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"
55 #include "support/debug.h"
56 #include "support/docstring_list.h"
57 #include "support/ExceptionMessage.h"
58 #include "support/gettext.h"
59 #include "support/lassert.h"
60 #include "support/lstrings.h"
61 #include "support/textutils.h"
67 using namespace lyx::support;
72 /// Inset identifier (above 0x10ffff, for ucs-4)
73 char_type const META_INSET = 0x200001;
76 /////////////////////////////////////////////////////////////////////
80 /////////////////////////////////////////////////////////////////////
82 class Paragraph::Private
86 Private(Paragraph * owner, Layout const & layout);
87 /// "Copy constructor"
88 Private(Private const &, Paragraph * owner);
89 /// Copy constructor from \p beg to \p end
90 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
93 void insertChar(pos_type pos, char_type c, Change const & change);
95 /// Output the surrogate pair formed by \p c and \p next to \p os.
96 /// \return the number of characters written.
97 int latexSurrogatePair(odocstream & os, char_type c, char_type next,
98 OutputParams const &);
100 /// Output a space in appropriate formatting (or a surrogate pair
101 /// if the next character is a combining character).
102 /// \return whether a surrogate pair was output.
103 bool simpleTeXBlanks(OutputParams const &,
104 odocstream &, TexRow & texrow,
106 unsigned int & column,
108 Layout const & style);
110 /// Output consecutive unicode chars, belonging to the same script as
111 /// specified by the latex macro \p ltx, to \p os starting from \p i.
112 /// \return the number of characters written.
113 int writeScriptChars(odocstream & os, docstring const & ltx,
114 Change const &, Encoding const &, pos_type & i);
116 /// This could go to ParagraphParameters if we want to.
117 int startTeXParParams(BufferParams const &, odocstream &, TexRow &,
118 OutputParams const &) const;
120 /// This could go to ParagraphParameters if we want to.
121 int endTeXParParams(BufferParams const &, odocstream &, TexRow &,
122 OutputParams const &) const;
125 void latexInset(BufferParams const &,
127 TexRow & texrow, OutputParams &,
130 Font const & outerfont,
132 Change & running_change,
133 Layout const & style,
135 unsigned int & column);
138 void latexSpecialChar(
140 OutputParams const & runparams,
141 Font const & running_font,
142 Change const & running_change,
143 Layout const & style,
145 unsigned int & column);
152 unsigned int & column);
154 bool latexSpecialTypewriter(
158 unsigned int & column);
160 bool latexSpecialPhrase(
163 unsigned int & column,
164 OutputParams const & runparams);
167 void validate(LaTeXFeatures & features) const;
169 /// Checks if the paragraph contains only text and no inset or font change.
170 bool onlyText(Buffer const & buf, Font const & outerfont,
171 pos_type initial) const;
173 /// match a string against a particular point in the paragraph
174 bool isTextAt(string const & str, pos_type pos) const;
177 InsetCode ownerCode() const
179 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
182 /// Which Paragraph owns us?
186 Inset const * inset_owner_;
195 ParagraphParameters params_;
197 /// for recording and looking up changes
201 InsetList insetlist_;
204 pos_type begin_of_body_;
206 typedef docstring TextContainer;
210 typedef set<docstring> Words;
211 typedef map<Language, Words> LangWordsMap;
215 Layout const * layout_;
221 struct special_phrase {
227 special_phrase const special_phrases[] = {
228 { "LyX", from_ascii("\\LyX{}"), false },
229 { "TeX", from_ascii("\\TeX{}"), true },
230 { "LaTeX2e", from_ascii("\\LaTeXe{}"), true },
231 { "LaTeX", from_ascii("\\LaTeX{}"), true },
234 size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase);
239 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
240 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
246 // Initialization of the counter for the paragraph id's,
248 // FIXME: There should be a more intelligent way to generate and use the
249 // paragraph ids per buffer instead a global static counter for all InsetText
250 // in the running program.
251 static int paragraph_id = -1;
253 Paragraph::Private::Private(Private const & p, Paragraph * owner)
254 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
255 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
256 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
259 id_ = ++paragraph_id;
263 Paragraph::Private::Private(Private const & p, Paragraph * owner,
264 pos_type beg, pos_type end)
265 : owner_(owner), inset_owner_(p.inset_owner_),
266 params_(p.params_), changes_(p.changes_),
267 insetlist_(p.insetlist_, beg, end),
268 begin_of_body_(p.begin_of_body_), words_(p.words_),
271 id_ = ++paragraph_id;
272 if (beg >= pos_type(p.text_.size()))
274 text_ = p.text_.substr(beg, end - beg);
276 FontList::const_iterator fcit = fontlist_.begin();
277 FontList::const_iterator fend = fontlist_.end();
278 for (; fcit != fend; ++fcit) {
279 if (fcit->pos() < beg)
281 if (fcit->pos() >= end) {
282 // Add last entry in the fontlist_.
283 fontlist_.set(text_.size() - 1, fcit->font());
286 // Add a new entry in the fontlist_.
287 fontlist_.set(fcit->pos() - beg, fcit->font());
292 void Paragraph::addChangesToToc(DocIterator const & cdit,
293 Buffer const & buf) const
295 d->changes_.addToToc(cdit, buf);
299 bool Paragraph::isDeleted(pos_type start, pos_type end) const
301 LASSERT(start >= 0 && start <= size(), /**/);
302 LASSERT(end > start && end <= size() + 1, /**/);
304 return d->changes_.isDeleted(start, end);
308 bool Paragraph::isChanged(pos_type start, pos_type end) const
310 LASSERT(start >= 0 && start <= size(), /**/);
311 LASSERT(end > start && end <= size() + 1, /**/);
313 return d->changes_.isChanged(start, end);
317 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
319 // keep the logic here in sync with the logic of eraseChars()
323 Change const change = d->changes_.lookup(size());
324 return change.inserted() && change.currentAuthor();
328 void Paragraph::setChange(Change const & change)
330 // beware of the imaginary end-of-par character!
331 d->changes_.set(change, 0, size() + 1);
334 * Propagate the change recursively - but not in case of DELETED!
336 * Imagine that your co-author makes changes in an existing inset. He
337 * sends your document to you and you come to the conclusion that the
338 * inset should go completely. If you erase it, LyX must not delete all
339 * text within the inset. Otherwise, the change tracked insertions of
340 * your co-author get lost and there is no way to restore them later.
342 * Conclusion: An inset's content should remain untouched if you delete it
345 if (!change.deleted()) {
346 for (pos_type pos = 0; pos < size(); ++pos) {
347 if (Inset * inset = getInset(pos))
348 inset->setChange(change);
354 void Paragraph::setChange(pos_type pos, Change const & change)
356 LASSERT(pos >= 0 && pos <= size(), /**/);
357 d->changes_.set(change, pos);
359 // see comment in setChange(Change const &) above
360 if (!change.deleted() && pos < size())
361 if (Inset * inset = getInset(pos))
362 inset->setChange(change);
366 Change const & Paragraph::lookupChange(pos_type pos) const
368 LASSERT(pos >= 0 && pos <= size(), /**/);
369 return d->changes_.lookup(pos);
373 void Paragraph::acceptChanges(pos_type start, pos_type end)
375 LASSERT(start >= 0 && start <= size(), /**/);
376 LASSERT(end > start && end <= size() + 1, /**/);
378 for (pos_type pos = start; pos < end; ++pos) {
379 switch (lookupChange(pos).type) {
380 case Change::UNCHANGED:
381 // accept changes in nested inset
382 if (Inset * inset = getInset(pos))
383 inset->acceptChanges();
386 case Change::INSERTED:
387 d->changes_.set(Change(Change::UNCHANGED), pos);
388 // also accept changes in nested inset
389 if (Inset * inset = getInset(pos))
390 inset->acceptChanges();
393 case Change::DELETED:
394 // Suppress access to non-existent
395 // "end-of-paragraph char"
397 eraseChar(pos, false);
408 void Paragraph::rejectChanges(pos_type start, pos_type end)
410 LASSERT(start >= 0 && start <= size(), /**/);
411 LASSERT(end > start && end <= size() + 1, /**/);
413 for (pos_type pos = start; pos < end; ++pos) {
414 switch (lookupChange(pos).type) {
415 case Change::UNCHANGED:
416 // reject changes in nested inset
417 if (Inset * inset = getInset(pos))
418 inset->rejectChanges();
421 case Change::INSERTED:
422 // Suppress access to non-existent
423 // "end-of-paragraph char"
425 eraseChar(pos, false);
431 case Change::DELETED:
432 d->changes_.set(Change(Change::UNCHANGED), pos);
434 // Do NOT reject changes within a deleted inset!
435 // There may be insertions of a co-author inside of it!
443 void Paragraph::Private::insertChar(pos_type pos, char_type c,
444 Change const & change)
446 LASSERT(pos >= 0 && pos <= int(text_.size()), /**/);
449 changes_.insert(change, pos);
451 // This is actually very common when parsing buffers (and
452 // maybe inserting ascii text)
453 if (pos == pos_type(text_.size())) {
454 // when appending characters, no need to update tables
459 text_.insert(text_.begin() + pos, c);
461 // Update the font table.
