
3539 d->insetlist_.resetBuffer();
3543 Inset * Paragraph::releaseInset(pos_type pos)
3545 Inset * inset = d->insetlist_.release(pos);
3546 /// does not honour change tracking!
3547 eraseChar(pos, false);
3552 Inset * Paragraph::getInset(pos_type pos)
3554 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3555 ? d->insetlist_.get(pos) : 0;
3559 Inset const * Paragraph::getInset(pos_type pos) const
3561 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3562 ? d->insetlist_.get(pos) : 0;
3566 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3567 pos_type & right, TextCase action)
3569 // process sequences of modified characters; in change
3570 // tracking mode, this approach results in much better
3571 // usability than changing case on a char-by-char basis
3572 // We also need to track the current font, since font
3573 // changes within sequences can occur.
3574 vector<pair<char_type, Font> > changes;
3576 bool const trackChanges = bparams.track_changes;
3578 bool capitalize = true;
3580 for (; pos < right; ++pos) {
3581 char_type oldChar = d->text_[pos];
3582 char_type newChar = oldChar;
3584 // ignore insets and don't play with deleted text!
3585 if (oldChar != META_INSET && !isDeleted(pos)) {
3587 case text_lowercase:
3588 newChar = lowercase(oldChar);
3590 case text_capitalization:
3592 newChar = uppercase(oldChar);
3596 case text_uppercase:
3597 newChar = uppercase(oldChar);
3602 if (isWordSeparator(pos) || isDeleted(pos)) {
3603 // permit capitalization again
3607 if (oldChar != newChar) {
3608 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3609 if (pos != right - 1)
3611 // step behind the changing area
3615 int erasePos = pos - changes.size();
3616 for (size_t i = 0; i < changes.size(); i++) {
3617 insertChar(pos, changes[i].first,
3620 if (!eraseChar(erasePos, trackChanges)) {
3623 ++right; // expand selection
3631 int Paragraph::find(docstring const & str, bool cs, bool mw,
3632 pos_type start_pos, bool del) const
3634 pos_type pos = start_pos;
3635 int const strsize = str.length();
3637 pos_type const parsize = d->text_.size();
3638 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3639 // Ignore "invisible" letters such as ligature breaks
3640 // and hyphenation chars while searching
3641 while (pos < parsize - 1 && isInset(pos)) {
3642 odocstringstream os;
3643 getInset(pos)->toString(os);
3644 if (!getInset(pos)->isLetter() || !os.str().empty())
3648 if (cs && str[i] != d->text_[pos])
3650 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3652 if (!del && isDeleted(pos))
3659 // if necessary, check whether string matches word
3661 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3664 && !isWordSeparator(pos))
3668 return pos - start_pos;
3672 char_type Paragraph::getChar(pos_type pos) const
3674 return d->text_[pos];
3678 pos_type Paragraph::size() const
3680 return d->text_.size();
3684 bool Paragraph::empty() const
3686 return d->text_.empty();
3690 bool Paragraph::isInset(pos_type pos) const
3692 return d->text_[pos] == META_INSET;
3696 bool Paragraph::isSeparator(pos_type pos) const
3698 //FIXME: Are we sure this can be the only separator?
3699 return d->text_[pos] == ' ';
3703 void Paragraph::deregisterWords()
3705 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3706 Private::LangWordsMap::const_iterator ite = d->words_.end();
3707 for (; itl != ite; ++itl) {
3708 WordList * wl = theWordList(itl->first);
3709 Private::Words::const_iterator it = (itl->second).begin();
3710 Private::Words::const_iterator et = (itl->second).end();
3711 for (; it != et; ++it)
3718 void Paragraph::locateWord(pos_type & from, pos_type & to,
3719 word_location const loc) const
3722 case WHOLE_WORD_STRICT:
3723 if (from == 0 || from == size()
3724 || isWordSeparator(from)
3725 || isWordSeparator(from - 1)) {
3729 // no break here, we go to the next
3732 // If we are already at the beginning of a word, do nothing
3733 if (!from || isWordSeparator(from - 1))
3735 // no break here, we go to the next
3738 // always move the cursor to the beginning of previous word
3739 while (from && !isWordSeparator(from - 1))
3743 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3746 // no need to move the 'from' cursor
3750 while (to < size() && !isWordSeparator(to))
3755 void Paragraph::collectWords()
3757 for (pos_type pos = 0; pos < size(); ++pos) {
3758 if (isWordSeparator(pos))
3760 pos_type from = pos;
3761 locateWord(from, pos, WHOLE_WORD);
3762 // Work around MSVC warning: The statement
3763 // if (pos < from + lyxrc.completion_minlength)
3764 // triggers a signed vs. unsigned warning.
