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"
29 #include "InsetList.h"
31 #include "LaTeXFeatures.h"
37 #include "OutputParams.h"
38 #include "output_latex.h"
39 #include "output_xhtml.h"
40 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
43 #include "TextClass.h"
47 #include "WordLangTuple.h"
50 #include "frontends/alert.h"
52 #include "insets/InsetBibitem.h"
53 #include "insets/InsetLabel.h"
54 #include "insets/InsetSpecialChar.h"
56 #include "support/debug.h"
57 #include "support/docstring_list.h"
58 #include "support/ExceptionMessage.h"
59 #include "support/gettext.h"
60 #include "support/lassert.h"
61 #include "support/lstrings.h"
62 #include "support/textutils.h"
68 using namespace lyx::support;
73 /// Inset identifier (above 0x10ffff, for ucs-4)
74 char_type const META_INSET = 0x200001;
78 /////////////////////////////////////////////////////////////////////
82 /////////////////////////////////////////////////////////////////////
84 class SpellResultRange {
86 SpellResultRange(FontSpan range, SpellChecker::Result result)
87 : range_(range), result_(result)
90 FontSpan const & range() const { return range_; }
92 void range(FontSpan const & r) { range_ = r; }
94 SpellChecker::Result result() const { return result_; }
96 void result(SpellChecker::Result r) { result_ = r; }
98 bool inside(pos_type pos) const { return range_.inside(pos); }
100 bool covered(FontSpan const & r) const
102 // 1. first of new range inside current range or
103 // 2. last of new range inside current range or
104 // 3. first of current range inside new range or
105 // 4. last of current range inside new range
106 return range_.inside(r.first) || range_.inside(r.last) ||
107 r.inside(range_.first) || r.inside(range_.last);
110 void shift(pos_type pos, int offset)
112 if (range_.first > pos) {
113 range_.first += offset;
114 range_.last += offset;
115 } else if (range_.last > pos) {
116 range_.last += offset;
121 SpellChecker::Result result_ ;
125 /////////////////////////////////////////////////////////////////////
129 /////////////////////////////////////////////////////////////////////
131 class SpellCheckerState {
133 SpellCheckerState() {
134 needs_refresh_ = true;
135 current_change_number_ = 0;
138 void setRange(FontSpan const fp, SpellChecker::Result state)
140 eraseCoveredRanges(fp);
141 if (state != SpellChecker::WORD_OK)
142 ranges_.push_back(SpellResultRange(fp, state));
145 void increasePosAfterPos(pos_type pos)
147 correctRangesAfterPos(pos, 1);
151 void decreasePosAfterPos(pos_type pos)
153 correctRangesAfterPos(pos, -1);
157 void refreshLast(pos_type pos)
159 if (pos < refresh_.last)
163 SpellChecker::Result getState(pos_type pos) const
165 SpellChecker::Result result = SpellChecker::WORD_OK;
166 RangesIterator et = ranges_.end();
167 RangesIterator it = ranges_.begin();
168 for (; it != et; ++it) {
169 if(it->inside(pos)) {
176 bool needsRefresh() const {
177 return needs_refresh_;
180 SpellChecker::ChangeNumber currentChangeNumber() const {
181 return current_change_number_;
184 void refreshRange(pos_type & first, pos_type & last) const {
185 first = refresh_.first;
186 last = refresh_.last;
189 void needsRefresh(pos_type pos) {
190 if (needs_refresh_ && pos != -1) {
191 if (pos < refresh_.first)
192 refresh_.first = pos;
193 if (pos > refresh_.last)
195 } else if (pos != -1) {
196 refresh_.first = pos;
199 needs_refresh_ = pos != -1;
202 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number) {
203 needs_refresh_ = true;
206 current_change_number_ = change_number;
210 typedef vector<SpellResultRange> Ranges;
211 typedef Ranges::const_iterator RangesIterator;
213 /// the area of the paragraph with pending spell check
216 /// spell state cache version number
217 SpellChecker::ChangeNumber current_change_number_;
220 void eraseCoveredRanges(FontSpan const fp)
223 RangesIterator et = ranges_.end();
224 RangesIterator it = ranges_.begin();
225 for (; it != et; ++it) {
226 if (!it->covered(fp))
227 result.push_back(SpellResultRange(it->range(), it->result()));
232 void correctRangesAfterPos(pos_type pos, int offset)
234 RangesIterator et = ranges_.end();
235 Ranges::iterator it = ranges_.begin();
236 for (; it != et; ++it) {
237 it->shift(pos, offset);
243 /////////////////////////////////////////////////////////////////////
245 // Paragraph::Private
247 /////////////////////////////////////////////////////////////////////
249 class Paragraph::Private
253 Private(Paragraph * owner, Layout const & layout);
254 /// "Copy constructor"
255 Private(Private const &, Paragraph * owner);
256 /// Copy constructor from \p beg to \p end
257 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
260 void insertChar(pos_type pos, char_type c, Change const & change);
262 /// Output the surrogate pair formed by \p c and \p next to \p os.
263 /// \return the number of characters written.
264 int latexSurrogatePair(odocstream & os, char_type c, char_type next,
265 OutputParams const &);
267 /// Output a space in appropriate formatting (or a surrogate pair
268 /// if the next character is a combining character).
269 /// \return whether a surrogate pair was output.
270 bool simpleTeXBlanks(OutputParams const &,
271 odocstream &, TexRow & texrow,
273 unsigned int & column,
275 Layout const & style);
277 /// Output consecutive unicode chars, belonging to the same script as
278 /// specified by the latex macro \p ltx, to \p os starting from \p i.
279 /// \return the number of characters written.
280 int writeScriptChars(odocstream & os, docstring const & ltx,
281 Change const &, Encoding const &, pos_type & i);
283 /// This could go to ParagraphParameters if we want to.
284 int startTeXParParams(BufferParams const &, odocstream &, TexRow &,
285 OutputParams const &) const;
287 /// This could go to ParagraphParameters if we want to.
288 int endTeXParParams(BufferParams const &, odocstream &, TexRow &,
289 OutputParams const &) const;
292 void latexInset(BufferParams const &,
294 TexRow & texrow, OutputParams &,
297 Font const & outerfont,
299 Change & running_change,
300 Layout const & style,
302 unsigned int & column);
305 void latexSpecialChar(
307 OutputParams const & runparams,
308 Font const & running_font,
309 Change const & running_change,
310 Layout const & style,
312 unsigned int & column);
319 unsigned int & column);
321 bool latexSpecialTypewriter(
325 unsigned int & column);
327 bool latexSpecialPhrase(
330 unsigned int & column,
331 OutputParams const & runparams);
334 void validate(LaTeXFeatures & features) const;
336 /// Checks if the paragraph contains only text and no inset or font change.
337 bool onlyText(Buffer const & buf, Font const & outerfont,
338 pos_type initial) const;
340 /// match a string against a particular point in the paragraph
341 bool isTextAt(string const & str, pos_type pos) const;
343 /// a vector of speller skip positions
344 typedef vector<FontSpan> SkipPositions;
345 typedef SkipPositions::const_iterator SkipPositionsIterator;
347 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
349 Language * getSpellLanguage(pos_type const from) const;
351 Language * locateSpellRange(pos_type & from, pos_type & to,
352 SkipPositions & skips) const;
354 bool hasSpellerChange() const {
355 SpellChecker::ChangeNumber speller_change_number = 0;
356 if (theSpellChecker())
357 speller_change_number = theSpellChecker()->changeNumber();
358 return speller_change_number > speller_state_.currentChangeNumber();
361 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
363 pos_type textsize = owner_->size();
364 // check for sane arguments
365 if (to < from || from >= textsize)
367 FontSpan fp = FontSpan(from, to);
368 // don't mark end of paragraph
369 if (fp.last >= textsize)
370 fp.last = textsize - 1;
371 speller_state_.setRange(fp, state);
374 void requestSpellCheck(pos_type pos) {
375 speller_state_.needsRefresh(pos);
378 void readySpellCheck() {
379 speller_state_.needsRefresh(-1);
382 bool needsSpellCheck() const
384 return speller_state_.needsRefresh();
387 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
389 speller_state_.refreshRange(first, last);
391 last = owner_->size();
394 pos_type endpos = last;
395 owner_->locateWord(first, endpos, WHOLE_WORD);
398 owner_->locateWord(last, endpos, WHOLE_WORD);
403 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
407 while (it != et && it->first < start) {
408 int skip = it->last - it->first + 1;
416 void markMisspelledWords(pos_type const & first, pos_type const & last,
417 SpellChecker::Result result,
418 docstring const & word,
419 SkipPositions const & skips);
421 InsetCode ownerCode() const
423 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
426 /// Which Paragraph owns us?
430 Inset const * inset_owner_;
439 ParagraphParameters params_;
441 /// for recording and looking up changes
445 InsetList insetlist_;
448 pos_type begin_of_body_;
450 typedef docstring TextContainer;
454 typedef set<docstring> Words;
455 typedef map<Language, Words> LangWordsMap;
459 Layout const * layout_;
461 SpellCheckerState speller_state_;
467 struct special_phrase {
473 special_phrase const special_phrases[] = {
474 { "LyX", from_ascii("\\LyX{}"), false },
475 { "TeX", from_ascii("\\TeX{}"), true },
476 { "LaTeX2e", from_ascii("\\LaTeXe{}"), true },
477 { "LaTeX", from_ascii("\\LaTeX{}"), true },
480 size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase);
485 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
486 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
492 // Initialization of the counter for the paragraph id's,
494 // FIXME: There should be a more intelligent way to generate and use the
495 // paragraph ids per buffer instead a global static counter for all InsetText
496 // in the running program.
497 static int paragraph_id = -1;
499 Paragraph::Private::Private(Private const & p, Paragraph * owner)
500 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
501 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
502 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
505 id_ = ++paragraph_id;
506 requestSpellCheck(p.text_.size());
510 Paragraph::Private::Private(Private const & p, Paragraph * owner,
511 pos_type beg, pos_type end)
512 : owner_(owner), inset_owner_(p.inset_owner_),
513 params_(p.params_), changes_(p.changes_),
514 insetlist_(p.insetlist_, beg, end),
515 begin_of_body_(p.begin_of_body_), words_(p.words_),
518 id_ = ++paragraph_id;
519 if (beg >= pos_type(p.text_.size()))
521 text_ = p.text_.substr(beg, end - beg);
523 FontList::const_iterator fcit = fontlist_.begin();
524 FontList::const_iterator fend = fontlist_.end();
525 for (; fcit != fend; ++fcit) {
526 if (fcit->pos() < beg)
528 if (fcit->pos() >= end) {
529 // Add last entry in the fontlist_.