462 fontlist_.increasePosAfterPos(pos);
465 insetlist_.increasePosAfterPos(pos);
469 bool Paragraph::insertInset(pos_type pos, Inset * inset,
470 Change const & change)
472 LASSERT(inset, /**/);
473 LASSERT(pos >= 0 && pos <= size(), /**/);
475 // Paragraph::insertInset() can be used in cut/copy/paste operation where
476 // d->inset_owner_ is not set yet.
477 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
480 d->insertChar(pos, META_INSET, change);
481 LASSERT(d->text_[pos] == META_INSET, /**/);
483 // Add a new entry in the insetlist_.
484 d->insetlist_.insert(inset, pos);
489 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
491 LASSERT(pos >= 0 && pos <= size(), return false);
493 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
496 Change change = d->changes_.lookup(pos);
498 // set the character to DELETED if
499 // a) it was previously unchanged or
500 // b) it was inserted by a co-author
502 if (!change.changed() ||
503 (change.inserted() && !change.currentAuthor())) {
504 setChange(pos, Change(Change::DELETED));
508 if (change.deleted())
512 // Don't physically access the imaginary end-of-paragraph character.
513 // eraseChar() can only mark it as DELETED. A physical deletion of
514 // end-of-par must be handled externally.
520 d->changes_.erase(pos);
522 // if it is an inset, delete the inset entry
523 if (d->text_[pos] == META_INSET)
524 d->insetlist_.erase(pos);
526 d->text_.erase(d->text_.begin() + pos);
528 // Update the fontlist_
529 d->fontlist_.erase(pos);
531 // Update the insetlist_
532 d->insetlist_.decreasePosAfterPos(pos);
538 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
540 LASSERT(start >= 0 && start <= size(), /**/);
541 LASSERT(end >= start && end <= size() + 1, /**/);
544 for (pos_type count = end - start; count; --count) {
545 if (!eraseChar(i, trackChanges))
552 int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
553 char_type next, OutputParams const & runparams)
555 // Writing next here may circumvent a possible font change between
556 // c and next. Since next is only output if it forms a surrogate pair
557 // with c we can ignore this:
558 // A font change inside a surrogate pair does not make sense and is
559 // hopefully impossible to input.
560 // FIXME: change tracking
561 // Is this correct WRT change tracking?
562 Encoding const & encoding = *(runparams.encoding);
563 docstring const latex1 = encoding.latexChar(next);
564 docstring const latex2 = encoding.latexChar(c);
565 if (docstring(1, next) == latex1) {
566 // the encoding supports the combination
567 os << latex2 << latex1;
568 return latex1.length() + latex2.length();
569 } else if (runparams.local_font &&
570 runparams.local_font->language()->lang() == "polutonikogreek") {
571 // polutonikogreek only works without the brackets
572 os << latex1 << latex2;
573 return latex1.length() + latex2.length();
575 os << latex1 << '{' << latex2 << '}';
576 return latex1.length() + latex2.length() + 2;
580 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
581 odocstream & os, TexRow & texrow,
583 unsigned int & column,
585 Layout const & style)
587 if (style.pass_thru || runparams.verbatim)
590 if (i + 1 < int(text_.size())) {
591 char_type next = text_[i + 1];
592 if (Encodings::isCombiningChar(next)) {
593 // This space has an accent, so we must always output it.
594 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
599 if (runparams.linelen > 0
600 && column > runparams.linelen
602 && text_[i - 1] != ' '
603 && (i + 1 < int(text_.size()))
604 // same in FreeSpacing mode
605 && !owner_->isFreeSpacing()
606 // In typewriter mode, we want to avoid
607 // ! . ? : at the end of a line
608 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
609 && (text_[i - 1] == '.'
610 || text_[i - 1] == '?'
611 || text_[i - 1] == ':'
612 || text_[i - 1] == '!'))) {
615 texrow.start(owner_->id(), i + 1);
617 } else if (style.free_spacing) {
626 int Paragraph::Private::writeScriptChars(odocstream & os,
627 docstring const & ltx,
628 Change const & runningChange,
629 Encoding const & encoding,
632 // FIXME: modifying i here is not very nice...
634 // We only arrive here when a proper language for character text_[i] has
635 // not been specified (i.e., it could not be translated in the current
636 // latex encoding) or its latex translation has been forced, and it
637 // belongs to a known script.
638 // Parameter ltx contains the latex translation of text_[i] as specified
639 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
640 // The latex macro name "textXXX" specifies the script to which text_[i]
641 // belongs and we use it in order to check whether characters from the
642 // same script immediately follow, such that we can collect them in a
643 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
644 // for the first char but only "<spec>" for all subsequent chars.
645 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
646 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
647 string script = to_ascii(ltx.substr(1, brace1 - 1));
650 bool closing_brace = true;
651 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
652 // Correct encoding is being used, so we can avoid \textgreek.
655 closing_brace = false;
657 os << ltx.substr(pos, length);
658 int size = text_.size();
659 while (i + 1 < size) {
660 char_type const next = text_[i + 1];
661 // Stop here if next character belongs to another script
662 // or there is a change in change tracking status.
663 if (!Encodings::isKnownScriptChar(next, script) ||
664 runningChange != owner_->lookupChange(i + 1))
668 FontList::const_iterator cit = fontlist_.begin();
669 FontList::const_iterator end = fontlist_.end();
670 for (; cit != end; ++cit) {
671 if (cit->pos() >= i && !found) {
672 prev_font = cit->font();
675 if (cit->pos() >= i + 1)
678 // Stop here if there is a font attribute or encoding change.
679 if (found && cit != end && prev_font != cit->font())
681 docstring const latex = encoding.latexChar(next);
682 docstring::size_type const b1 =
683 latex.find_first_of(from_ascii("{"));
684 docstring::size_type const b2 =
685 latex.find_last_of(from_ascii("}"));
686 int const len = b2 - b1 - 1;
687 os << latex.substr(b1 + 1, len);
699 bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
701 pos_type const len = str.length();
703 // is the paragraph large enough?
704 if (pos + len > int(text_.size()))
707 // does the wanted text start at point?
708 for (string::size_type i = 0; i < str.length(); ++i) {
709 // Caution: direct comparison of characters works only
710 // because str is pure ASCII.
711 if (str[i] != text_[pos + i])
715 return fontlist_.hasChangeInRange(pos, len);
719 void Paragraph::Private::latexInset(
720 BufferParams const & bparams,
723 OutputParams & runparams,
726 Font const & outerfont,
728 Change & running_change,
729 Layout const & style,
731 unsigned int & column)
733 Inset * inset = owner_->getInset(i);
734 LASSERT(inset, /**/);
736 if (style.pass_thru) {
737 inset->plaintext(os, runparams);
741 // FIXME: move this to InsetNewline::latex
742 if (inset->lyxCode() == NEWLINE_CODE) {
743 // newlines are handled differently here than
744 // the default in simpleTeXSpecialChars().
745 if (!style.newline_allowed) {
749 column += running_font.latexWriteEndChanges(
750 os, bparams, runparams,
755 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
758 basefont = owner_->getLayoutFont(bparams, outerfont);
759 running_font = basefont;
761 if (runparams.moving_arg)
766 texrow.start(owner_->id(), i + 1);
770 if (owner_->isDeleted(i)) {
771 if( ++runparams.inDeletedInset == 1)
772 runparams.changeOfDeletedInset = owner_->lookupChange(i);
775 if (inset->canTrackChanges()) {
776 column += Changes::latexMarkChange(os, bparams, running_change,
777 Change(Change::UNCHANGED), runparams);
778 running_change = Change(Change::UNCHANGED);
782 odocstream::pos_type const len = os.tellp();
784 if (inset->forceLTR()
785 && running_font.isRightToLeft()
786 // ERT is an exception, it should be output with no
787 // decorations at all
788 && inset->lyxCode() != ERT_CODE) {
789 if (running_font.language()->lang() == "farsi")
796 // FIXME: Bug: we can have an empty font change here!
797 // if there has just been a font change, we are going to close it
798 // right now, which means stupid latex code like \textsf{}. AFAIK,
799 // this does not harm dvi output. A minor bug, thus (JMarc)
801 // Some insets cannot be inside a font change command.
802 // However, even such insets *can* be placed in \L or \R
803 // or their equivalents (for RTL language switches), so we don't
804 // close the language in those cases.
805 // ArabTeX, though, cannot handle this special behavior, it seems.
806 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
807 || running_font.language()->lang() == "arabic_arabtex";
808 if (open_font && inset->noFontChange()) {
809 bool closeLanguage = arabtex
810 || basefont.isRightToLeft() == running_font.isRightToLeft();
811 unsigned int count = running_font.latexWriteEndChanges(os,
812 bparams, runparams, basefont, basefont, closeLanguage);
814 // if any font properties were closed, update the running_font,
815 // making sure, however, to leave the language as it was
817 // FIXME: probably a better way to keep track of the old
818 // language, than copying the entire font?
