3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Richard Heck (XHTML output)
9 * \author Jean-Marc Lasgouttes
10 * \author Angus Leeming
12 * \author André Pönitz
14 * \author Jürgen Vigna
16 * Full author contact details are available in file CREDITS.
21 #include "Paragraph.h"
23 #include "LayoutFile.h"
25 #include "BufferParams.h"
28 #include "BufferEncodings.h"
29 #include "InsetList.h"
31 #include "LaTeXFeatures.h"
37 #include "OutputParams.h"
38 #include "output_latex.h"
39 #include "output_xhtml.h"
40 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
43 #include "TextClass.h"
46 #include "WordLangTuple.h"
49 #include "frontends/alert.h"
51 #include "insets/InsetBibitem.h"
52 #include "insets/InsetLabel.h"
53 #include "insets/InsetSpecialChar.h"
55 #include "support/debug.h"
56 #include "support/docstring_list.h"
57 #include "support/ExceptionMessage.h"
58 #include "support/gettext.h"
59 #include "support/lassert.h"
60 #include "support/lstrings.h"
61 #include "support/textutils.h"
67 using namespace lyx::support;
73 /// Inset identifier (above 0x10ffff, for ucs-4)
74 char_type const META_INSET = 0x200001;
79 /////////////////////////////////////////////////////////////////////
83 /////////////////////////////////////////////////////////////////////
85 class SpellResultRange {
87 SpellResultRange(FontSpan range, SpellChecker::Result result)
88 : range_(range), result_(result)
91 FontSpan const & range() const { return range_; }
93 void range(FontSpan const & r) { range_ = r; }
95 SpellChecker::Result result() const { return result_; }
97 void result(SpellChecker::Result r) { result_ = r; }
99 bool inside(pos_type pos) const { return range_.inside(pos); }
101 bool covered(FontSpan const & r) const
103 // 1. first of new range inside current range or
104 // 2. last of new range inside current range or
105 // 3. first of current range inside new range or
106 // 4. last of current range inside new range
107 return range_.inside(r.first) || range_.inside(r.last) ||
108 r.inside(range_.first) || r.inside(range_.last);
111 void shift(pos_type pos, int offset)
113 if (range_.first > pos) {
114 range_.first += offset;
115 range_.last += offset;
116 } else if (range_.last >= pos) {
117 range_.last += offset;
122 SpellChecker::Result result_ ;
126 /////////////////////////////////////////////////////////////////////
130 /////////////////////////////////////////////////////////////////////
132 class SpellCheckerState {
136 needs_refresh_ = true;
137 current_change_number_ = 0;
140 void setRange(FontSpan const & fp, SpellChecker::Result state)
143 RangesIterator et = ranges_.end();
144 RangesIterator it = ranges_.begin();
145 for (; it != et; ++it) {
146 if (!it->covered(fp))
147 result.push_back(SpellResultRange(it->range(), it->result()));
148 else if (state == SpellChecker::WORD_OK) {
149 // trim or split the current misspelled range
150 // store misspelled ranges only
151 FontSpan range = it->range();
152 if (fp.first > range.first) {
153 // misspelled area in front of WORD_OK
154 range.last = fp.first - 1;
155 result.push_back(SpellResultRange(range, it->result()));
158 if (fp.last < range.last) {
159 // misspelled area after WORD_OK range
160 range.first = fp.last + 1;
161 result.push_back(SpellResultRange(range, it->result()));
166 if (state != SpellChecker::WORD_OK)
167 ranges_.push_back(SpellResultRange(fp, state));
170 void increasePosAfterPos(pos_type pos)
172 correctRangesAfterPos(pos, 1);
176 void decreasePosAfterPos(pos_type pos)
178 correctRangesAfterPos(pos, -1);
182 void refreshLast(pos_type pos)
184 if (pos < refresh_.last)
188 SpellChecker::Result getState(pos_type pos) const
190 SpellChecker::Result result = SpellChecker::WORD_OK;
191 RangesIterator et = ranges_.end();
192 RangesIterator it = ranges_.begin();
193 for (; it != et; ++it) {
194 if(it->inside(pos)) {
201 FontSpan const & getRange(pos_type pos) const
203 /// empty span to indicate mismatch
204 static FontSpan empty_;
205 RangesIterator et = ranges_.end();
206 RangesIterator it = ranges_.begin();
207 for (; it != et; ++it) {
208 if(it->inside(pos)) {
215 bool needsRefresh() const
217 return needs_refresh_;
220 SpellChecker::ChangeNumber currentChangeNumber() const
222 return current_change_number_;