530 fontlist_.set(text_.size() - 1, fcit->font());
533 // Add a new entry in the fontlist_.
534 fontlist_.set(fcit->pos() - beg, fcit->font());
536 requestSpellCheck(p.text_.size());
540 void Paragraph::addChangesToToc(DocIterator const & cdit,
541 Buffer const & buf) const
543 d->changes_.addToToc(cdit, buf);
547 bool Paragraph::isDeleted(pos_type start, pos_type end) const
549 LASSERT(start >= 0 && start <= size(), /**/);
550 LASSERT(end > start && end <= size() + 1, /**/);
552 return d->changes_.isDeleted(start, end);
556 bool Paragraph::isChanged(pos_type start, pos_type end) const
558 LASSERT(start >= 0 && start <= size(), /**/);
559 LASSERT(end > start && end <= size() + 1, /**/);
561 return d->changes_.isChanged(start, end);
565 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
567 // keep the logic here in sync with the logic of eraseChars()
571 Change const change = d->changes_.lookup(size());
572 return change.inserted() && change.currentAuthor();
576 void Paragraph::setChange(Change const & change)
578 // beware of the imaginary end-of-par character!
579 d->changes_.set(change, 0, size() + 1);
582 * Propagate the change recursively - but not in case of DELETED!
584 * Imagine that your co-author makes changes in an existing inset. He
585 * sends your document to you and you come to the conclusion that the
586 * inset should go completely. If you erase it, LyX must not delete all
587 * text within the inset. Otherwise, the change tracked insertions of
588 * your co-author get lost and there is no way to restore them later.
590 * Conclusion: An inset's content should remain untouched if you delete it
593 if (!change.deleted()) {
594 for (pos_type pos = 0; pos < size(); ++pos) {
595 if (Inset * inset = getInset(pos))
596 inset->setChange(change);
602 void Paragraph::setChange(pos_type pos, Change const & change)
604 LASSERT(pos >= 0 && pos <= size(), /**/);
605 d->changes_.set(change, pos);
607 // see comment in setChange(Change const &) above
608 if (!change.deleted() && pos < size())
609 if (Inset * inset = getInset(pos))
610 inset->setChange(change);
614 Change const & Paragraph::lookupChange(pos_type pos) const
616 LASSERT(pos >= 0 && pos <= size(), /**/);
617 return d->changes_.lookup(pos);
621 void Paragraph::acceptChanges(pos_type start, pos_type end)
623 LASSERT(start >= 0 && start <= size(), /**/);
624 LASSERT(end > start && end <= size() + 1, /**/);
626 for (pos_type pos = start; pos < end; ++pos) {
627 switch (lookupChange(pos).type) {
628 case Change::UNCHANGED:
629 // accept changes in nested inset
630 if (Inset * inset = getInset(pos))
631 inset->acceptChanges();
634 case Change::INSERTED:
635 d->changes_.set(Change(Change::UNCHANGED), pos);
636 // also accept changes in nested inset
637 if (Inset * inset = getInset(pos))
638 inset->acceptChanges();
641 case Change::DELETED:
642 // Suppress access to non-existent
643 // "end-of-paragraph char"
645 eraseChar(pos, false);
656 void Paragraph::rejectChanges(pos_type start, pos_type end)
658 LASSERT(start >= 0 && start <= size(), /**/);
659 LASSERT(end > start && end <= size() + 1, /**/);
661 for (pos_type pos = start; pos < end; ++pos) {
662 switch (lookupChange(pos).type) {
663 case Change::UNCHANGED:
664 // reject changes in nested inset
665 if (Inset * inset = getInset(pos))
666 inset->rejectChanges();
669 case Change::INSERTED:
670 // Suppress access to non-existent
671 // "end-of-paragraph char"
673 eraseChar(pos, false);
679 case Change::DELETED:
680 d->changes_.set(Change(Change::UNCHANGED), pos);
682 // Do NOT reject changes within a deleted inset!
683 // There may be insertions of a co-author inside of it!
691 void Paragraph::Private::insertChar(pos_type pos, char_type c,
692 Change const & change)
694 LASSERT(pos >= 0 && pos <= int(text_.size()), /**/);
697 changes_.insert(change, pos);
699 // This is actually very common when parsing buffers (and
700 // maybe inserting ascii text)
701 if (pos == pos_type(text_.size())) {
702 // when appending characters, no need to update tables
704 // but we want spell checking
705 requestSpellCheck(pos);
709 text_.insert(text_.begin() + pos, c);
711 // Update the font table.
712 fontlist_.increasePosAfterPos(pos);
715 insetlist_.increasePosAfterPos(pos);
717 // Update list of misspelled positions
718 speller_state_.increasePosAfterPos(pos);
722 bool Paragraph::insertInset(pos_type pos, Inset * inset,
723 Change const & change)
725 LASSERT(inset, /**/);
726 LASSERT(pos >= 0 && pos <= size(), /**/);
728 // Paragraph::insertInset() can be used in cut/copy/paste operation where
729 // d->inset_owner_ is not set yet.
730 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
733 d->insertChar(pos, META_INSET, change);
734 LASSERT(d->text_[pos] == META_INSET, /**/);
736 // Add a new entry in the insetlist_.
737 d->insetlist_.insert(inset, pos);
739 // Some insets require run of spell checker
740 requestSpellCheck(pos);
745 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
747 LASSERT(pos >= 0 && pos <= size(), return false);
749 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
752 Change change = d->changes_.lookup(pos);
754 // set the character to DELETED if
755 // a) it was previously unchanged or
756 // b) it was inserted by a co-author
758 if (!change.changed() ||
759 (change.inserted() && !change.currentAuthor())) {
760 setChange(pos, Change(Change::DELETED));
761 // request run of spell checker
762 requestSpellCheck(pos);
766 if (change.deleted())
770 // Don't physically access the imaginary end-of-paragraph character.
771 // eraseChar() can only mark it as DELETED. A physical deletion of
772 // end-of-par must be handled externally.
778 d->changes_.erase(pos);
780 // if it is an inset, delete the inset entry
781 if (d->text_[pos] == META_INSET)
782 d->insetlist_.erase(pos);
784 d->text_.erase(d->text_.begin() + pos);
786 // Update the fontlist_
787 d->fontlist_.erase(pos);
789 // Update the insetlist_
790 d->insetlist_.decreasePosAfterPos(pos);
792 // Update list of misspelled positions
793 d->speller_state_.decreasePosAfterPos(pos);
794 d->speller_state_.refreshLast(size());
800 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
802 LASSERT(start >= 0 && start <= size(), /**/);
803 LASSERT(end >= start && end <= size() + 1, /**/);
806 for (pos_type count = end - start; count; --count) {
807 if (!eraseChar(i, trackChanges))
814 int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
815 char_type next, OutputParams const & runparams)
817 // Writing next here may circumvent a possible font change between
818 // c and next. Since next is only output if it forms a surrogate pair
819 // with c we can ignore this:
820 // A font change inside a surrogate pair does not make sense and is
821 // hopefully impossible to input.
822 // FIXME: change tracking
823 // Is this correct WRT change tracking?
824 Encoding const & encoding = *(runparams.encoding);
825 docstring const latex1 = encoding.latexChar(next);
826 docstring const latex2 = encoding.latexChar(c);
827 if (docstring(1, next) == latex1) {
828 // the encoding supports the combination
829 os << latex2 << latex1;
830 return latex1.length() + latex2.length();
831 } else if (runparams.local_font &&
832 runparams.local_font->language()->lang() == "polutonikogreek") {
833 // polutonikogreek only works without the brackets
834 os << latex1 << latex2;
835 return latex1.length() + latex2.length();
837 os << latex1 << '{' << latex2 << '}';
838 return latex1.length() + latex2.length() + 2;
842 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
843 odocstream & os, TexRow & texrow,
845 unsigned int & column,
847 Layout const & style)
849 if (style.pass_thru || runparams.pass_thru)
852 if (i + 1 < int(text_.size())) {
853 char_type next = text_[i + 1];
854 if (Encodings::isCombiningChar(next)) {
855 // This space has an accent, so we must always output it.
856 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
861 if (runparams.linelen > 0
862 && column > runparams.linelen
864 && text_[i - 1] != ' '
865 && (i + 1 < int(text_.size()))
866 // same in FreeSpacing mode
867 && !owner_->isFreeSpacing()
868 // In typewriter mode, we want to avoid
869 // ! . ? : at the end of a line
870 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
871 && (text_[i - 1] == '.'
872 || text_[i - 1] == '?'
873 || text_[i - 1] == ':'
874 || text_[i - 1] == '!'))) {
877 texrow.start(owner_->id(), i + 1);
879 } else if (style.free_spacing) {
888 int Paragraph::Private::writeScriptChars(odocstream & os,
889 docstring const & ltx,
890 Change const & runningChange,
891 Encoding const & encoding,
894 // FIXME: modifying i here is not very nice...
896 // We only arrive here when a proper language for character text_[i] has
897 // not been specified (i.e., it could not be translated in the current
898 // latex encoding) or its latex translation has been forced, and it
899 // belongs to a known script.
900 // Parameter ltx contains the latex translation of text_[i] as specified
901 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
902 // The latex macro name "textXXX" specifies the script to which text_[i]
903 // belongs and we use it in order to check whether characters from the
904 // same script immediately follow, such that we can collect them in a
905 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
906 // for the first char but only "<spec>" for all subsequent chars.
907 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
908 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
909 string script = to_ascii(ltx.substr(1, brace1 - 1));
912 bool closing_brace = true;
913 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
914 // Correct encoding is being used, so we can avoid \textgreek.
917 closing_brace = false;
919 os << ltx.substr(pos, length);
920 int size = text_.size();
921 while (i + 1 < size) {
922 char_type const next = text_[i + 1];
923 // Stop here if next character belongs to another script
924 // or there is a change in change tracking status.
925 if (!Encodings::isKnownScriptChar(next, script) ||
926 runningChange != owner_->lookupChange(i + 1))
930 FontList::const_iterator cit = fontlist_.begin();
931 FontList::const_iterator end = fontlist_.end();
932 for (; cit != end; ++cit) {
933 if (cit->pos() >= i && !found) {
934 prev_font = cit->font();
937 if (cit->pos() >= i + 1)
940 // Stop here if there is a font attribute or encoding change.