819 Font const copy_font(running_font);
820 basefont = owner_->getLayoutFont(bparams, outerfont);
821 running_font = basefont;
823 running_font.setLanguage(copy_font.language());
824 // leave font open if language is still open
825 open_font = (running_font.language() == basefont.language());
827 runparams.local_font = &basefont;
834 tmp = inset->latex(os, runparams);
835 } catch (EncodingException & e) {
836 // add location information and throw again.
843 if (running_font.language()->lang() == "farsi")
850 texrow.newlines(tmp);
851 texrow.start(owner_->id(), i + 1);
854 column += os.tellp() - len;
857 if (owner_->isDeleted(i))
858 --runparams.inDeletedInset;
862 void Paragraph::Private::latexSpecialChar(
864 OutputParams const & runparams,
865 Font const & running_font,
866 Change const & running_change,
867 Layout const & style,
869 unsigned int & column)
871 char_type const c = text_[i];
873 if (style.pass_thru) {
875 // FIXME UNICODE: This can fail if c cannot
876 // be encoded in the current encoding.
881 if (runparams.verbatim) {
882 // FIXME UNICODE: This can fail if c cannot
883 // be encoded in the current encoding.
888 // If T1 font encoding is used, use the special
889 // characters it provides.
890 // NOTE: some languages reset the font encoding
892 if (!running_font.language()->internalFontEncoding()
893 && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
896 // \tt font needs special treatment
897 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
898 && latexSpecialTypewriter(c, os, i, column))
901 // Otherwise, we use what LaTeX provides us.
904 os << "\\textbackslash{}";
908 os << "\\textless{}";
912 os << "\\textgreater{}";
923 os << "\\char`\\\"{}";
928 case '%': case '#': case '{':
936 os << "\\textasciitilde{}";
941 os << "\\textasciicircum{}";
948 // avoid being mistaken for optional arguments
956 // Blanks are printed before font switching.
957 // Sure? I am not! (try nice-latex)
958 // I am sure it's correct. LyX might be smarter
959 // in the future, but for now, nothing wrong is
965 if (latexSpecialPhrase(os, i, column, runparams))
971 Encoding const & encoding = *(runparams.encoding);
972 if (i + 1 < int(text_.size())) {
973 char_type next = text_[i + 1];
974 if (Encodings::isCombiningChar(next)) {
975 column += latexSurrogatePair(os, c, next, runparams) - 1;
981 docstring const latex = encoding.latexChar(c);
982 if (Encodings::isKnownScriptChar(c, script)
983 && prefixIs(latex, from_ascii("\\" + script)))
984 column += writeScriptChars(os, latex,
985 running_change, encoding, i) - 1;
986 else if (latex.length() > 1 && latex[latex.length() - 1] != '}') {
987 // Prevent eating of a following
988 // space or command corruption by
989 // following characters
990 column += latex.length() + 1;
993 column += latex.length() - 1;
1001 bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
1002 pos_type i, unsigned int & column)
1008 // In T1 encoding, these characters exist
1009 // but we should avoid ligatures
1010 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1012 os << "\\textcompwordmark{}";
1019 // soul.sty breaks with \char`\"
1020 os << "\\textquotedbl{}";
1029 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream & os,
1030 pos_type i, unsigned int & column)
1034 // within \ttfamily, "--" is merged to "-" (no endash)
1035 // so we avoid this rather irritating ligature
1036 if (i + 1 < int(text_.size()) && text_[i + 1] == '-') {
1043 // everything else has to be checked separately
1044 // (depending on the encoding)
1051 bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
1052 unsigned int & column, OutputParams const & runparams)
1054 // FIXME: if we have "LaTeX" with a font
1055 // change in the middle (before the 'T', then
1056 // the "TeX" part is still special cased.
1057 // Really we should only operate this on
1058 // "words" for some definition of word
1060 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1061 if (!isTextAt(special_phrases[pnr].phrase, i))
1063 if (runparams.moving_arg)
1065 os << special_phrases[pnr].macro;
1066 i += special_phrases[pnr].phrase.length() - 1;
1067 column += special_phrases[pnr].macro.length() - 1;
1074 void Paragraph::Private::validate(LaTeXFeatures & features) const
1076 if (layout_->inpreamble && inset_owner_) {
1077 Buffer const & buf = inset_owner_->buffer();
1078 BufferParams const & bp = buf.params();
1081 odocstringstream ods;
1082 // we have to provide all the optional arguments here, even though
1083 // the last one is the only one we care about.
1084 owner_->latex(bp, f, ods, tr, features.runparams(), 0, -1, true);
1085 docstring const d = ods.str();
1087 // this will have "{" at the beginning, but not at the end
1088 string const content = to_utf8(d);
1089 string const cmd = layout_->latexname();
1090 features.addPreambleSnippet("\\" + cmd + content + "}");
1094 if (features.runparams().flavor == OutputParams::HTML
1095 && layout_->htmltitle()) {
1096 features.setHTMLTitle(owner_->asString(AS_STR_INSETS));
1099 // check the params.
1100 if (!params_.spacing().isDefault())
1101 features.require("setspace");
1104 features.useLayout(layout_->name());
1107 fontlist_.validate(features);
1109 // then the indentation
1110 if (!params_.leftIndent().zero())
1111 features.require("ParagraphLeftIndent");
1114 InsetList::const_iterator icit = insetlist_.begin();
1115 InsetList::const_iterator iend = insetlist_.end();
1116 for (; icit != iend; ++icit) {
1118 icit->inset->validate(features);
1119 if (layout_->needprotect &&
1120 icit->inset->lyxCode() == FOOT_CODE)
1121 features.require("NeedLyXFootnoteCode");
1125 // then the contents
1126 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1127 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1128 if (!special_phrases[pnr].builtin
1129 && isTextAt(special_phrases[pnr].phrase, i)) {
1130 features.require(special_phrases[pnr].phrase);
1134 Encodings::validate(text_[i], features);
1138 /////////////////////////////////////////////////////////////////////
1142 /////////////////////////////////////////////////////////////////////
1145 Layout const emptyParagraphLayout;
1148 Paragraph::Paragraph()
1149 : d(new Paragraph::Private(this, emptyParagraphLayout))
1156 Paragraph::Paragraph(Paragraph const & par)
1157 : itemdepth(par.itemdepth),
1158 d(new Paragraph::Private(*par.d, this))
1164 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1165 : itemdepth(par.itemdepth),
1166 d(new Paragraph::Private(*par.d, this, beg, end))
1172 Paragraph & Paragraph::operator=(Paragraph const & par)
1174 // needed as we will destroy the private part before copying it
1176 itemdepth = par.itemdepth;
1180 d = new Private(*par.d, this);
1187 Paragraph::~Paragraph()
1196 // this shall be called just before every "os << ..." action.
1197 void flushString(ostream & os, docstring & s)
1206 void Paragraph::write(ostream & os, BufferParams const & bparams,
1207 depth_type & dth) const
1209 // The beginning or end of a deeper (i.e. nested) area?
1210 if (dth != d->params_.depth()) {
1211 if (d->params_.depth() > dth) {
1212 while (d->params_.depth() > dth) {
1213 os << "\n\\begin_deeper";
1217 while (d->params_.depth() < dth) {
1218 os << "\n\\end_deeper";
1224 // First write the layout
1225 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1227 d->params_.write(os);
1229 Font font1(inherit_font, bparams.language);
1231 Change running_change = Change(Change::UNCHANGED);
1233 // this string is used as a buffer to avoid repetitive calls
1234 // to to_utf8(), which turn out to be expensive (JMarc)
1235 docstring write_buffer;
1238 for (pos_type i = 0; i <= size(); ++i) {
1240 Change const change = lookupChange(i);
1241 if (change != running_change)
1242 flushString(os, write_buffer);
1243 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1244 running_change = change;
1249 // Write font changes (ignore spelling markers)
1250 Font font2 = getFontSettings(bparams, i);
1251 font2.setMisspelled(false);
1252 if (font2 != font1) {
1253 flushString(os, write_buffer);
1254 font2.lyxWriteChanges(font1, os);
1259 char_type const c = d->text_[i];
1262 if (Inset const * inset = getInset(i)) {
1263 flushString(os, write_buffer);
1264 if (inset->directWrite()) {
1265 // international char, let it write
1266 // code directly so it's shorter in
1272 os << "\\begin_inset ";
1274 os << "\n\\end_inset\n\n";
1280 flushString(os, write_buffer);
1281 os << "\n\\backslash\n";
1285 flushString(os, write_buffer);
1286 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1293 if ((column > 70 && c == ' ')
1295 flushString(os, write_buffer);
1299 // this check is to amend a bug. LyX sometimes
1300 // inserts '\0' this could cause problems.