3765 // I don't know why this happens, it could be a MSVC bug, or
3766 // related to LLP64 (windows) vs. LP64 (unix) programming
3767 // model, or the C++ standard might be ambigous in the section
3768 // defining the "usual arithmetic conversions". However, using
3769 // a temporary variable is safe and works on all compilers.
3770 pos_type const endpos = from + lyxrc.completion_minlength;
3773 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
3774 if (cit == d->fontlist_.end())
3776 Language const * lang = cit->font().language();
3777 docstring const word = asString(from, pos, AS_STR_NONE);
3778 d->words_[lang->lang()].insert(word);
3783 void Paragraph::registerWords()
3785 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3786 Private::LangWordsMap::const_iterator ite = d->words_.end();
3787 for (; itl != ite; ++itl) {
3788 WordList * wl = theWordList(itl->first);
3789 Private::Words::const_iterator it = (itl->second).begin();
3790 Private::Words::const_iterator et = (itl->second).end();
3791 for (; it != et; ++it)
3797 void Paragraph::updateWords()
3805 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3807 SkipPositionsIterator begin = skips.begin();
3808 SkipPositions::iterator end = skips.end();
3809 if (pos > 0 && begin < end) {
3811 if (end->last == pos - 1) {
3816 skips.insert(end, FontSpan(pos, pos));
3820 Language * Paragraph::Private::locateSpellRange(
3821 pos_type & from, pos_type & to,
3822 SkipPositions & skips) const
3824 // skip leading white space
3825 while (from < to && owner_->isWordSeparator(from))
3827 // don't check empty range
3830 // get current language
3831 Language * lang = getSpellLanguage(from);
3832 pos_type last = from;
3833 bool samelang = true;
3834 bool sameinset = true;
3835 while (last < to && samelang && sameinset) {
3836 // hop to end of word
3837 while (last < to && !owner_->isWordSeparator(last)) {
3838 if (owner_->getInset(last)) {
3839 appendSkipPosition(skips, last);
3840 } else if (owner_->isDeleted(last)) {
3841 appendSkipPosition(skips, last);
3845 // hop to next word while checking for insets
3846 while (sameinset && last < to && owner_->isWordSeparator(last)) {
3847 if (Inset const * inset = owner_->getInset(last))
3848 sameinset = inset->isChar() && inset->isLetter();
3849 if (sameinset && owner_->isDeleted(last)) {
3850 appendSkipPosition(skips, last);
3855 if (sameinset && last < to) {
3856 // now check for language change
3857 samelang = lang == getSpellLanguage(last);
3860 // if language change detected backstep is needed
3868 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
3871 const_cast<Language *>(owner_->getFontSettings(
3872 inset_owner_->buffer().params(), from).language());
3873 if (lang == inset_owner_->buffer().params().language
3874 && !lyxrc.spellchecker_alt_lang.empty()) {
3876 string const lang_variety =
3877 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3878 lang->setCode(lang_code);
3879 lang->setVariety(lang_variety);
3885 void Paragraph::requestSpellCheck(pos_type pos)
3887 d->requestSpellCheck(pos);
3891 bool Paragraph::needsSpellCheck() const
3893 SpellChecker::ChangeNumber speller_change_number = 0;
3894 if (theSpellChecker())
3895 speller_change_number = theSpellChecker()->changeNumber();
3896 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
3897 d->speller_state_.needsCompleteRefresh(speller_change_number);
3899 return d->needsSpellCheck();
3903 bool Paragraph::Private::ignoreWord(docstring const & word) const
3905 // Ignore words with digits
3906 // FIXME: make this customizable
3907 // (note that some checkers ignore words with digits by default)
3908 docstring::const_iterator cit = word.begin();
3909 docstring::const_iterator const end = word.end();
3910 for (; cit != end; ++cit) {
3911 if (isNumber((*cit)))
3918 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
3919 WordLangTuple & wl, docstring_list & suggestions,
3920 bool do_suggestion, bool check_learned) const
3922 SpellChecker::Result result = SpellChecker::WORD_OK;
3923 SpellChecker * speller = theSpellChecker();
3927 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