941 if (found && cit != end && prev_font != cit->font())
943 docstring const latex = encoding.latexChar(next);
944 docstring::size_type const b1 =
945 latex.find_first_of(from_ascii("{"));
946 docstring::size_type const b2 =
947 latex.find_last_of(from_ascii("}"));
948 int const len = b2 - b1 - 1;
949 os << latex.substr(b1 + 1, len);
961 bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
963 pos_type const len = str.length();
965 // is the paragraph large enough?
966 if (pos + len > int(text_.size()))
969 // does the wanted text start at point?
970 for (string::size_type i = 0; i < str.length(); ++i) {
971 // Caution: direct comparison of characters works only
972 // because str is pure ASCII.
973 if (str[i] != text_[pos + i])
977 return fontlist_.hasChangeInRange(pos, len);
981 void Paragraph::Private::latexInset(BufferParams const & bparams,
984 OutputParams & runparams,
987 Font const & outerfont,
989 Change & running_change,
990 Layout const & style,
992 unsigned int & column)
994 Inset * inset = owner_->getInset(i);
995 LASSERT(inset, /**/);
997 if (style.pass_thru) {
998 inset->plaintext(os, runparams);
1002 // FIXME: move this to InsetNewline::latex
1003 if (inset->lyxCode() == NEWLINE_CODE) {
1004 // newlines are handled differently here than
1005 // the default in simpleTeXSpecialChars().
1006 if (!style.newline_allowed) {
1010 column += running_font.latexWriteEndChanges(
1011 os, bparams, runparams,
1012 basefont, basefont);
1016 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1019 basefont = owner_->getLayoutFont(bparams, outerfont);
1020 running_font = basefont;
1022 if (runparams.moving_arg)
1027 texrow.start(owner_->id(), i + 1);
1031 if (owner_->isDeleted(i)) {
1032 if( ++runparams.inDeletedInset == 1)
1033 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1036 if (inset->canTrackChanges()) {
1037 column += Changes::latexMarkChange(os, bparams, running_change,
1038 Change(Change::UNCHANGED), runparams);
1039 running_change = Change(Change::UNCHANGED);
1043 odocstream::pos_type const len = os.tellp();
1045 if (inset->forceLTR()
1046 && running_font.isRightToLeft()
1047 // ERT is an exception, it should be output with no
1048 // decorations at all
1049 && inset->lyxCode() != ERT_CODE) {
1050 if (running_font.language()->lang() == "farsi")
1057 // FIXME: Bug: we can have an empty font change here!
1058 // if there has just been a font change, we are going to close it
1059 // right now, which means stupid latex code like \textsf{}. AFAIK,
1060 // this does not harm dvi output. A minor bug, thus (JMarc)
1062 // Some insets cannot be inside a font change command.
1063 // However, even such insets *can* be placed in \L or \R
1064 // or their equivalents (for RTL language switches), so we don't
1065 // close the language in those cases.
1066 // ArabTeX, though, cannot handle this special behavior, it seems.
1067 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1068 || running_font.language()->lang() == "arabic_arabtex";
1069 if (open_font && inset->noFontChange()) {
1070 bool closeLanguage = arabtex
1071 || basefont.isRightToLeft() == running_font.isRightToLeft();
1072 unsigned int count = running_font.latexWriteEndChanges(os,
1073 bparams, runparams, basefont, basefont, closeLanguage);
1075 // if any font properties were closed, update the running_font,
1076 // making sure, however, to leave the language as it was
1078 // FIXME: probably a better way to keep track of the old
1079 // language, than copying the entire font?
1080 Font const copy_font(running_font);
1081 basefont = owner_->getLayoutFont(bparams, outerfont);
1082 running_font = basefont;
1084 running_font.setLanguage(copy_font.language());
1085 // leave font open if language is still open
1086 open_font = (running_font.language() == basefont.language());
1088 runparams.local_font = &basefont;
1095 tmp = inset->latex(os, runparams);
1096 } catch (EncodingException & e) {
1097 // add location information and throw again.
1104 if (running_font.language()->lang() == "farsi")
1111 texrow.newlines(tmp);
1112 texrow.start(owner_->id(), i + 1);
1115 column += (unsigned int)(os.tellp() - len);
1118 if (owner_->isDeleted(i))
1119 --runparams.inDeletedInset;
1123 void Paragraph::Private::latexSpecialChar(
1125 OutputParams const & runparams,
1126 Font const & running_font,
1127 Change const & running_change,
1128 Layout const & style,
1130 unsigned int & column)
1132 char_type const c = text_[i];
1134 if (style.pass_thru || runparams.pass_thru) {
1136 // FIXME UNICODE: This can fail if c cannot
1137 // be encoded in the current encoding.
1142 // If T1 font encoding is used, use the special
1143 // characters it provides.
1144 // NOTE: some languages reset the font encoding
1146 if (!running_font.language()->internalFontEncoding()
1147 && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
1150 // \tt font needs special treatment
1151 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
1152 && latexSpecialTypewriter(c, os, i, column))
1155 // Otherwise, we use what LaTeX provides us.
1158 os << "\\textbackslash{}";
1162 os << "\\textless{}";
1166 os << "\\textgreater{}";
1170 os << "\\textbar{}";
1177 os << "\\char`\\\"{}";
1182 case '%': case '#': case '{':
1190 os << "\\textasciitilde{}";
1195 os << "\\textasciicircum{}";
1202 // avoid being mistaken for optional arguments
1210 // Blanks are printed before font switching.
1211 // Sure? I am not! (try nice-latex)
1212 // I am sure it's correct. LyX might be smarter
1213 // in the future, but for now, nothing wrong is
1219 if (latexSpecialPhrase(os, i, column, runparams))
1225 Encoding const & encoding = *(runparams.encoding);
1226 if (i + 1 < int(text_.size())) {
1227 char_type next = text_[i + 1];
1228 if (Encodings::isCombiningChar(next)) {
1229 column += latexSurrogatePair(os, c, next, runparams) - 1;
1235 docstring const latex = encoding.latexChar(c);
1236 if (Encodings::isKnownScriptChar(c, script)
1237 && prefixIs(latex, from_ascii("\\" + script)))
1238 column += writeScriptChars(os, latex,
1239 running_change, encoding, i) - 1;
1240 else if (latex.length() > 1 && latex[latex.length() - 1] != '}') {
1241 // Prevent eating of a following
1242 // space or command corruption by
1243 // following characters
1244 column += latex.length() + 1;
1245 os << latex << "{}";
1247 column += latex.length() - 1;
1255 bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
1256 pos_type i, unsigned int & column)
1262 // In T1 encoding, these characters exist
1263 // but we should avoid ligatures
1264 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1266 os << "\\textcompwordmark{}";
1273 // soul.sty breaks with \char`\"
1274 os << "\\textquotedbl{}";
1283 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream & os,
1284 pos_type i, unsigned int & column)
1288 // within \ttfamily, "--" is merged to "-" (no endash)
1289 // so we avoid this rather irritating ligature
1290 if (i + 1 < int(text_.size()) && text_[i + 1] == '-') {
1297 // everything else has to be checked separately
1298 // (depending on the encoding)
1305 bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
1306 unsigned int & column, OutputParams const & runparams)
1308 // FIXME: if we have "LaTeX" with a font
1309 // change in the middle (before the 'T', then
1310 // the "TeX" part is still special cased.
1311 // Really we should only operate this on
1312 // "words" for some definition of word
1314 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1315 if (!isTextAt(special_phrases[pnr].phrase, i))
1317 if (runparams.moving_arg)
1319 os << special_phrases[pnr].macro;
1320 i += special_phrases[pnr].phrase.length() - 1;
1321 column += special_phrases[pnr].macro.length() - 1;
1328 void Paragraph::Private::validate(LaTeXFeatures & features) const
1330 if (layout_->inpreamble && inset_owner_) {
1331 Buffer const & buf = inset_owner_->buffer();
1332 BufferParams const & bp = buf.params();
1335 odocstringstream ods;
1336 // we have to provide all the optional arguments here, even though
1337 // the last one is the only one we care about.
1338 owner_->latex(bp, f, ods, tr, features.runparams(), 0, -1, true);
1339 docstring const d = ods.str();
1341 // this will have "{" at the beginning, but not at the end
1342 string const content = to_utf8(d);
1343 string const cmd = layout_->latexname();
1344 features.addPreambleSnippet("\\" + cmd + content + "}");
1348 if (features.runparams().flavor == OutputParams::HTML
1349 && layout_->htmltitle()) {
1350 features.setHTMLTitle(owner_->asString(AS_STR_INSETS));
1353 // check the params.
1354 if (!params_.spacing().isDefault())
1355 features.require("setspace");
1358 features.useLayout(layout_->name());
1361 fontlist_.validate(features);
1363 // then the indentation
1364 if (!params_.leftIndent().zero())
1365 features.require("ParagraphLeftIndent");
1368 InsetList::const_iterator icit = insetlist_.begin();
1369 InsetList::const_iterator iend = insetlist_.end();
1370 for (; icit != iend; ++icit) {
1372 icit->inset->validate(features);
1373 if (layout_->needprotect &&
1374 icit->inset->lyxCode() == FOOT_CODE)
1375 features.require("NeedLyXFootnoteCode");
1379 // then the contents
1380 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1381 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1382 if (!special_phrases[pnr].builtin
1383 && isTextAt(special_phrases[pnr].phrase, i)) {
1384 features.require(special_phrases[pnr].phrase);
1388 Encodings::validate(text_[i], features);
1392 /////////////////////////////////////////////////////////////////////
1396 /////////////////////////////////////////////////////////////////////
1399 Layout const emptyParagraphLayout;
1402 Paragraph::Paragraph()
1403 : d(new Paragraph::Private(this, emptyParagraphLayout))
1410 Paragraph::Paragraph(Paragraph const & par)
1411 : itemdepth(par.itemdepth),
1412 d(new Paragraph::Private(*par.d, this))
1418 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1419 : itemdepth(par.itemdepth),
1420 d(new Paragraph::Private(*par.d, this, beg, end))
1426 Paragraph & Paragraph::operator=(Paragraph const & par)
1428 // needed as we will destroy the private part before copying it
1430 itemdepth = par.itemdepth;
1434 d = new Private(*par.d, this);
1441 Paragraph::~Paragraph()
1450 // this shall be called just before every "os << ..." action.