1302 write_buffer.push_back(c);
1304 LYXERR0("NUL char in structure.");
1310 flushString(os, write_buffer);
1311 os << "\n\\end_layout\n";
1315 void Paragraph::validate(LaTeXFeatures & features) const
1317 d->validate(features);
1321 void Paragraph::insert(pos_type start, docstring const & str,
1322 Font const & font, Change const & change)
1324 for (size_t i = 0, n = str.size(); i != n ; ++i)
1325 insertChar(start + i, str[i], font, change);
1329 void Paragraph::appendChar(char_type c, Font const & font,
1330 Change const & change)
1333 d->changes_.insert(change, d->text_.size());
1334 // when appending characters, no need to update tables
1335 d->text_.push_back(c);
1336 setFont(d->text_.size() - 1, font);
1340 void Paragraph::appendString(docstring const & s, Font const & font,
1341 Change const & change)
1343 pos_type end = s.size();
1344 size_t oldsize = d->text_.size();
1345 size_t newsize = oldsize + end;
1346 size_t capacity = d->text_.capacity();
1347 if (newsize >= capacity)
1348 d->text_.reserve(max(capacity + 100, newsize));
1350 // when appending characters, no need to update tables
1353 // FIXME: Optimize this!
1354 for (size_t i = oldsize; i != newsize; ++i) {
1356 d->changes_.insert(change, i);
1358 d->fontlist_.set(oldsize, font);
1359 d->fontlist_.set(newsize - 1, font);
1363 void Paragraph::insertChar(pos_type pos, char_type c,
1366 d->insertChar(pos, c, Change(trackChanges ?
1367 Change::INSERTED : Change::UNCHANGED));
1371 void Paragraph::insertChar(pos_type pos, char_type c,
1372 Font const & font, bool trackChanges)
1374 d->insertChar(pos, c, Change(trackChanges ?
1375 Change::INSERTED : Change::UNCHANGED));
1380 void Paragraph::insertChar(pos_type pos, char_type c,
1381 Font const & font, Change const & change)
1383 d->insertChar(pos, c, change);
1388 bool Paragraph::insertInset(pos_type pos, Inset * inset,
1389 Font const & font, Change const & change)
1391 bool const success = insertInset(pos, inset, change);
1392 // Set the font/language of the inset...
1398 void Paragraph::resetFonts(Font const & font)
1400 d->fontlist_.clear();
1401 d->fontlist_.set(0, font);
1402 d->fontlist_.set(d->text_.size() - 1, font);
1405 // Gets uninstantiated font setting at position.
1406 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1410 LYXERR0("pos: " << pos << " size: " << size());
1411 LASSERT(pos <= size(), /**/);
1414 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1415 if (cit != d->fontlist_.end())
1418 if (pos == size() && !empty())
1419 return getFontSettings(bparams, pos - 1);
1421 // Optimisation: avoid a full font instantiation if there is no
1422 // language change from previous call.
1423 static Font previous_font;
1424 static Language const * previous_lang = 0;
1425 Language const * lang = getParLanguage(bparams);
1426 if (lang != previous_lang) {
1427 previous_lang = lang;
1428 previous_font = Font(inherit_font, lang);
1430 return previous_font;
1434 FontSpan Paragraph::fontSpan(pos_type pos) const
1436 LASSERT(pos <= size(), /**/);
1439 FontList::const_iterator cit = d->fontlist_.begin();
1440 FontList::const_iterator end = d->fontlist_.end();
1441 for (; cit != end; ++cit) {
1442 if (cit->pos() >= pos) {
1443 if (pos >= beginOfBody())
1444 return FontSpan(max(start, beginOfBody()),
1447 return FontSpan(start,
1448 min(beginOfBody() - 1,
1451 start = cit->pos() + 1;
1454 // This should not happen, but if so, we take no chances.
1455 // LYXERR0("Paragraph::getEndPosOfFontSpan: This should not happen!");
1456 return FontSpan(pos, pos);
1460 // Gets uninstantiated font setting at position 0
1461 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1463 if (!empty() && !d->fontlist_.empty())
1464 return d->fontlist_.begin()->font();
1466 // Optimisation: avoid a full font instantiation if there is no
1467 // language change from previous call.
1468 static Font previous_font;
1469 static Language const * previous_lang = 0;
1470 if (bparams.language != previous_lang) {
1471 previous_lang = bparams.language;
1472 previous_font = Font(inherit_font, bparams.language);
1475 return previous_font;
1479 // Gets the fully instantiated font at a given position in a paragraph
1480 // This is basically the same function as Text::GetFont() in text2.cpp.
1481 // The difference is that this one is used for generating the LaTeX file,
1482 // and thus cosmetic "improvements" are disallowed: This has to deliver
1483 // the true picture of the buffer. (Asger)
1484 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1485 Font const & outerfont) const
1487 LASSERT(pos >= 0, /**/);
1489 Font font = getFontSettings(bparams, pos);
1491 pos_type const body_pos = beginOfBody();
1492 FontInfo & fi = font.fontInfo();
1494 fi.realize(d->layout_->labelfont);
1496 fi.realize(d->layout_->font);
1498 fi.realize(outerfont.fontInfo());
1499 fi.realize(bparams.getFont().fontInfo());
1505 Font const Paragraph::getLabelFont
1506 (BufferParams const & bparams, Font const & outerfont) const
1508 FontInfo tmpfont = d->layout_->labelfont;
1509 tmpfont.realize(outerfont.fontInfo());
1510 tmpfont.realize(bparams.getFont().fontInfo());
1511 return Font(tmpfont, getParLanguage(bparams));
1515 Font const Paragraph::getLayoutFont
1516 (BufferParams const & bparams, Font const & outerfont) const
1518 FontInfo tmpfont = d->layout_->font;
1519 tmpfont.realize(outerfont.fontInfo());
1520 tmpfont.realize(bparams.getFont().fontInfo());
1521 return Font(tmpfont, getParLanguage(bparams));
1525 /// Returns the height of the highest font in range
1526 FontSize Paragraph::highestFontInRange
1527 (pos_type startpos, pos_type endpos, FontSize def_size) const
1529 return d->fontlist_.highestInRange(startpos, endpos, def_size);
1533 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1535 char_type c = d->text_[pos];
1536 if (!lyxrc.rtl_support)
1566 if (uc != c && getFontSettings(bparams, pos).isRightToLeft())
1572 void Paragraph::setFont(pos_type pos, Font const & font)
1574 LASSERT(pos <= size(), /**/);
1576 // First, reduce font against layout/label font
1577 // Update: The setCharFont() routine in text2.cpp already
1578 // reduces font, so we don't need to do that here. (Asger)
1580 d->fontlist_.set(pos, font);
1584 void Paragraph::makeSameLayout(Paragraph const & par)
1586 d->layout_ = par.d->layout_;
1587 d->params_ = par.d->params_;
1591 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1593 if (isFreeSpacing())
1599 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1600 if (eraseChar(pos, trackChanges))
1606 return count > 0 || pos > 0;
1610 bool Paragraph::hasSameLayout(Paragraph const & par) const
1612 return par.d->layout_ == d->layout_
1613 && d->params_.sameLayout(par.d->params_);
1617 depth_type Paragraph::getDepth() const
1619 return d->params_.depth();
1623 depth_type Paragraph::getMaxDepthAfter() const
1625 if (d->layout_->isEnvironment())
1626 return d->params_.depth() + 1;
1628 return d->params_.depth();
1632 char Paragraph::getAlign() const
1634 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1635 return d->layout_->align;
1637 return d->params_.align();
1641 docstring const & Paragraph::labelString() const
1643 return d->params_.labelString();
1647 // the next two functions are for the manual labels
1648 docstring const Paragraph::getLabelWidthString() const
1650 if (d->layout_->margintype == MARGIN_MANUAL
1651 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
1652 return d->params_.labelWidthString();
1654 return _("Senseless with this layout!");
1658 void Paragraph::setLabelWidthString(docstring const & s)
1660 d->params_.labelWidthString(s);
1664 docstring Paragraph::expandLabel(Layout const & layout,
1665 BufferParams const & bparams) const
1667 return expandParagraphLabel(layout, bparams, true);
1671 docstring Paragraph::expandDocBookLabel(Layout const & layout,
1672 BufferParams const & bparams) const
1674 return expandParagraphLabel(layout, bparams, false);
1678 docstring Paragraph::expandParagraphLabel(Layout const & layout,
1679 BufferParams const & bparams, bool process_appendix) const
1681 DocumentClass const & tclass = bparams.documentClass();
1682 string const & lang = getParLanguage(bparams)->code();
1683 bool const in_appendix = process_appendix && d->params_.appendix();
1684 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
1686 if (fmt.empty() && layout.labeltype == LABEL_COUNTER
1687 && !layout.counter.empty())
1688 return tclass.counters().theCounter(layout.counter, lang);
1690 // handle 'inherited level parts' in 'fmt',
1691 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
1692 size_t const i = fmt.find('@', 0);
1693 if (i != docstring::npos) {
1694 size_t const j = fmt.find('@', i + 1);