3930 locateWord(from, to, WHOLE_WORD);
3931 if (from == to || from >= size())
3934 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3935 Language * lang = d->getSpellLanguage(from);
3937 wl = WordLangTuple(word, lang);
3942 if (needsSpellCheck() || check_learned) {
3944 if (!d->ignoreWord(word)) {
3945 bool const trailing_dot = to < size() && d->text_[to] == '.';
3946 result = speller->check(wl);
3947 if (SpellChecker::misspelled(result) && trailing_dot) {
3948 wl = WordLangTuple(word.append(from_ascii(".")), lang);
3949 result = speller->check(wl);
3950 if (!SpellChecker::misspelled(result)) {
3951 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
3953 from << ".." << to << "]");
3955 // spell check with dot appended failed too
3956 // restore original word/lang value
3957 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3958 wl = WordLangTuple(word, lang);
3962 if (!SpellChecker::misspelled(result)) {
3963 // area up to the begin of the next word is not misspelled
3964 while (end < size() && isWordSeparator(end))
3967 d->setMisspelled(from, end, result);
3969 result = d->speller_state_.getState(from);
3973 suggestions.clear();
3975 if (SpellChecker::misspelled(result)) {
3976 LYXERR(Debug::GUI, "misspelled word: \"" <<
3978 from << ".." << to << "]");
3980 speller->suggest(wl, suggestions);
3986 void Paragraph::Private::markMisspelledWords(
3987 pos_type const & first, pos_type const & last,
3988 SpellChecker::Result result,
3989 docstring const & word,
3990 SkipPositions const & skips)
3992 if (!SpellChecker::misspelled(result)) {
3993 setMisspelled(first, last, SpellChecker::WORD_OK);
3997 SpellChecker * speller = theSpellChecker();
3998 // locate and enumerate the error positions
3999 int nerrors = speller->numMisspelledWords();
4001 SkipPositionsIterator it = skips.begin();
4002 SkipPositionsIterator et = skips.end();
4003 for (int index = 0; index < nerrors; ++index) {
4006 speller->misspelledWord(index, wstart, wlen);
4007 /// should not happen if speller supports range checks
4008 if (!wlen) continue;
4009 docstring const misspelled = word.substr(wstart, wlen);
4010 wstart += first + numskipped;
4011 if (snext < wstart) {
4012 /// mark the range of correct spelling
4013 numskipped += countSkips(it, et, wstart);
4014 setMisspelled(snext,
4015 wstart - 1, SpellChecker::WORD_OK);
4017 snext = wstart + wlen;
4018 numskipped += countSkips(it, et, snext);
4019 /// mark the range of misspelling
4020 setMisspelled(wstart, snext, result);
4021 LYXERR(Debug::GUI, "misspelled word: \"" <<
4022 misspelled << "\" [" <<
4023 wstart << ".." << (snext-1) << "]");
4026 if (snext <= last) {
4027 /// mark the range of correct spelling at end
4028 setMisspelled(snext, last, SpellChecker::WORD_OK);
4033 void Paragraph::spellCheck() const
4035 SpellChecker * speller = theSpellChecker();
4036 if (!speller || empty() ||!needsSpellCheck())
4040 d->rangeOfSpellCheck(start, endpos);
4041 if (speller->canCheckParagraph()) {
4042 // loop until we leave the range
4043 for (pos_type first = start; first < endpos; ) {
4044 pos_type last = endpos;
4045 Private::SkipPositions skips;
4046 Language * lang = d->locateSpellRange(first, last, skips);
4047 if (first >= endpos)
4049 // start the spell checker on the unit of meaning
4050 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4051 WordLangTuple wl = WordLangTuple(word, lang);
4052 SpellChecker::Result result = word.size() ?
4053 speller->check(wl) : SpellChecker::WORD_OK;
4054 d->markMisspelledWords(first, last, result, word, skips);
4058 static docstring_list suggestions;
4059 pos_type to = endpos;
4060 while (start < endpos) {
4062 spellCheck(start, to, wl, suggestions, false);
4066 d->readySpellCheck();
4070 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4072 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4073 if (result || pos <= 0 || pos > size())
4075 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4076 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4081 string Paragraph::magicLabel() const
4084 ss << "magicparlabel-" << id();