1451 void flushString(ostream & os, docstring & s)
1460 void Paragraph::write(ostream & os, BufferParams const & bparams,
1461 depth_type & dth) const
1463 // The beginning or end of a deeper (i.e. nested) area?
1464 if (dth != d->params_.depth()) {
1465 if (d->params_.depth() > dth) {
1466 while (d->params_.depth() > dth) {
1467 os << "\n\\begin_deeper";
1471 while (d->params_.depth() < dth) {
1472 os << "\n\\end_deeper";
1478 // First write the layout
1479 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1481 d->params_.write(os);
1483 Font font1(inherit_font, bparams.language);
1485 Change running_change = Change(Change::UNCHANGED);
1487 // this string is used as a buffer to avoid repetitive calls
1488 // to to_utf8(), which turn out to be expensive (JMarc)
1489 docstring write_buffer;
1492 for (pos_type i = 0; i <= size(); ++i) {
1494 Change const change = lookupChange(i);
1495 if (change != running_change)
1496 flushString(os, write_buffer);
1497 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1498 running_change = change;
1503 // Write font changes
1504 Font font2 = getFontSettings(bparams, i);
1505 if (font2 != font1) {
1506 flushString(os, write_buffer);
1507 font2.lyxWriteChanges(font1, os);
1512 char_type const c = d->text_[i];
1515 if (Inset const * inset = getInset(i)) {
1516 flushString(os, write_buffer);
1517 if (inset->directWrite()) {
1518 // international char, let it write
1519 // code directly so it's shorter in
1525 os << "\\begin_inset ";
1527 os << "\n\\end_inset\n\n";
1533 flushString(os, write_buffer);
1534 os << "\n\\backslash\n";
1538 flushString(os, write_buffer);
1539 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1546 if ((column > 70 && c == ' ')
1548 flushString(os, write_buffer);
1552 // this check is to amend a bug. LyX sometimes
1553 // inserts '\0' this could cause problems.
1555 write_buffer.push_back(c);
1557 LYXERR0("NUL char in structure.");
1563 flushString(os, write_buffer);
1564 os << "\n\\end_layout\n";
1568 void Paragraph::validate(LaTeXFeatures & features) const
1570 d->validate(features);
1574 void Paragraph::insert(pos_type start, docstring const & str,
1575 Font const & font, Change const & change)
1577 for (size_t i = 0, n = str.size(); i != n ; ++i)
1578 insertChar(start + i, str[i], font, change);
1582 void Paragraph::appendChar(char_type c, Font const & font,
1583 Change const & change)
1586 d->changes_.insert(change, d->text_.size());
1587 // when appending characters, no need to update tables
1588 d->text_.push_back(c);
1589 setFont(d->text_.size() - 1, font);
1593 void Paragraph::appendString(docstring const & s, Font const & font,
1594 Change const & change)
1596 pos_type end = s.size();
1597 size_t oldsize = d->text_.size();
1598 size_t newsize = oldsize + end;
1599 size_t capacity = d->text_.capacity();
1600 if (newsize >= capacity)
1601 d->text_.reserve(max(capacity + 100, newsize));
1603 // when appending characters, no need to update tables
1606 // FIXME: Optimize this!
1607 for (size_t i = oldsize; i != newsize; ++i) {
1609 d->changes_.insert(change, i);
1611 d->fontlist_.set(oldsize, font);
1612 d->fontlist_.set(newsize - 1, font);
1616 void Paragraph::insertChar(pos_type pos, char_type c,
1619 d->insertChar(pos, c, Change(trackChanges ?
1620 Change::INSERTED : Change::UNCHANGED));
1624 void Paragraph::insertChar(pos_type pos, char_type c,
1625 Font const & font, bool trackChanges)
1627 d->insertChar(pos, c, Change(trackChanges ?
1628 Change::INSERTED : Change::UNCHANGED));
1633 void Paragraph::insertChar(pos_type pos, char_type c,
1634 Font const & font, Change const & change)
1636 d->insertChar(pos, c, change);
1641 bool Paragraph::insertInset(pos_type pos, Inset * inset,
1642 Font const & font, Change const & change)
1644 bool const success = insertInset(pos, inset, change);
1645 // Set the font/language of the inset...
1651 void Paragraph::resetFonts(Font const & font)
1653 d->fontlist_.clear();
1654 d->fontlist_.set(0, font);
1655 d->fontlist_.set(d->text_.size() - 1, font);
1658 // Gets uninstantiated font setting at position.
1659 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1663 LYXERR0("pos: " << pos << " size: " << size());
1664 LASSERT(pos <= size(), /**/);
1667 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1668 if (cit != d->fontlist_.end())
1671 if (pos == size() && !empty())
1672 return getFontSettings(bparams, pos - 1);
1674 // Optimisation: avoid a full font instantiation if there is no
1675 // language change from previous call.
1676 static Font previous_font;
1677 static Language const * previous_lang = 0;
1678 Language const * lang = getParLanguage(bparams);
1679 if (lang != previous_lang) {
1680 previous_lang = lang;
1681 previous_font = Font(inherit_font, lang);
1683 return previous_font;
1687 FontSpan Paragraph::fontSpan(pos_type pos) const
1689 LASSERT(pos <= size(), /**/);
1692 FontList::const_iterator cit = d->fontlist_.begin();
1693 FontList::const_iterator end = d->fontlist_.end();
1694 for (; cit != end; ++cit) {
1695 if (cit->pos() >= pos) {
1696 if (pos >= beginOfBody())
1697 return FontSpan(max(start, beginOfBody()),
1700 return FontSpan(start,
1701 min(beginOfBody() - 1,
1704 start = cit->pos() + 1;
1707 // This should not happen, but if so, we take no chances.
1708 // LYXERR0("Paragraph::getEndPosOfFontSpan: This should not happen!");
1709 return FontSpan(pos, pos);
1713 // Gets uninstantiated font setting at position 0
1714 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1716 if (!empty() && !d->fontlist_.empty())
1717 return d->fontlist_.begin()->font();
1719 // Optimisation: avoid a full font instantiation if there is no
1720 // language change from previous call.
1721 static Font previous_font;
1722 static Language const * previous_lang = 0;
1723 if (bparams.language != previous_lang) {
1724 previous_lang = bparams.language;
1725 previous_font = Font(inherit_font, bparams.language);
1728 return previous_font;
1732 // Gets the fully instantiated font at a given position in a paragraph
1733 // This is basically the same function as Text::GetFont() in text2.cpp.
1734 // The difference is that this one is used for generating the LaTeX file,
1735 // and thus cosmetic "improvements" are disallowed: This has to deliver
1736 // the true picture of the buffer. (Asger)
1737 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1738 Font const & outerfont) const
1740 LASSERT(pos >= 0, /**/);
1742 Font font = getFontSettings(bparams, pos);
1744 pos_type const body_pos = beginOfBody();
1745 FontInfo & fi = font.fontInfo();
1747 fi.realize(d->layout_->labelfont);
1749 fi.realize(d->layout_->font);
1751 fi.realize(outerfont.fontInfo());
1752 fi.realize(bparams.getFont().fontInfo());
1758 Font const Paragraph::getLabelFont
1759 (BufferParams const & bparams, Font const & outerfont) const
1761 FontInfo tmpfont = d->layout_->labelfont;
1762 tmpfont.realize(outerfont.fontInfo());
1763 tmpfont.realize(bparams.getFont().fontInfo());
1764 return Font(tmpfont, getParLanguage(bparams));
1768 Font const Paragraph::getLayoutFont
1769 (BufferParams const & bparams, Font const & outerfont) const
1771 FontInfo tmpfont = d->layout_->font;
1772 tmpfont.realize(outerfont.fontInfo());
1773 tmpfont.realize(bparams.getFont().fontInfo());
1774 return Font(tmpfont, getParLanguage(bparams));
1778 /// Returns the height of the highest font in range
1779 FontSize Paragraph::highestFontInRange
1780 (pos_type startpos, pos_type endpos, FontSize def_size) const
1782 return d->fontlist_.highestInRange(startpos, endpos, def_size);
1786 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1788 char_type c = d->text_[pos];
1789 if (!lyxrc.rtl_support)
1819 if (uc != c && getFontSettings(bparams, pos).isRightToLeft())
1825 void Paragraph::setFont(pos_type pos, Font const & font)
1827 LASSERT(pos <= size(), /**/);
1829 // First, reduce font against layout/label font
1830 // Update: The setCharFont() routine in text2.cpp already
1831 // reduces font, so we don't need to do that here. (Asger)
1833 d->fontlist_.set(pos, font);
1837 void Paragraph::makeSameLayout(Paragraph const & par)
1839 d->layout_ = par.d->layout_;
1840 d->params_ = par.d->params_;
1844 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1846 if (isFreeSpacing())
1852 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1853 if (eraseChar(pos, trackChanges))
1859 return count > 0 || pos > 0;
1863 bool Paragraph::hasSameLayout(Paragraph const & par) const
1865 return par.d->layout_ == d->layout_
1866 && d->params_.sameLayout(par.d->params_);
1870 depth_type Paragraph::getDepth() const
1872 return d->params_.depth();
1876 depth_type Paragraph::getMaxDepthAfter() const
1878 if (d->layout_->isEnvironment())
1879 return d->params_.depth() + 1;
1881 return d->params_.depth();
1885 char Paragraph::getAlign() const
1887 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1888 return d->layout_->align;
1890 return d->params_.align();
1894 docstring const & Paragraph::labelString() const
1896 return d->params_.labelString();
1900 // the next two functions are for the manual labels
1901 docstring const Paragraph::getLabelWidthString() const
1903 if (d->layout_->margintype == MARGIN_MANUAL
1904 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
1905 return d->params_.labelWidthString();
1907 return _("Senseless with this layout!");
1911 void Paragraph::setLabelWidthString(docstring const & s)
1913 d->params_.labelWidthString(s);
1917 docstring Paragraph::expandLabel(Layout const & layout,
1918 BufferParams const & bparams) const
1920 return expandParagraphLabel(layout, bparams, true);
1924 docstring Paragraph::expandDocBookLabel(Layout const & layout,
1925 BufferParams const & bparams) const
1927 return expandParagraphLabel(layout, bparams, false);