1695 if (j != docstring::npos) {
1696 docstring parent(fmt, i + 1, j - i - 1);
1697 docstring label = from_ascii("??");
1698 if (tclass.hasLayout(parent))
1699 docstring label = expandParagraphLabel(tclass[parent], bparams,
1701 fmt = docstring(fmt, 0, i) + label
1702 + docstring(fmt, j + 1, docstring::npos);
1706 return tclass.counters().counterLabel(fmt, lang);
1710 void Paragraph::applyLayout(Layout const & new_layout)
1712 d->layout_ = &new_layout;
1713 LyXAlignment const oldAlign = d->params_.align();
1715 if (!(oldAlign & d->layout_->alignpossible)) {
1716 frontend::Alert::warning(_("Alignment not permitted"),
1717 _("The new layout does not permit the alignment previously used.\nSetting to default."));
1718 d->params_.align(LYX_ALIGN_LAYOUT);
1723 pos_type Paragraph::beginOfBody() const
1725 return d->begin_of_body_;
1729 void Paragraph::setBeginOfBody()
1731 if (d->layout_->labeltype != LABEL_MANUAL) {
1732 d->begin_of_body_ = 0;
1736 // Unroll the first two cycles of the loop
1737 // and remember the previous character to
1738 // remove unnecessary getChar() calls
1740 pos_type end = size();
1741 if (i < end && !isNewline(i)) {
1743 char_type previous_char = 0;
1746 previous_char = d->text_[i];
1747 if (!isNewline(i)) {
1749 while (i < end && previous_char != ' ') {
1754 previous_char = temp;
1760 d->begin_of_body_ = i;
1764 bool Paragraph::allowParagraphCustomization() const
1766 return inInset().allowParagraphCustomization();
1770 bool Paragraph::usePlainLayout() const
1772 return inInset().usePlainLayout();
1778 // paragraphs inside floats need different alignment tags to avoid
1781 bool noTrivlistCentering(InsetCode code)
1783 return code == FLOAT_CODE
1784 || code == WRAP_CODE
1785 || code == CELL_CODE;
1789 string correction(string const & orig)
1791 if (orig == "flushleft")
1792 return "raggedright";
1793 if (orig == "flushright")
1794 return "raggedleft";
1795 if (orig == "center")
1801 string const corrected_env(string const & suffix, string const & env,
1802 InsetCode code, bool const lastpar)
1804 string output = suffix + "{";
1805 if (noTrivlistCentering(code)) {
1807 // the last paragraph in non-trivlist-aligned
1808 // context is special (to avoid unwanted whitespace)
1809 if (suffix == "\\begin")
1810 return "\\" + correction(env) + "{}";
1813 output += correction(env);
1817 if (suffix == "\\begin")
1823 void adjust_row_column(string const & str, TexRow & texrow, int & column)
1825 if (!contains(str, "\n"))
1826 column += str.size();
1830 column = rsplit(str, tmp, '\n').size();
1837 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
1838 odocstream & os, TexRow & texrow,
1839 OutputParams const & runparams) const
1843 if (params_.noindent()) {
1844 os << "\\noindent ";
1848 LyXAlignment const curAlign = params_.align();
1850 if (curAlign == layout_->align)
1854 case LYX_ALIGN_NONE:
1855 case LYX_ALIGN_BLOCK:
1856 case LYX_ALIGN_LAYOUT:
1857 case LYX_ALIGN_SPECIAL:
1858 case LYX_ALIGN_DECIMAL:
1860 case LYX_ALIGN_LEFT:
1861 case LYX_ALIGN_RIGHT:
1862 case LYX_ALIGN_CENTER:
1863 if (runparams.moving_arg) {
1870 string const begin_tag = "\\begin";
1871 InsetCode code = ownerCode();
1872 bool const lastpar = runparams.isLastPar;
1875 case LYX_ALIGN_NONE:
1876 case LYX_ALIGN_BLOCK:
1877 case LYX_ALIGN_LAYOUT:
1878 case LYX_ALIGN_SPECIAL:
1879 case LYX_ALIGN_DECIMAL:
1881 case LYX_ALIGN_LEFT: {
1883 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1884 output = corrected_env(begin_tag, "flushleft", code, lastpar);
1886 output = corrected_env(begin_tag, "flushright", code, lastpar);
1887 os << from_ascii(output);
1888 adjust_row_column(output, texrow, column);
1890 } case LYX_ALIGN_RIGHT: {
1892 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1893 output = corrected_env(begin_tag, "flushright", code, lastpar);
1895 output = corrected_env(begin_tag, "flushleft", code, lastpar);
1896 os << from_ascii(output);
1897 adjust_row_column(output, texrow, column);
1899 } case LYX_ALIGN_CENTER: {
1901 output = corrected_env(begin_tag, "center", code, lastpar);
1902 os << from_ascii(output);
1903 adjust_row_column(output, texrow, column);
1912 int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
1913 odocstream & os, TexRow & texrow,
1914 OutputParams const & runparams) const
1918 LyXAlignment const curAlign = params_.align();
1920 if (curAlign == layout_->align)
1924 case LYX_ALIGN_NONE:
1925 case LYX_ALIGN_BLOCK:
1926 case LYX_ALIGN_LAYOUT:
1927 case LYX_ALIGN_SPECIAL:
1928 case LYX_ALIGN_DECIMAL:
1930 case LYX_ALIGN_LEFT:
1931 case LYX_ALIGN_RIGHT:
1932 case LYX_ALIGN_CENTER:
1933 if (runparams.moving_arg) {
1940 string const end_tag = "\n\\par\\end";
1941 InsetCode code = ownerCode();
1942 bool const lastpar = runparams.isLastPar;
1945 case LYX_ALIGN_NONE:
1946 case LYX_ALIGN_BLOCK:
1947 case LYX_ALIGN_LAYOUT:
1948 case LYX_ALIGN_SPECIAL:
1949 case LYX_ALIGN_DECIMAL:
1951 case LYX_ALIGN_LEFT: {
1953 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1954 output = corrected_env(end_tag, "flushleft", code, lastpar);
1956 output = corrected_env(end_tag, "flushright", code, lastpar);
1957 os << from_ascii(output);
1958 adjust_row_column(output, texrow, column);
1960 } case LYX_ALIGN_RIGHT: {
1962 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1963 output = corrected_env(end_tag, "flushright", code, lastpar);
1965 output = corrected_env(end_tag, "flushleft", code, lastpar);
1966 os << from_ascii(output);
1967 adjust_row_column(output, texrow, column);
1969 } case LYX_ALIGN_CENTER: {
1971 output = corrected_env(end_tag, "center", code, lastpar);
1972 os << from_ascii(output);
1973 adjust_row_column(output, texrow, column);
1982 // This one spits out the text of the paragraph
1983 void Paragraph::latex(BufferParams const & bparams,
1984 Font const & outerfont,
1985 odocstream & os, TexRow & texrow,
1986 OutputParams const & runparams,
1987 int start_pos, int end_pos, bool force) const
1989 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
1991 // FIXME This check should not be needed. Perhaps issue an
1992 // error if it triggers.
1993 Layout const & style = inInset().forcePlainLayout() ?
1994 bparams.documentClass().plainLayout() : *d->layout_;
1996 if (!force && style.inpreamble)
1999 bool const allowcust = allowParagraphCustomization();
2001 // Current base font for all inherited font changes, without any
2002 // change caused by an individual character, except for the language:
2003 // It is set to the language of the first character.
2004 // As long as we are in the label, this font is the base font of the
2005 // label. Before the first body character it is set to the base font
2009 // Maybe we have to create a optional argument.
2010 pos_type body_pos = beginOfBody();
2011 unsigned int column = 0;
2014 // the optional argument is kept in curly brackets in
2015 // case it contains a ']'
2018 basefont = getLabelFont(bparams, outerfont);
2020 basefont = getLayoutFont(bparams, outerfont);
2023 // Which font is currently active?
2024 Font running_font(basefont);
2025 // Do we have an open font change?
2026 bool open_font = false;
2028 Change runningChange = Change(Change::UNCHANGED);
2030 Encoding const * const prev_encoding = runparams.encoding;
2032 texrow.start(id(), 0);
2034 // if the paragraph is empty, the loop will not be entered at all
2036 if (style.isCommand()) {
2041 column += d->startTeXParParams(bparams, os, texrow,
2045 for (pos_type i = 0; i < size(); ++i) {
2046 // First char in paragraph or after label?
2047 if (i == body_pos) {
2050 column += running_font.latexWriteEndChanges(
2051 os, bparams, runparams,
2052 basefont, basefont);
2055 basefont = getLayoutFont(bparams, outerfont);
2056 running_font = basefont;
2058 column += Changes::latexMarkChange(os, bparams,
2059 runningChange, Change(Change::UNCHANGED),
2061 runningChange = Change(Change::UNCHANGED);
2066 if (style.isCommand()) {
2072 column += d->startTeXParParams(bparams, os,
2077 Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
2080 if (bparams.outputChanges && runningChange != change) {
2082 column += running_font.latexWriteEndChanges(
2083 os, bparams, runparams, basefont, basefont);
2086 basefont = getLayoutFont(bparams, outerfont);
2087 running_font = basefont;
2089 column += Changes::latexMarkChange(os, bparams, runningChange,
2091 runningChange = change;
2094 // do not output text which is marked deleted
2095 // if change tracking output is disabled
2096 if (!bparams.outputChanges && change.deleted()) {
2102 // Fully instantiated font
2103 Font const font = getFont(bparams, i, outerfont);
2105 Font const last_font = running_font;
2107 // Do we need to close the previous font?