1931 docstring Paragraph::expandParagraphLabel(Layout const & layout,
1932 BufferParams const & bparams, bool process_appendix) const
1934 DocumentClass const & tclass = bparams.documentClass();
1935 string const & lang = getParLanguage(bparams)->code();
1936 bool const in_appendix = process_appendix && d->params_.appendix();
1937 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
1939 if (fmt.empty() && layout.labeltype == LABEL_COUNTER
1940 && !layout.counter.empty())
1941 return tclass.counters().theCounter(layout.counter, lang);
1943 // handle 'inherited level parts' in 'fmt',
1944 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
1945 size_t const i = fmt.find('@', 0);
1946 if (i != docstring::npos) {
1947 size_t const j = fmt.find('@', i + 1);
1948 if (j != docstring::npos) {
1949 docstring parent(fmt, i + 1, j - i - 1);
1950 docstring label = from_ascii("??");
1951 if (tclass.hasLayout(parent))
1952 docstring label = expandParagraphLabel(tclass[parent], bparams,
1954 fmt = docstring(fmt, 0, i) + label
1955 + docstring(fmt, j + 1, docstring::npos);
1959 return tclass.counters().counterLabel(fmt, lang);
1963 void Paragraph::applyLayout(Layout const & new_layout)
1965 d->layout_ = &new_layout;
1966 LyXAlignment const oldAlign = d->params_.align();
1968 if (!(oldAlign & d->layout_->alignpossible)) {
1969 frontend::Alert::warning(_("Alignment not permitted"),
1970 _("The new layout does not permit the alignment previously used.\nSetting to default."));
1971 d->params_.align(LYX_ALIGN_LAYOUT);
1976 pos_type Paragraph::beginOfBody() const
1978 return d->begin_of_body_;
1982 void Paragraph::setBeginOfBody()
1984 if (d->layout_->labeltype != LABEL_MANUAL) {
1985 d->begin_of_body_ = 0;
1989 // Unroll the first two cycles of the loop
1990 // and remember the previous character to
1991 // remove unnecessary getChar() calls
1993 pos_type end = size();
1994 if (i < end && !isNewline(i)) {
1996 char_type previous_char = 0;
1999 previous_char = d->text_[i];
2000 if (!isNewline(i)) {
2002 while (i < end && previous_char != ' ') {
2007 previous_char = temp;
2013 d->begin_of_body_ = i;
2017 bool Paragraph::allowParagraphCustomization() const
2019 return inInset().allowParagraphCustomization();
2023 bool Paragraph::usePlainLayout() const
2025 return inInset().usePlainLayout();
2031 // paragraphs inside floats need different alignment tags to avoid
2034 bool noTrivlistCentering(InsetCode code)
2036 return code == FLOAT_CODE
2037 || code == WRAP_CODE
2038 || code == CELL_CODE;
2042 string correction(string const & orig)
2044 if (orig == "flushleft")
2045 return "raggedright";
2046 if (orig == "flushright")
2047 return "raggedleft";
2048 if (orig == "center")
2054 string const corrected_env(string const & suffix, string const & env,
2055 InsetCode code, bool const lastpar)
2057 string output = suffix + "{";
2058 if (noTrivlistCentering(code)) {
2060 // the last paragraph in non-trivlist-aligned
2061 // context is special (to avoid unwanted whitespace)
2062 if (suffix == "\\begin")
2063 return "\\" + correction(env) + "{}";
2066 output += correction(env);
2070 if (suffix == "\\begin")
2076 void adjust_row_column(string const & str, TexRow & texrow, int & column)
2078 if (!contains(str, "\n"))
2079 column += str.size();
2083 column = rsplit(str, tmp, '\n').size();
2090 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2091 odocstream & os, TexRow & texrow,
2092 OutputParams const & runparams) const
2096 if (params_.noindent()) {
2097 os << "\\noindent ";
2101 LyXAlignment const curAlign = params_.align();
2103 if (curAlign == layout_->align)
2107 case LYX_ALIGN_NONE:
2108 case LYX_ALIGN_BLOCK:
2109 case LYX_ALIGN_LAYOUT:
2110 case LYX_ALIGN_SPECIAL:
2111 case LYX_ALIGN_DECIMAL:
2113 case LYX_ALIGN_LEFT:
2114 case LYX_ALIGN_RIGHT:
2115 case LYX_ALIGN_CENTER:
2116 if (runparams.moving_arg) {
2123 string const begin_tag = "\\begin";
2124 InsetCode code = ownerCode();
2125 bool const lastpar = runparams.isLastPar;
2128 case LYX_ALIGN_NONE:
2129 case LYX_ALIGN_BLOCK:
2130 case LYX_ALIGN_LAYOUT:
2131 case LYX_ALIGN_SPECIAL:
2132 case LYX_ALIGN_DECIMAL:
2134 case LYX_ALIGN_LEFT: {
2136 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2137 output = corrected_env(begin_tag, "flushleft", code, lastpar);
2139 output = corrected_env(begin_tag, "flushright", code, lastpar);
2140 os << from_ascii(output);
2141 adjust_row_column(output, texrow, column);
2143 } case LYX_ALIGN_RIGHT: {
2145 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2146 output = corrected_env(begin_tag, "flushright", code, lastpar);
2148 output = corrected_env(begin_tag, "flushleft", code, lastpar);
2149 os << from_ascii(output);
2150 adjust_row_column(output, texrow, column);
2152 } case LYX_ALIGN_CENTER: {
2154 output = corrected_env(begin_tag, "center", code, lastpar);
2155 os << from_ascii(output);
2156 adjust_row_column(output, texrow, column);
2165 int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2166 odocstream & os, TexRow & texrow,
2167 OutputParams const & runparams) const
2171 LyXAlignment const curAlign = params_.align();
2173 if (curAlign == layout_->align)
2177 case LYX_ALIGN_NONE:
2178 case LYX_ALIGN_BLOCK:
2179 case LYX_ALIGN_LAYOUT:
2180 case LYX_ALIGN_SPECIAL:
2181 case LYX_ALIGN_DECIMAL:
2183 case LYX_ALIGN_LEFT:
2184 case LYX_ALIGN_RIGHT:
2185 case LYX_ALIGN_CENTER:
2186 if (runparams.moving_arg) {
2193 string const end_tag = "\n\\par\\end";
2194 InsetCode code = ownerCode();
2195 bool const lastpar = runparams.isLastPar;
2198 case LYX_ALIGN_NONE:
2199 case LYX_ALIGN_BLOCK:
2200 case LYX_ALIGN_LAYOUT:
2201 case LYX_ALIGN_SPECIAL:
2202 case LYX_ALIGN_DECIMAL:
2204 case LYX_ALIGN_LEFT: {
2206 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2207 output = corrected_env(end_tag, "flushleft", code, lastpar);
2209 output = corrected_env(end_tag, "flushright", code, lastpar);
2210 os << from_ascii(output);
2211 adjust_row_column(output, texrow, column);
2213 } case LYX_ALIGN_RIGHT: {
2215 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2216 output = corrected_env(end_tag, "flushright", code, lastpar);
2218 output = corrected_env(end_tag, "flushleft", code, lastpar);
2219 os << from_ascii(output);
2220 adjust_row_column(output, texrow, column);
2222 } case LYX_ALIGN_CENTER: {
2224 output = corrected_env(end_tag, "center", code, lastpar);
2225 os << from_ascii(output);
2226 adjust_row_column(output, texrow, column);
2235 // This one spits out the text of the paragraph
2236 void Paragraph::latex(BufferParams const & bparams,
2237 Font const & outerfont,
2238 odocstream & os, TexRow & texrow,
2239 OutputParams const & runparams,
2240 int start_pos, int end_pos, bool force) const
2242 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2244 // FIXME This check should not be needed. Perhaps issue an
2245 // error if it triggers.
2246 Layout const & style = inInset().forcePlainLayout() ?
2247 bparams.documentClass().plainLayout() : *d->layout_;
2249 if (!force && style.inpreamble)
2252 bool const allowcust = allowParagraphCustomization();
2254 // Current base font for all inherited font changes, without any
2255 // change caused by an individual character, except for the language:
2256 // It is set to the language of the first character.
2257 // As long as we are in the label, this font is the base font of the
2258 // label. Before the first body character it is set to the base font
2262 // Maybe we have to create a optional argument.
2263 pos_type body_pos = beginOfBody();
2264 unsigned int column = 0;
2267 // the optional argument is kept in curly brackets in
2268 // case it contains a ']'
2271 basefont = getLabelFont(bparams, outerfont);
2273 basefont = getLayoutFont(bparams, outerfont);
2276 // Which font is currently active?
2277 Font running_font(basefont);
2278 // Do we have an open font change?
2279 bool open_font = false;
2281 Change runningChange = Change(Change::UNCHANGED);
2283 Encoding const * const prev_encoding = runparams.encoding;
2285 texrow.start(id(), 0);
2287 // if the paragraph is empty, the loop will not be entered at all
2289 if (style.isCommand()) {
2294 column += d->startTeXParParams(bparams, os, texrow,
2298 for (pos_type i = 0; i < size(); ++i) {
2299 // First char in paragraph or after label?
2300 if (i == body_pos) {
2303 column += running_font.latexWriteEndChanges(
2304 os, bparams, runparams,
2305 basefont, basefont);
2308 basefont = getLayoutFont(bparams, outerfont);
2309 running_font = basefont;
2311 column += Changes::latexMarkChange(os, bparams,
2312 runningChange, Change(Change::UNCHANGED),
2314 runningChange = Change(Change::UNCHANGED);
2319 if (style.isCommand()) {
2325 column += d->startTeXParParams(bparams, os,
2330 Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
2333 if (bparams.outputChanges && runningChange != change) {
2335 column += running_font.latexWriteEndChanges(
2336 os, bparams, runparams, basefont, basefont);
2339 basefont = getLayoutFont(bparams, outerfont);
2340 running_font = basefont;
2342 column += Changes::latexMarkChange(os, bparams, runningChange,
2344 runningChange = change;
2347 // do not output text which is marked deleted
2348 // if change tracking output is disabled
2349 if (!bparams.outputChanges && change.deleted()) {
2355 // Fully instantiated font
2356 Font const font = getFont(bparams, i, outerfont);
2358 Font const last_font = running_font;
2360 // Do we need to close the previous font?
2362 (font != running_font ||
2363 font.language() != running_font.language()))
2365 column += running_font.latexWriteEndChanges(
2366 os, bparams, runparams, basefont,
2367 (i == body_pos-1) ? basefont : font);
2368 running_font = basefont;
2372 string const running_lang = runparams.use_polyglossia ?