2109 (font != running_font ||
2110 font.language() != running_font.language()))
2112 column += running_font.latexWriteEndChanges(
2113 os, bparams, runparams, basefont,
2114 (i == body_pos-1) ? basefont : font);
2115 running_font = basefont;
2119 // close babel's font environment before opening CJK.
2120 if (!running_font.language()->babel().empty() &&
2121 font.language()->encoding()->package() == Encoding::CJK) {
2122 string end_tag = subst(lyxrc.language_command_end,
2124 running_font.language()->babel());
2125 os << from_ascii(end_tag);
2126 column += end_tag.length();
2129 // Switch file encoding if necessary (and allowed)
2130 if (!runparams.verbatim &&
2131 runparams.encoding->package() != Encoding::none &&
2132 font.language()->encoding()->package() != Encoding::none) {
2133 pair<bool, int> const enc_switch = switchEncoding(os, bparams,
2134 runparams, *(font.language()->encoding()));
2135 if (enc_switch.first) {
2136 column += enc_switch.second;
2137 runparams.encoding = font.language()->encoding();
2141 char_type const c = d->text_[i];
2143 // Do we need to change font?
2144 if ((font != running_font ||
2145 font.language() != running_font.language()) &&
2148 odocstringstream ods;
2149 column += font.latexWriteStartChanges(ods, bparams,
2150 runparams, basefont,
2152 running_font = font;
2154 docstring fontchange = ods.str();
2155 // check whether the fontchange ends with a \\textcolor
2156 // modifier and the text starts with a space (bug 4473)
2157 docstring const last_modifier = rsplit(fontchange, '\\');
2158 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2159 os << fontchange << from_ascii("{}");
2160 // check if the fontchange ends with a trailing blank
2161 // (like "\small " (see bug 3382)
2162 else if (suffixIs(fontchange, ' ') && c == ' ')
2163 os << fontchange.substr(0, fontchange.size() - 1)
2164 << from_ascii("{}");
2169 // FIXME: think about end_pos implementation...
2170 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2171 // FIXME: integrate this case in latexSpecialChar
2172 // Do not print the separation of the optional argument
2173 // if style.pass_thru is false. This works because
2174 // latexSpecialChar ignores spaces if
2175 // style.pass_thru is false.
2176 if (i != body_pos - 1) {
2177 if (d->simpleTeXBlanks(
2178 runparams, os, texrow,
2179 i, column, font, style)) {
2180 // A surrogate pair was output. We
2181 // must not call latexSpecialChar
2182 // in this iteration, since it would output
2183 // the combining character again.
2190 OutputParams rp = runparams;
2191 rp.free_spacing = style.free_spacing;
2192 rp.local_font = &font;
2193 rp.intitle = style.intitle;
2195 // Two major modes: LaTeX or plain
2196 // Handle here those cases common to both modes
2197 // and then split to handle the two modes separately.
2198 if (c == META_INSET) {
2199 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2200 d->latexInset(bparams, os,
2201 texrow, rp, running_font,
2202 basefont, outerfont, open_font,
2203 runningChange, style, i, column);
2206 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2208 d->latexSpecialChar(os, rp, running_font, runningChange,
2210 } catch (EncodingException & e) {
2211 if (runparams.dryrun) {
2212 os << "<" << _("LyX Warning: ")
2213 << _("uncodable character") << " '";
2217 // add location information and throw again.
2226 // Set the encoding to that returned from latexSpecialChar (see
2227 // comment for encoding member in OutputParams.h)
2228 runparams.encoding = rp.encoding;
2231 // If we have an open font definition, we have to close it
2233 #ifdef FIXED_LANGUAGE_END_DETECTION
2235 running_font.latexWriteEndChanges(os, bparams,
2236 runparams, basefont,
2237 next_->getFont(bparams, 0, outerfont));
2239 running_font.latexWriteEndChanges(os, bparams,
2240 runparams, basefont, basefont);
2243 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2244 //FIXME: there as we start another \selectlanguage with the next paragraph if
2245 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2246 running_font.latexWriteEndChanges(os, bparams, runparams,
2247 basefont, basefont);
2251 column += Changes::latexMarkChange(os, bparams, runningChange,
2252 Change(Change::UNCHANGED), runparams);
2254 // Needed if there is an optional argument but no contents.
2255 if (body_pos > 0 && body_pos == size()) {
2259 if (allowcust && d->endTeXParParams(bparams, os, texrow, runparams)
2260 && runparams.encoding != prev_encoding) {
2261 runparams.encoding = prev_encoding;
2262 if (!bparams.useXetex)
2263 os << setEncoding(prev_encoding->iconvName());
2266 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2270 bool Paragraph::emptyTag() const
2272 for (pos_type i = 0; i < size(); ++i) {
2273 if (Inset const * inset = getInset(i)) {
2274 InsetCode lyx_code = inset->lyxCode();
2275 // FIXME testing like that is wrong. What is
2277 if (lyx_code != TOC_CODE &&
2278 lyx_code != INCLUDE_CODE &&
2279 lyx_code != GRAPHICS_CODE &&
2280 lyx_code != ERT_CODE &&
2281 lyx_code != LISTINGS_CODE &&
2282 lyx_code != FLOAT_CODE &&
2283 lyx_code != TABULAR_CODE) {
2287 char_type c = d->text_[i];
2288 if (c != ' ' && c != '\t')
2296 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2299 for (pos_type i = 0; i < size(); ++i) {
2300 if (Inset const * inset = getInset(i)) {
2301 InsetCode lyx_code = inset->lyxCode();
2302 if (lyx_code == LABEL_CODE) {
2303 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2304 docstring const & id = il->getParam("name");
2305 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2313 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2317 for (i = 0; i < size(); ++i) {
2318 if (Inset const * inset = getInset(i)) {
2319 inset->docbook(os, runparams);
2321 char_type c = d->text_[i];
2324 os << sgml::escapeChar(c);
2331 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2335 for (i = 0; i < size(); ++i) {
2336 if (Inset const * inset = getInset(i)) {
2337 inset->xhtml(xs, runparams);
2339 char_type c = d->text_[i];
2349 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2352 pos_type size = text_.size();
2353 for (pos_type i = initial; i < size; ++i) {
2354 Font font = owner_->getFont(buf.params(), i, outerfont);
2355 if (text_[i] == META_INSET)
2357 if (i != initial && font != font_old)
2366 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2368 OutputParams const & runparams,
2369 Font const & outerfont,
2370 pos_type initial) const
2372 bool emph_flag = false;
2374 Layout const & style = *d->layout_;
2376 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2378 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2381 // parsing main loop
2382 for (pos_type i = initial; i < size(); ++i) {
2383 Font font = getFont(buf.params(), i, outerfont);
2385 // handle <emphasis> tag
2386 if (font_old.emph() != font.fontInfo().emph()) {
2387 if (font.fontInfo().emph() == FONT_ON) {
2390 } else if (i != initial) {
2391 os << "</emphasis>";
2396 if (Inset const * inset = getInset(i)) {
2397 inset->docbook(os, runparams);
2399 char_type c = d->text_[i];
2401 if (style.pass_thru)
2404 os << sgml::escapeChar(c);
2406 font_old = font.fontInfo();
2410 os << "</emphasis>";
2413 if (style.free_spacing)
2415 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2420 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2422 OutputParams const & runparams,
2423 Font const & outerfont,
2424 pos_type initial) const
2428 bool emph_flag = false;
2429 bool bold_flag = false;
2432 Layout const & style = *d->layout_;
2434 if (!runparams.for_toc && runparams.html_make_pars) {
2435 // generate a magic label for this paragraph
2436 string const attr = "id='" + magicLabel() + "'";
2437 xs << html::CompTag("a", attr);
2441 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2443 // parsing main loop
2444 for (pos_type i = initial; i < size(); ++i) {
2445 // let's not show deleted material in the output
2449 Font font = getFont(buf.params(), i, outerfont);
2452 if (font_old.emph() != font.fontInfo().emph()) {
2453 if (font.fontInfo().emph() == FONT_ON) {
2454 xs << html::StartTag("em");
2456 } else if (emph_flag && i != initial) {
2457 xs << html::EndTag("em");
2462 if (font_old.series() != font.fontInfo().series()) {
2463 if (font.fontInfo().series() == BOLD_SERIES) {
2464 xs << html::StartTag("strong");
2466 } else if (bold_flag && i != initial) {
2467 xs << html::EndTag("strong");
2472 // Other such tags? What about the other text ranges?
2474 Inset const * inset = getInset(i);
2476 InsetCommand const * ic = inset->asInsetCommand();
2477 InsetLayout const & il = inset->getLayout();
2478 InsetMath const * im = inset->asInsetMath();
2479 if (!runparams.for_toc
2480 || im || il.isInToc() || (ic && ic->isInToc())) {
2481 OutputParams np = runparams;
2482 if (!il.htmlisblock())
2483 np.html_in_par = true;
2484 retval += inset->xhtml(xs, np);
2487 char_type c = d->text_[i];
2489 if (style.pass_thru)
2491 else if (c == '-') {
2494 if (j < size() && d->text_[j] == '-') {
2496 if (j < size() && d->text_[j] == '-') {
2497 str += from_ascii("—");
2500 str += from_ascii("–");
2506 // We don't want to escape the entities. Note that
2507 // it is safe to do this, since str can otherwise
2508 // only be "-". E.g., it can't be "<".
2509 xs << XHTMLStream::NextRaw() << str;
2513 font_old = font.fontInfo();
2521 bool Paragraph::isHfill(pos_type pos) const
2523 Inset const * inset = getInset(pos);
2524 return inset && (inset->lyxCode() == SPACE_CODE &&
2525 inset->isStretchableSpace());
2529 bool Paragraph::isNewline(pos_type pos) const
2531 Inset const * inset = getInset(pos);
2532 return inset && inset->lyxCode() == NEWLINE_CODE;
2536 bool Paragraph::isLineSeparator(pos_type pos) const
2538 char_type const c = d->text_[pos];
2539 if (isLineSeparatorChar(c))
2541 Inset const * inset = getInset(pos);
2542 return inset && inset->isLineSeparator();
2546 bool Paragraph::isWordSeparator(pos_type pos) const
2548 if (Inset const * inset = getInset(pos))
2549 return !inset->isLetter();
2550 char_type const c = d->text_[pos];
2551 // We want to pass the ' and escape chars to the spellchecker
2552 static docstring const quote = from_utf8(lyxrc.spellchecker_esc_chars + '\'');
2553 return (!isLetterChar(c) && !isDigit(c) && !contains(quote, c))
2558 bool Paragraph::isChar(pos_type pos) const
2560 if (Inset const * inset = getInset(pos))
2561 return inset->isChar();
2562 char_type const c = d->text_[pos];
2563 return !isLetterChar(c) && !isDigit(c) && !lyx::isSpace(c);
2567 bool Paragraph::isSpace(pos_type pos) const
2569 if (Inset const * inset = getInset(pos))
2570 return inset->isSpace();
2571 char_type const c = d->text_[pos];
2572 return lyx::isSpace(c);
2577 Paragraph::getParLanguage(BufferParams const & bparams) const
2580 return getFirstFontSettings(bparams).language();
2581 // FIXME: we should check the prev par as well (Lgb)
2582 return bparams.language;
2586 bool Paragraph::isRTL(BufferParams const & bparams) const
2588 return lyxrc.rtl_support
2589 && getParLanguage(bparams)->rightToLeft()
2590 && !inInset().getLayout().forceLTR();
2594 void Paragraph::changeLanguage(BufferParams const & bparams,
2595 Language const * from, Language const * to)
2597 // change language including dummy font change at the end
2598 for (pos_type i = 0; i <= size(); ++i) {
2599 Font font = getFontSettings(bparams, i);
2600 if (font.language() == from) {
2601 font.setLanguage(to);
2608 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
2610 Language const * doc_language = bparams.language;
2611 FontList::const_iterator cit = d->fontlist_.begin();
2612 FontList::const_iterator end = d->fontlist_.end();
2614 for (; cit != end; ++cit)
2615 if (cit->font().language() != ignore_language &&
2616 cit->font().language() != latex_language &&
2617 cit->font().language() != doc_language)
2623 void Paragraph::getLanguages(std::set<Language const *> & languages) const
2625 FontList::const_iterator cit = d->fontlist_.begin();
2626 FontList::const_iterator end = d->fontlist_.end();
2628 for (; cit != end; ++cit) {
2629 Language const * lang = cit->font().language();
2630 if (lang != ignore_language &&
2631 lang != latex_language)
2632 languages.insert(lang);
2637 docstring Paragraph::asString(int options) const
2639 return asString(0, size(), options);
2643 docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
2645 odocstringstream os;
2648 && options & AS_STR_LABEL
2649 && !d->params_.labelString().empty())
2650 os << d->params_.labelString() << ' ';
2652 for (pos_type i = beg; i < end; ++i) {
2653 char_type const c = d->text_[i];
2654 if (isPrintable(c) || c == '\t'
2655 || (c == '\n' && (options & AS_STR_NEWLINES)))
2657 else if (c == META_INSET && (options & AS_STR_INSETS)) {
2658 getInset(i)->tocString(os);
2659 if (getInset(i)->asInsetMath())
2668 docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputParams & runparams) const
2670 odocstringstream os;
2673 && options & AS_STR_LABEL
2674 && !d->params_.labelString().empty())
2675 os << d->params_.labelString() << ' ';
2677 for (pos_type i = beg; i < end; ++i) {
2678 char_type const c = d->text_[i];
2679 if (isPrintable(c) || c == '\t'
2680 || (c == '\n' && (options & AS_STR_NEWLINES)))
2682 else if (c == META_INSET && (options & AS_STR_INSETS)) {
2683 getInset(i)->plaintext(os, runparams);
2691 void Paragraph::setInsetOwner(Inset const * inset)
2693 d->inset_owner_ = inset;
2697 int Paragraph::id() const
2703 void Paragraph::setId(int id)
2709 Layout const & Paragraph::layout() const
2715 void Paragraph::setLayout(Layout const & layout)
2717 d->layout_ = &layout;
2721 void Paragraph::setDefaultLayout(DocumentClass const & tc)
2723 setLayout(tc.defaultLayout());
2727 void Paragraph::setPlainLayout(DocumentClass const & tc)
2729 setLayout(tc.plainLayout());
2733 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
2735 if (usePlainLayout())
2736 setPlainLayout(tclass);
2738 setDefaultLayout(tclass);
2742 Inset const & Paragraph::inInset() const
2744 LASSERT(d->inset_owner_, throw ExceptionMessage(BufferException,
2745 _("Memory problem"), _("Paragraph not properly initialized")));
2746 return *d->inset_owner_;
2750 ParagraphParameters & Paragraph::params()
2756 ParagraphParameters const & Paragraph::params() const
2762 bool Paragraph::isFreeSpacing() const
2764 if (d->layout_->free_spacing)
2766 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
2770 bool Paragraph::allowEmpty() const
2772 if (d->layout_->keepempty)
2774 return d->inset_owner_ && d->inset_owner_->allowEmpty();
2778 char_type Paragraph::transformChar(char_type c, pos_type pos) const
2780 if (!Encodings::isArabicChar(c))
2783 char_type prev_char = ' ';
2784 char_type next_char = ' ';
2786 for (pos_type i = pos - 1; i >= 0; --i) {
2787 char_type const par_char = d->text_[i];
2788 if (!Encodings::isArabicComposeChar(par_char)) {
2789 prev_char = par_char;
2794 for (pos_type i = pos + 1, end = size(); i < end; ++i) {
2795 char_type const par_char = d->text_[i];
2796 if (!Encodings::isArabicComposeChar(par_char)) {
2797 next_char = par_char;
2802 if (Encodings::isArabicChar(next_char)) {
2803 if (Encodings::isArabicChar(prev_char) &&
2804 !Encodings::isArabicSpecialChar(prev_char))
2805 return Encodings::transformChar(c, Encodings::FORM_MEDIAL);
2807 return Encodings::transformChar(c, Encodings::FORM_INITIAL);
2809 if (Encodings::isArabicChar(prev_char) &&
2810 !Encodings::isArabicSpecialChar(prev_char))
2811 return Encodings::transformChar(c, Encodings::FORM_FINAL);
2813 return Encodings::transformChar(c, Encodings::FORM_ISOLATED);
2818 int Paragraph::checkBiblio(Buffer const & buffer)
2821 // This is getting more and more a mess. ...We really should clean
2822 // up this bibitem issue for 1.6. See also bug 2743.
2824 // Add bibitem insets if necessary
2825 if (d->layout_->labeltype != LABEL_BIBLIO)
2828 bool hasbibitem = !d->insetlist_.empty()
2829 // Insist on it being in pos 0
2830 && d->text_[0] == META_INSET
2831 && d->insetlist_.begin()->inset->lyxCode() == BIBITEM_CODE;
2833 bool track_changes = buffer.params().trackChanges;
2838 // remove a bibitem in pos != 0
2839 // restore it later in pos 0 if necessary
2840 // (e.g. if a user inserts contents _before_ the item)
2841 // we're assuming there's only one of these, which there
2843 int erasedInsetPosition = -1;
2844 InsetList::iterator it = d->insetlist_.begin();
2845 InsetList::iterator end = d->insetlist_.end();
2846 for (; it != end; ++it)
2847 if (it->inset->lyxCode() == BIBITEM_CODE
2849 InsetBibitem * olditem = static_cast<InsetBibitem *>(it->inset);
2850 oldkey = olditem->getParam("key");
2851 oldlabel = olditem->getParam("label");
2852 erasedInsetPosition = it->pos;
2853 eraseChar(erasedInsetPosition, track_changes);
2857 // There was an InsetBibitem at the beginning, and we didn't
2858 // have to erase one.