2373 running_font.language()->polyglossia() : running_font.language()->babel();
2374 // close babel's font environment before opening CJK.
2375 string const lang_end_command = runparams.use_polyglossia ?
2376 "\\end{$$lang}" : lyxrc.language_command_end;
2377 if (!running_lang.empty() &&
2378 font.language()->encoding()->package() == Encoding::CJK) {
2379 string end_tag = subst(lang_end_command,
2382 os << from_ascii(end_tag);
2383 column += end_tag.length();
2386 // Switch file encoding if necessary (and allowed)
2387 if (!runparams.pass_thru && !style.pass_thru &&
2388 runparams.encoding->package() != Encoding::none &&
2389 font.language()->encoding()->package() != Encoding::none) {
2390 pair<bool, int> const enc_switch = switchEncoding(os, bparams,
2391 runparams, *(font.language()->encoding()));
2392 if (enc_switch.first) {
2393 column += enc_switch.second;
2394 runparams.encoding = font.language()->encoding();
2398 char_type const c = d->text_[i];
2400 // Do we need to change font?
2401 if ((font != running_font ||
2402 font.language() != running_font.language()) &&
2405 odocstringstream ods;
2406 column += font.latexWriteStartChanges(ods, bparams,
2407 runparams, basefont,
2409 running_font = font;
2411 docstring fontchange = ods.str();
2412 // check whether the fontchange ends with a \\textcolor
2413 // modifier and the text starts with a space (bug 4473)
2414 docstring const last_modifier = rsplit(fontchange, '\\');
2415 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2416 os << fontchange << from_ascii("{}");
2417 // check if the fontchange ends with a trailing blank
2418 // (like "\small " (see bug 3382)
2419 else if (suffixIs(fontchange, ' ') && c == ' ')
2420 os << fontchange.substr(0, fontchange.size() - 1)
2421 << from_ascii("{}");
2426 // FIXME: think about end_pos implementation...
2427 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2428 // FIXME: integrate this case in latexSpecialChar
2429 // Do not print the separation of the optional argument
2430 // if style.pass_thru is false. This works because
2431 // latexSpecialChar ignores spaces if
2432 // style.pass_thru is false.
2433 if (i != body_pos - 1) {
2434 if (d->simpleTeXBlanks(
2435 runparams, os, texrow,
2436 i, column, font, style)) {
2437 // A surrogate pair was output. We
2438 // must not call latexSpecialChar
2439 // in this iteration, since it would output
2440 // the combining character again.
2447 OutputParams rp = runparams;
2448 rp.free_spacing = style.free_spacing;
2449 rp.local_font = &font;
2450 rp.intitle = style.intitle;
2452 // Two major modes: LaTeX or plain
2453 // Handle here those cases common to both modes
2454 // and then split to handle the two modes separately.
2455 if (c == META_INSET) {
2456 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2457 d->latexInset(bparams, os,
2458 texrow, rp, running_font,
2459 basefont, outerfont, open_font,
2460 runningChange, style, i, column);
2463 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2465 d->latexSpecialChar(os, rp, running_font, runningChange,
2467 } catch (EncodingException & e) {
2468 if (runparams.dryrun) {
2469 os << "<" << _("LyX Warning: ")
2470 << _("uncodable character") << " '";
2474 // add location information and throw again.
2483 // Set the encoding to that returned from latexSpecialChar (see
2484 // comment for encoding member in OutputParams.h)
2485 runparams.encoding = rp.encoding;
2488 // If we have an open font definition, we have to close it
2490 #ifdef FIXED_LANGUAGE_END_DETECTION
2492 running_font.latexWriteEndChanges(os, bparams,
2493 runparams, basefont,
2494 next_->getFont(bparams, 0, outerfont));
2496 running_font.latexWriteEndChanges(os, bparams,
2497 runparams, basefont, basefont);
2500 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2501 //FIXME: there as we start another \selectlanguage with the next paragraph if
2502 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2503 running_font.latexWriteEndChanges(os, bparams, runparams,
2504 basefont, basefont);
2508 column += Changes::latexMarkChange(os, bparams, runningChange,
2509 Change(Change::UNCHANGED), runparams);
2511 // Needed if there is an optional argument but no contents.
2512 if (body_pos > 0 && body_pos == size()) {
2516 if (allowcust && d->endTeXParParams(bparams, os, texrow, runparams)
2517 && runparams.encoding != prev_encoding) {
2518 runparams.encoding = prev_encoding;
2519 if (!bparams.useXetex)
2520 os << setEncoding(prev_encoding->iconvName());
2523 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2527 bool Paragraph::emptyTag() const
2529 for (pos_type i = 0; i < size(); ++i) {
2530 if (Inset const * inset = getInset(i)) {
2531 InsetCode lyx_code = inset->lyxCode();
2532 // FIXME testing like that is wrong. What is
2534 if (lyx_code != TOC_CODE &&
2535 lyx_code != INCLUDE_CODE &&
2536 lyx_code != GRAPHICS_CODE &&
2537 lyx_code != ERT_CODE &&
2538 lyx_code != LISTINGS_CODE &&
2539 lyx_code != FLOAT_CODE &&
2540 lyx_code != TABULAR_CODE) {
2544 char_type c = d->text_[i];
2545 if (c != ' ' && c != '\t')
2553 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2556 for (pos_type i = 0; i < size(); ++i) {
2557 if (Inset const * inset = getInset(i)) {
2558 InsetCode lyx_code = inset->lyxCode();
2559 if (lyx_code == LABEL_CODE) {
2560 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2561 docstring const & id = il->getParam("name");
2562 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2570 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2574 for (i = 0; i < size(); ++i) {
2575 if (Inset const * inset = getInset(i)) {
2576 inset->docbook(os, runparams);
2578 char_type c = d->text_[i];
2581 os << sgml::escapeChar(c);
2588 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2592 for (i = 0; i < size(); ++i) {
2593 if (Inset const * inset = getInset(i)) {
2594 inset->xhtml(xs, runparams);
2596 char_type c = d->text_[i];
2606 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2609 pos_type size = text_.size();
2610 for (pos_type i = initial; i < size; ++i) {
2611 Font font = owner_->getFont(buf.params(), i, outerfont);
2612 if (text_[i] == META_INSET)
2614 if (i != initial && font != font_old)
2623 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2625 OutputParams const & runparams,
2626 Font const & outerfont,
2627 pos_type initial) const
2629 bool emph_flag = false;
2631 Layout const & style = *d->layout_;
2633 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2635 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2638 // parsing main loop
2639 for (pos_type i = initial; i < size(); ++i) {
2640 Font font = getFont(buf.params(), i, outerfont);
2642 // handle <emphasis> tag
2643 if (font_old.emph() != font.fontInfo().emph()) {
2644 if (font.fontInfo().emph() == FONT_ON) {
2647 } else if (i != initial) {
2648 os << "</emphasis>";
2653 if (Inset const * inset = getInset(i)) {
2654 inset->docbook(os, runparams);
2656 char_type c = d->text_[i];
2658 if (style.pass_thru)
2661 os << sgml::escapeChar(c);
2663 font_old = font.fontInfo();
2667 os << "</emphasis>";
2670 if (style.free_spacing)
2672 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2677 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2679 OutputParams const & runparams,
2680 Font const & outerfont,
2681 pos_type initial) const
2685 bool emph_flag = false;
2686 bool bold_flag = false;
2689 Layout const & style = *d->layout_;
2691 if (!runparams.for_toc && runparams.html_make_pars) {
2692 // generate a magic label for this paragraph
2693 string const attr = "id='" + magicLabel() + "'";
2694 xs << html::CompTag("a", attr);
2698 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2700 // parsing main loop
2701 for (pos_type i = initial; i < size(); ++i) {
2702 // let's not show deleted material in the output
2706 Font font = getFont(buf.params(), i, outerfont);
2709 if (font_old.emph() != font.fontInfo().emph()) {
2710 if (font.fontInfo().emph() == FONT_ON) {
2711 xs << html::StartTag("em");
2713 } else if (emph_flag && i != initial) {
2714 xs << html::EndTag("em");
2719 if (font_old.series() != font.fontInfo().series()) {
2720 if (font.fontInfo().series() == BOLD_SERIES) {
2721 xs << html::StartTag("strong");
2723 } else if (bold_flag && i != initial) {
2724 xs << html::EndTag("strong");
2729 // Other such tags? What about the other text ranges?
2731 Inset const * inset = getInset(i);
2733 if (!runparams.for_toc || inset->isInToc()) {
2734 OutputParams np = runparams;
2735 if (!inset->getLayout().htmlisblock())
2736 np.html_in_par = true;
2737 retval += inset->xhtml(xs, np);
2740 char_type c = d->text_[i];
2742 if (style.pass_thru)
2744 else if (c == '-') {
2747 if (j < size() && d->text_[j] == '-') {
2749 if (j < size() && d->text_[j] == '-') {
2750 str += from_ascii("—");
2753 str += from_ascii("–");
2759 // We don't want to escape the entities. Note that
2760 // it is safe to do this, since str can otherwise
2761 // only be "-". E.g., it can't be "<".