2859 if (hasbibitem && erasedInsetPosition < 0)
2862 // There was an InsetBibitem at the beginning and we did have to
2863 // erase one. So we give its properties to the beginning inset.
2865 InsetBibitem * inset =
2866 static_cast<InsetBibitem *>(d->insetlist_.begin()->inset);
2867 if (!oldkey.empty())
2868 inset->setParam("key", oldkey);
2869 inset->setParam("label", oldlabel);
2870 return -erasedInsetPosition;
2873 // There was no inset at the beginning, so we need to create one with
2874 // the key and label of the one we erased.
2875 InsetBibitem * inset =
2876 new InsetBibitem(const_cast<Buffer *>(&buffer), InsetCommandParams(BIBITEM_CODE));
2877 // restore values of previously deleted item in this par.
2878 if (!oldkey.empty())
2879 inset->setParam("key", oldkey);
2880 inset->setParam("label", oldlabel);
2881 insertInset(0, static_cast<Inset *>(inset),
2882 Change(track_changes ? Change::INSERTED : Change::UNCHANGED));
2888 void Paragraph::checkAuthors(AuthorList const & authorList)
2890 d->changes_.checkAuthors(authorList);
2894 bool Paragraph::isChanged(pos_type pos) const
2896 return lookupChange(pos).changed();
2900 bool Paragraph::isInserted(pos_type pos) const
2902 return lookupChange(pos).inserted();
2906 bool Paragraph::isDeleted(pos_type pos) const
2908 return lookupChange(pos).deleted();
2912 InsetList const & Paragraph::insetList() const
2914 return d->insetlist_;
2918 void Paragraph::setBuffer(Buffer & b)
2920 d->insetlist_.setBuffer(b);
2924 Inset * Paragraph::releaseInset(pos_type pos)
2926 Inset * inset = d->insetlist_.release(pos);
2927 /// does not honour change tracking!
2928 eraseChar(pos, false);
2933 Inset * Paragraph::getInset(pos_type pos)
2935 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
2936 ? d->insetlist_.get(pos) : 0;
2940 Inset const * Paragraph::getInset(pos_type pos) const
2942 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
2943 ? d->insetlist_.get(pos) : 0;
2947 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
2948 pos_type & right, TextCase action)
2950 // process sequences of modified characters; in change
2951 // tracking mode, this approach results in much better
2952 // usability than changing case on a char-by-char basis
2955 bool const trackChanges = bparams.trackChanges;
2957 bool capitalize = true;
2959 for (; pos < right; ++pos) {
2960 char_type oldChar = d->text_[pos];
2961 char_type newChar = oldChar;
2963 // ignore insets and don't play with deleted text!
2964 if (oldChar != META_INSET && !isDeleted(pos)) {
2966 case text_lowercase:
2967 newChar = lowercase(oldChar);
2969 case text_capitalization:
2971 newChar = uppercase(oldChar);
2975 case text_uppercase:
2976 newChar = uppercase(oldChar);
2981 if (isWordSeparator(pos) || isDeleted(pos)) {
2982 // permit capitalization again
2986 if (oldChar != newChar) {
2988 if (pos != right - 1)
2990 // step behind the changing area
2994 int erasePos = pos - changes.size();
2995 for (size_t i = 0; i < changes.size(); i++) {
2996 insertChar(pos, changes[i],
2997 getFontSettings(bparams,
3000 if (!eraseChar(erasePos, trackChanges)) {
3003 ++right; // expand selection
3011 bool Paragraph::find(docstring const & str, bool cs, bool mw,
3012 pos_type pos, bool del) const
3014 int const strsize = str.length();
3016 pos_type const parsize = d->text_.size();
3017 for (i = 0; pos + i < parsize; ++i) {
3020 if (cs && str[i] != d->text_[pos + i])
3022 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos + i]))
3024 if (!del && isDeleted(pos + i))
3031 // if necessary, check whether string matches word
3033 if (pos > 0 && !isWordSeparator(pos - 1))
3035 if (pos + strsize < parsize
3036 && !isWordSeparator(pos + strsize))
3044 char_type Paragraph::getChar(pos_type pos) const
3046 return d->text_[pos];
3050 pos_type Paragraph::size() const
3052 return d->text_.size();
3056 bool Paragraph::empty() const
3058 return d->text_.empty();
3062 bool Paragraph::isInset(pos_type pos) const
3064 return d->text_[pos] == META_INSET;
3068 bool Paragraph::isSeparator(pos_type pos) const
3070 //FIXME: Are we sure this can be the only separator?
3071 return d->text_[pos] == ' ';
3075 void Paragraph::deregisterWords()
3077 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3078 Private::LangWordsMap::const_iterator ite = d->words_.end();
3079 for (; itl != ite; ++itl) {
3080 WordList * wl = theWordList(itl->first);
3081 Private::Words::const_iterator it = (itl->second).begin();
3082 Private::Words::const_iterator et = (itl->second).end();
3083 for (; it != et; ++it)
3090 void Paragraph::locateWord(pos_type & from, pos_type & to,
3091 word_location const loc) const
3094 case WHOLE_WORD_STRICT:
3095 if (from == 0 || from == size()
3096 || isWordSeparator(from)
3097 || isWordSeparator(from - 1)) {
3101 // no break here, we go to the next
3104 // If we are already at the beginning of a word, do nothing
3105 if (!from || isWordSeparator(from - 1))
3107 // no break here, we go to the next
3110 // always move the cursor to the beginning of previous word
3111 while (from && !isWordSeparator(from - 1))
3115 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3118 // no need to move the 'from' cursor
3122 while (to < size() && !isWordSeparator(to))
3127 void Paragraph::collectWords()
3129 // This is the value that needs to be exposed in the preferences
3130 // to resolve bug #6760.
3131 static int minlength = 6;
3132 pos_type n = size();
3133 for (pos_type pos = 0; pos < n; ++pos) {
3134 if (isWordSeparator(pos))
3136 pos_type from = pos;
3137 locateWord(from, pos, WHOLE_WORD);
3138 if (pos - from >= minlength) {
3139 docstring word = asString(from, pos, AS_STR_NONE);
3140 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
3141 if (cit == d->fontlist_.end())
3143 Language const * lang = cit->font().language();
3144 d->words_[*lang].insert(word);
3150 void Paragraph::registerWords()
3152 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3153 Private::LangWordsMap::const_iterator ite = d->words_.end();
3154 for (; itl != ite; ++itl) {
3155 WordList * wl = theWordList(itl->first);
3156 Private::Words::const_iterator it = (itl->second).begin();
3157 Private::Words::const_iterator et = (itl->second).end();
3158 for (; it != et; ++it)
3164 void Paragraph::updateWords()
3172 bool Paragraph::spellCheck(pos_type & from, pos_type & to, WordLangTuple & wl,
3173 docstring_list & suggestions, bool do_suggestion) const
3175 SpellChecker * speller = theSpellChecker();
3179 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
3182 locateWord(from, to, WHOLE_WORD);
3183 if (from == to || from >= pos_type(d->text_.size()))
3186 docstring word = asString(from, to, AS_STR_INSETS);
3187 // Ignore words with digits
3188 // FIXME: make this customizable
3189 // (note that hunspell ignores words with digits by default)
3190 bool const ignored = hasDigit(word);
3191 Language * lang = const_cast<Language *>(getFontSettings(
3192 d->inset_owner_->buffer().params(), from).language());
3193 if (lang == d->inset_owner_->buffer().params().language
3194 && !lyxrc.spellchecker_alt_lang.empty()) {
3196 string const lang_variety =
3197 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3198 lang->setCode(lang_code);
3199 lang->setVariety(lang_variety);
3201 wl = WordLangTuple(word, lang);
3202 SpellChecker::Result res = ignored ?
3203 SpellChecker::OK : speller->check(wl);
3205 // FIXME: the code below makes aspell abort if a word in an unknown
3206 // language is checked.
3207 // Just ignore any error that the spellchecker reports.
3208 // FIXME: we should through out an exception and catch it in the GUI to
3209 // display the error.
3210 if (!speller->error().empty())
3214 bool const misspelled = res != SpellChecker::OK
3215 && res != SpellChecker::IGNORED_WORD;
3217 if (lyxrc.spellcheck_continuously)
3218 d->fontlist_.setMisspelled(from, to, misspelled);
3220 if (misspelled && do_suggestion)
3221 speller->suggest(wl, suggestions);
3223 suggestions.clear();
3229 bool Paragraph::isMisspelled(pos_type pos) const
3231 pos_type from = pos;
3234 docstring_list suggestions;
3235 return spellCheck(from, to, wl, suggestions, false);
3239 string Paragraph::magicLabel() const
3242 ss << "magicparlabel-" << id();