2762 xs << XHTMLStream::NextRaw() << str;
2766 font_old = font.fontInfo();
2774 bool Paragraph::isHfill(pos_type pos) const
2776 Inset const * inset = getInset(pos);
2777 return inset && (inset->lyxCode() == SPACE_CODE &&
2778 inset->isStretchableSpace());
2782 bool Paragraph::isNewline(pos_type pos) const
2784 Inset const * inset = getInset(pos);
2785 return inset && inset->lyxCode() == NEWLINE_CODE;
2789 bool Paragraph::isLineSeparator(pos_type pos) const
2791 char_type const c = d->text_[pos];
2792 if (isLineSeparatorChar(c))
2794 Inset const * inset = getInset(pos);
2795 return inset && inset->isLineSeparator();
2799 bool Paragraph::isWordSeparator(pos_type pos) const
2801 if (Inset const * inset = getInset(pos))
2802 return !inset->isLetter();
2803 char_type const c = d->text_[pos];
2804 // We want to pass the ' and escape chars to the spellchecker
2805 static docstring const quote = from_utf8(lyxrc.spellchecker_esc_chars + '\'');
2806 return (!isLetterChar(c) && !isDigit(c) && !contains(quote, c))
2811 bool Paragraph::isChar(pos_type pos) const
2813 if (Inset const * inset = getInset(pos))
2814 return inset->isChar();
2815 char_type const c = d->text_[pos];
2816 return !isLetterChar(c) && !isDigit(c) && !lyx::isSpace(c);
2820 bool Paragraph::isSpace(pos_type pos) const
2822 if (Inset const * inset = getInset(pos))
2823 return inset->isSpace();
2824 char_type const c = d->text_[pos];
2825 return lyx::isSpace(c);
2830 Paragraph::getParLanguage(BufferParams const & bparams) const
2833 return getFirstFontSettings(bparams).language();
2834 // FIXME: we should check the prev par as well (Lgb)
2835 return bparams.language;
2839 bool Paragraph::isRTL(BufferParams const & bparams) const
2841 return lyxrc.rtl_support
2842 && getParLanguage(bparams)->rightToLeft()
2843 && !inInset().getLayout().forceLTR();
2847 void Paragraph::changeLanguage(BufferParams const & bparams,
2848 Language const * from, Language const * to)
2850 // change language including dummy font change at the end
2851 for (pos_type i = 0; i <= size(); ++i) {
2852 Font font = getFontSettings(bparams, i);
2853 if (font.language() == from) {
2854 font.setLanguage(to);
2858 d->requestSpellCheck(size());
2862 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
2864 Language const * doc_language = bparams.language;
2865 FontList::const_iterator cit = d->fontlist_.begin();
2866 FontList::const_iterator end = d->fontlist_.end();
2868 for (; cit != end; ++cit)
2869 if (cit->font().language() != ignore_language &&
2870 cit->font().language() != latex_language &&
2871 cit->font().language() != doc_language)
2877 void Paragraph::getLanguages(std::set<Language const *> & languages) const
2879 FontList::const_iterator cit = d->fontlist_.begin();
2880 FontList::const_iterator end = d->fontlist_.end();
2882 for (; cit != end; ++cit) {
2883 Language const * lang = cit->font().language();
2884 if (lang != ignore_language &&
2885 lang != latex_language)
2886 languages.insert(lang);
2891 docstring Paragraph::asString(int options) const
2893 return asString(0, size(), options);
2897 docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
2899 odocstringstream os;
2902 && options & AS_STR_LABEL
2903 && !d->params_.labelString().empty())
2904 os << d->params_.labelString() << ' ';
2906 for (pos_type i = beg; i < end; ++i) {
2907 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
2909 char_type const c = d->text_[i];
2910 if (isPrintable(c) || c == '\t'
2911 || (c == '\n' && (options & AS_STR_NEWLINES)))
2913 else if (c == META_INSET && (options & AS_STR_INSETS)) {
2914 getInset(i)->tocString(os);
2915 if (getInset(i)->asInsetMath())
2924 docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputParams & runparams) const
2926 odocstringstream os;
2929 && options & AS_STR_LABEL
2930 && !d->params_.labelString().empty())
2931 os << d->params_.labelString() << ' ';
2933 for (pos_type i = beg; i < end; ++i) {
2934 char_type const c = d->text_[i];
2935 if (isPrintable(c) || c == '\t'
2936 || (c == '\n' && (options & AS_STR_NEWLINES)))
2938 else if (c == META_INSET && (options & AS_STR_INSETS)) {
2939 getInset(i)->plaintext(os, runparams);
2947 void Paragraph::setInsetOwner(Inset const * inset)
2949 d->inset_owner_ = inset;
2953 int Paragraph::id() const
2959 void Paragraph::setId(int id)
2965 Layout const & Paragraph::layout() const
2971 void Paragraph::setLayout(Layout const & layout)
2973 d->layout_ = &layout;
2977 void Paragraph::setDefaultLayout(DocumentClass const & tc)
2979 setLayout(tc.defaultLayout());
2983 void Paragraph::setPlainLayout(DocumentClass const & tc)
2985 setLayout(tc.plainLayout());
2989 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
2991 if (usePlainLayout())
2992 setPlainLayout(tclass);
2994 setDefaultLayout(tclass);
2998 Inset const & Paragraph::inInset() const
3000 LASSERT(d->inset_owner_, throw ExceptionMessage(BufferException,
3001 _("Memory problem"), _("Paragraph not properly initialized")));
3002 return *d->inset_owner_;
3006 ParagraphParameters & Paragraph::params()
3012 ParagraphParameters const & Paragraph::params() const
3018 bool Paragraph::isFreeSpacing() const
3020 if (d->layout_->free_spacing)
3022 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3026 bool Paragraph::allowEmpty() const
3028 if (d->layout_->keepempty)
3030 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3034 char_type Paragraph::transformChar(char_type c, pos_type pos) const
3036 if (!Encodings::isArabicChar(c))
3039 char_type prev_char = ' ';
3040 char_type next_char = ' ';
3042 for (pos_type i = pos - 1; i >= 0; --i) {
3043 char_type const par_char = d->text_[i];
3044 if (!Encodings::isArabicComposeChar(par_char)) {
3045 prev_char = par_char;
3050 for (pos_type i = pos + 1, end = size(); i < end; ++i) {
3051 char_type const par_char = d->text_[i];
3052 if (!Encodings::isArabicComposeChar(par_char)) {
3053 next_char = par_char;
3058 if (Encodings::isArabicChar(next_char)) {
3059 if (Encodings::isArabicChar(prev_char) &&
3060 !Encodings::isArabicSpecialChar(prev_char))
3061 return Encodings::transformChar(c, Encodings::FORM_MEDIAL);
3063 return Encodings::transformChar(c, Encodings::FORM_INITIAL);
3065 if (Encodings::isArabicChar(prev_char) &&
3066 !Encodings::isArabicSpecialChar(prev_char))
3067 return Encodings::transformChar(c, Encodings::FORM_FINAL);
3069 return Encodings::transformChar(c, Encodings::FORM_ISOLATED);
3074 int Paragraph::checkBiblio(Buffer const & buffer)
3077 // This is getting more and more a mess. ...We really should clean
3078 // up this bibitem issue for 1.6. See also bug 2743.
3080 // Add bibitem insets if necessary
3081 if (d->layout_->labeltype != LABEL_BIBLIO)
3084 bool hasbibitem = !d->insetlist_.empty()
3085 // Insist on it being in pos 0
3086 && d->text_[0] == META_INSET
3087 && d->insetlist_.begin()->inset->lyxCode() == BIBITEM_CODE;
3089 bool track_changes = buffer.params().trackChanges;
3094 // remove a bibitem in pos != 0
3095 // restore it later in pos 0 if necessary
3096 // (e.g. if a user inserts contents _before_ the item)
3097 // we're assuming there's only one of these, which there
3099 int erasedInsetPosition = -1;
3100 InsetList::iterator it = d->insetlist_.begin();
3101 InsetList::iterator end = d->insetlist_.end();
3102 for (; it != end; ++it)
3103 if (it->inset->lyxCode() == BIBITEM_CODE
3105 InsetCommand * olditem = it->inset->asInsetCommand();
3106 oldkey = olditem->getParam("key");
3107 oldlabel = olditem->getParam("label");
3108 erasedInsetPosition = it->pos;
3109 eraseChar(erasedInsetPosition, track_changes);
3113 // There was an InsetBibitem at the beginning, and we didn't
3114 // have to erase one.
3115 if (hasbibitem && erasedInsetPosition < 0)
3118 // There was an InsetBibitem at the beginning and we did have to
3119 // erase one. So we give its properties to the beginning inset.
3121 InsetCommand * inset = d->insetlist_.begin()->inset->asInsetCommand();
3122 if (!oldkey.empty())
3123 inset->setParam("key", oldkey);
3124 inset->setParam("label", oldlabel);
3125 return -erasedInsetPosition;
3128 // There was no inset at the beginning, so we need to create one with
3129 // the key and label of the one we erased.
3130 InsetBibitem * inset =
3131 new InsetBibitem(const_cast<Buffer *>(&buffer), InsetCommandParams(BIBITEM_CODE));
3132 // restore values of previously deleted item in this par.
3133 if (!oldkey.empty())
3134 inset->setParam("key", oldkey);
3135 inset->setParam("label", oldlabel);
3136 insertInset(0, inset,
3137 Change(track_changes ? Change::INSERTED : Change::UNCHANGED));
3143 void Paragraph::checkAuthors(AuthorList const & authorList)
3145 d->changes_.checkAuthors(authorList);
3149 bool Paragraph::isChanged(pos_type pos) const
3151 return lookupChange(pos).changed();
3155 bool Paragraph::isInserted(pos_type pos) const
3157 return lookupChange(pos).inserted();
3161 bool Paragraph::isDeleted(pos_type pos) const
3163 return lookupChange(pos).deleted();
3167 InsetList const & Paragraph::insetList() const
3169 return d->insetlist_;
3173 void Paragraph::setBuffer(Buffer & b)
3175 d->insetlist_.setBuffer(b);
3179 Inset * Paragraph::releaseInset(pos_type pos)
3181 Inset * inset = d->insetlist_.release(pos);
3182 /// does not honour change tracking!
3183 eraseChar(pos, false);
3188 Inset * Paragraph::getInset(pos_type pos)
3190 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3191 ? d->insetlist_.get(pos) : 0;
3195 Inset const * Paragraph::getInset(pos_type pos) const
3197 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3198 ? d->insetlist_.get(pos) : 0;
3202 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3203 pos_type & right, TextCase action)
3205 // process sequences of modified characters; in change
3206 // tracking mode, this approach results in much better
3207 // usability than changing case on a char-by-char basis
3210 bool const trackChanges = bparams.trackChanges;
3212 bool capitalize = true;
3214 for (; pos < right; ++pos) {
3215 char_type oldChar = d->text_[pos];
3216 char_type newChar = oldChar;
3218 // ignore insets and don't play with deleted text!
3219 if (oldChar != META_INSET && !isDeleted(pos)) {
3221 case text_lowercase:
3222 newChar = lowercase(oldChar);
3224 case text_capitalization:
3226 newChar = uppercase(oldChar);
3230 case text_uppercase:
3231 newChar = uppercase(oldChar);
3236 if (isWordSeparator(pos) || isDeleted(pos)) {
3237 // permit capitalization again
3241 if (oldChar != newChar) {
3243 if (pos != right - 1)
3245 // step behind the changing area
3249 int erasePos = pos - changes.size();
3250 for (size_t i = 0; i < changes.size(); i++) {
3251 insertChar(pos, changes[i],
3252 getFontSettings(bparams,
3255 if (!eraseChar(erasePos, trackChanges)) {
3258 ++right; // expand selection
3266 int Paragraph::find(docstring const & str, bool cs, bool mw,
3267 pos_type start_pos, bool del) const
3269 pos_type pos = start_pos;
3270 int const strsize = str.length();
3272 pos_type const parsize = d->text_.size();
3273 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3274 // Ignore ligature break and hyphenation chars while searching
3275 while (pos < parsize - 1 && isInset(pos)) {
3276 const InsetSpecialChar *isc = dynamic_cast<const InsetSpecialChar*>(getInset(pos));
3278 || (isc->kind() != InsetSpecialChar::HYPHENATION
3279 && isc->kind() != InsetSpecialChar::LIGATURE_BREAK))
3283 if (cs && str[i] != d->text_[pos])
3285 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3287 if (!del && isDeleted(pos))
3294 // if necessary, check whether string matches word
3296 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3299 && !isWordSeparator(pos))
3303 return pos - start_pos;
3307 char_type Paragraph::getChar(pos_type pos) const
3309 return d->text_[pos];
3313 pos_type Paragraph::size() const
3315 return d->text_.size();
3319 bool Paragraph::empty() const
3321 return d->text_.empty();
3325 bool Paragraph::isInset(pos_type pos) const
3327 return d->text_[pos] == META_INSET;
3331 bool Paragraph::isSeparator(pos_type pos) const
3333 //FIXME: Are we sure this can be the only separator?
3334 return d->text_[pos] == ' ';
3338 void Paragraph::deregisterWords()
3340 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3341 Private::LangWordsMap::const_iterator ite = d->words_.end();
3342 for (; itl != ite; ++itl) {
3343 WordList * wl = theWordList(itl->first);
3344 Private::Words::const_iterator it = (itl->second).begin();
3345 Private::Words::const_iterator et = (itl->second).end();
3346 for (; it != et; ++it)
3353 void Paragraph::locateWord(pos_type & from, pos_type & to,
3354 word_location const loc) const
3357 case WHOLE_WORD_STRICT:
3358 if (from == 0 || from == size()
3359 || isWordSeparator(from)
3360 || isWordSeparator(from - 1)) {
3364 // no break here, we go to the next
3367 // If we are already at the beginning of a word, do nothing
3368 if (!from || isWordSeparator(from - 1))
3370 // no break here, we go to the next
3373 // always move the cursor to the beginning of previous word
3374 while (from && !isWordSeparator(from - 1))
3378 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3381 // no need to move the 'from' cursor
3385 while (to < size() && !isWordSeparator(to))
3390 void Paragraph::collectWords()
3392 // This is the value that needs to be exposed in the preferences
3393 // to resolve bug #6760.
3394 static int minlength = 6;
3395 pos_type n = size();
3396 for (pos_type pos = 0; pos < n; ++pos) {
3397 if (isWordSeparator(pos))
3399 pos_type from = pos;
3400 locateWord(from, pos, WHOLE_WORD);
3401 if (pos - from >= minlength) {
3402 docstring word = asString(from, pos, AS_STR_NONE);
3403 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
3404 if (cit == d->fontlist_.end())
3406 Language const * lang = cit->font().language();
3407 d->words_[*lang].insert(word);
3413 void Paragraph::registerWords()
3415 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3416 Private::LangWordsMap::const_iterator ite = d->words_.end();
3417 for (; itl != ite; ++itl) {
3418 WordList * wl = theWordList(itl->first);
3419 Private::Words::const_iterator it = (itl->second).begin();
3420 Private::Words::const_iterator et = (itl->second).end();
3421 for (; it != et; ++it)
3427 void Paragraph::updateWords()
3435 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3437 SkipPositionsIterator begin = skips.begin();
3438 SkipPositions::iterator end = skips.end();
3439 if (pos > 0 && begin < end) {
3441 if (end->last == pos - 1) {
3446 skips.insert(end, FontSpan(pos, pos));
3450 Language * Paragraph::Private::locateSpellRange(
3451 pos_type & from, pos_type & to,
3452 SkipPositions & skips) const
3454 // skip leading white space
3455 while (from < to && owner_->isWordSeparator(from))
3457 // don't check empty range
3460 // get current language
3461 Language * lang = getSpellLanguage(from);
3462 pos_type last = from;
3463 bool samelang = true;
3464 bool sameinset = true;
3465 while (last < to && samelang && sameinset) {
3466 // hop to end of word
3467 while (last < to && !owner_->isWordSeparator(last)) {
3468 if (owner_->getInset(last)) {
3469 appendSkipPosition(skips, last);
3470 } else if (owner_->isDeleted(last)) {
3471 appendSkipPosition(skips, last);
3475 // hop to next word while checking for insets
3476 while (sameinset && last < to && owner_->isWordSeparator(last)) {
3477 if (Inset const * inset = owner_->getInset(last))
3478 sameinset = inset->isChar() && inset->isLetter();
3479 if (sameinset && owner_->isDeleted(last)) {
3480 appendSkipPosition(skips, last);
3485 if (sameinset && last < to) {
3486 // now check for language change
3487 samelang = lang == getSpellLanguage(last);
3490 // if language change detected backstep is needed
3498 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
3501 const_cast<Language *>(owner_->getFontSettings(
3502 inset_owner_->buffer().params(), from).language());
3503 if (lang == inset_owner_->buffer().params().language
3504 && !lyxrc.spellchecker_alt_lang.empty()) {
3506 string const lang_variety =
3507 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3508 lang->setCode(lang_code);
3509 lang->setVariety(lang_variety);
3515 void Paragraph::requestSpellCheck(pos_type pos)
3517 d->requestSpellCheck(pos == -1 ? size() : pos);
3521 bool Paragraph::needsSpellCheck() const
3523 SpellChecker::ChangeNumber speller_change_number = 0;
3524 if (theSpellChecker())
3525 speller_change_number = theSpellChecker()->changeNumber();
3526 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
3527 d->speller_state_.needsCompleteRefresh(speller_change_number);
3529 return d->needsSpellCheck();
3533 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
3534 WordLangTuple & wl, docstring_list & suggestions,
3535 bool do_suggestion, bool check_learned) const
3537 SpellChecker::Result result = SpellChecker::WORD_OK;
3538 SpellChecker * speller = theSpellChecker();
3542 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
3545 locateWord(from, to, WHOLE_WORD);
3546 if (from == to || from >= size())
3549 docstring word = asString(from, to, AS_STR_INSETS + AS_STR_SKIPDELETE);
3550 Language * lang = d->getSpellLanguage(from);
3552 wl = WordLangTuple(word, lang);
3557 if (needsSpellCheck() || check_learned) {
3558 // Ignore words with digits
3559 // FIXME: make this customizable
3560 // (note that some checkers ignore words with digits by default)
3561 if (!hasDigit(word)) {
3562 bool const trailing_dot = to < size() && d->text_[to] == '.';
3563 result = speller->check(wl);
3564 if (SpellChecker::misspelled(result) && trailing_dot) {
3565 wl = WordLangTuple(word.append(from_ascii(".")), lang);
3566 result = speller->check(wl);
3567 if (!SpellChecker::misspelled(result)) {
3568 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
3570 from << ".." << to << "]");
3574 d->setMisspelled(from, to, result);
3576 result = d->speller_state_.getState(from);
3579 bool const misspelled_ = SpellChecker::misspelled(result) ;
3580 if (misspelled_ && do_suggestion)
3581 speller->suggest(wl, suggestions);
3582 else if (misspelled_)
3583 LYXERR(Debug::GUI, "misspelled word: \"" <<
3585 from << ".." << to << "]");
3587 suggestions.clear();
3593 void Paragraph::Private::markMisspelledWords(
3594 pos_type const & first, pos_type const & last,
3595 SpellChecker::Result result,
3596 docstring const & word,
3597 SkipPositions const & skips)
3599 if (!SpellChecker::misspelled(result)) {
3600 setMisspelled(first, last, SpellChecker::WORD_OK);
3604 SpellChecker * speller = theSpellChecker();
3605 // locate and enumerate the error positions
3606 int nerrors = speller->numMisspelledWords();
3608 SkipPositionsIterator it = skips.begin();
3609 SkipPositionsIterator et = skips.end();
3610 for (int index = 0; index < nerrors; ++index) {
3613 speller->misspelledWord(index, wstart, wlen);
3614 /// should not happen if speller supports range checks
3615 if (!wlen) continue;
3616 docstring const misspelled = word.substr(wstart, wlen);
3617 wstart += first + numskipped;
3618 if (snext < wstart) {
3619 /// mark the range of correct spelling
3620 numskipped += countSkips(it, et, wstart);
3621 setMisspelled(snext,
3622 wstart - 1, SpellChecker::WORD_OK);
3624 snext = wstart + wlen;
3625 numskipped += countSkips(it, et, snext);
3626 /// mark the range of misspelling
3627 setMisspelled(wstart, snext, result);
3628 LYXERR(Debug::GUI, "misspelled word: \"" <<
3629 misspelled << "\" [" <<
3630 wstart << ".." << (snext-1) << "]");
3633 if (snext <= last) {
3634 /// mark the range of correct spelling at end
3635 setMisspelled(snext, last, SpellChecker::WORD_OK);
3640 void Paragraph::spellCheck() const
3642 SpellChecker * speller = theSpellChecker();
3643 if (!speller || !size() ||!needsSpellCheck())
3647 d->rangeOfSpellCheck(start, endpos);
3648 if (speller->canCheckParagraph()) {
3649 // loop until we leave the range
3650 for (pos_type first = start; first < endpos; ) {
3651 pos_type last = endpos;
3652 Private::SkipPositions skips;
3653 Language * lang = d->locateSpellRange(first, last, skips);
3654 if (first >= endpos)
3656 // start the spell checker on the unit of meaning
3657 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
3658 WordLangTuple wl = WordLangTuple(word, lang);
3659 SpellChecker::Result result = word.size() ?
3660 speller->check(wl) : SpellChecker::WORD_OK;
3661 d->markMisspelledWords(first, last, result, word, skips);
3665 static docstring_list suggestions;
3666 pos_type to = endpos;
3667 while (start < endpos) {
3669 spellCheck(start, to, wl, suggestions, false);
3673 d->readySpellCheck();
3677 bool Paragraph::isMisspelled(pos_type pos) const
3679 return SpellChecker::misspelled(d->speller_state_.getState(pos));
3683 string Paragraph::magicLabel() const
3686 ss << "magicparlabel-" << id();