225 void refreshRange(pos_type & first, pos_type & last) const
227 first = refresh_.first;
228 last = refresh_.last;
231 void needsRefresh(pos_type pos)
233 if (needs_refresh_ && pos != -1) {
234 if (pos < refresh_.first)
235 refresh_.first = pos;
236 if (pos > refresh_.last)
238 } else if (pos != -1) {
239 // init request check for neighbour positions too
240 refresh_.first = pos > 0 ? pos - 1 : 0;
241 // no need for special end of paragraph check
242 refresh_.last = pos + 1;
244 needs_refresh_ = pos != -1;
247 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
249 needs_refresh_ = true;
252 current_change_number_ = change_number;
255 typedef vector<SpellResultRange> Ranges;
256 typedef Ranges::const_iterator RangesIterator;
258 /// the area of the paragraph with pending spell check
261 /// spell state cache version number
262 SpellChecker::ChangeNumber current_change_number_;
265 void correctRangesAfterPos(pos_type pos, int offset)
267 RangesIterator et = ranges_.end();
268 Ranges::iterator it = ranges_.begin();
269 for (; it != et; ++it) {
270 it->shift(pos, offset);
276 /////////////////////////////////////////////////////////////////////
278 // Paragraph::Private
280 /////////////////////////////////////////////////////////////////////
282 class Paragraph::Private
286 Private(Paragraph * owner, Layout const & layout);
287 /// "Copy constructor"
288 Private(Private const &, Paragraph * owner);
289 /// Copy constructor from \p beg to \p end
290 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
293 void insertChar(pos_type pos, char_type c, Change const & change);
295 /// Output the surrogate pair formed by \p c and \p next to \p os.
296 /// \return the number of characters written.
297 int latexSurrogatePair(otexstream & os, char_type c, char_type next,
298 OutputParams const &);
300 /// Output a space in appropriate formatting (or a surrogate pair
301 /// if the next character is a combining character).
302 /// \return whether a surrogate pair was output.
303 bool simpleTeXBlanks(OutputParams const &,
306 unsigned int & column,
308 Layout const & style);
310 /// Output consecutive unicode chars, belonging to the same script as
311 /// specified by the latex macro \p ltx, to \p os starting from \p i.
312 /// \return the number of characters written.
313 int writeScriptChars(otexstream & os, docstring const & ltx,
314 Change const &, Encoding const &, pos_type & i);
316 /// This could go to ParagraphParameters if we want to.
317 int startTeXParParams(BufferParams const &, otexstream &,
318 OutputParams const &) const;
320 /// This could go to ParagraphParameters if we want to.
321 bool endTeXParParams(BufferParams const &, otexstream &,
322 OutputParams const &) const;
325 void latexInset(BufferParams const &,
330 Font const & outerfont,
332 Change & running_change,
333 Layout const & style,
335 unsigned int & column);
338 void latexSpecialChar(
340 BufferParams const & bparams,
341 OutputParams const & runparams,
342 Font const & running_font,
343 Change const & running_change,
344 Layout const & style,
347 unsigned int & column);
354 unsigned int & column);
360 unsigned int & column);
362 bool latexSpecialTypewriter(
366 unsigned int & column);
368 bool latexSpecialPhrase(
372 unsigned int & column,
373 OutputParams const & runparams);
376 void validate(LaTeXFeatures & features) const;
378 /// Checks if the paragraph contains only text and no inset or font change.
379 bool onlyText(Buffer const & buf, Font const & outerfont,
380 pos_type initial) const;
382 /// match a string against a particular point in the paragraph
383 bool isTextAt(string const & str, pos_type pos) const;
385 /// a vector of speller skip positions
386 typedef vector<FontSpan> SkipPositions;
387 typedef SkipPositions::const_iterator SkipPositionsIterator;
389 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
391 Language * getSpellLanguage(pos_type const from) const;
393 Language * locateSpellRange(pos_type & from, pos_type & to,
394 SkipPositions & skips) const;
396 bool hasSpellerChange() const
398 SpellChecker::ChangeNumber speller_change_number = 0;
399 if (theSpellChecker())
400 speller_change_number = theSpellChecker()->changeNumber();
401 return speller_change_number > speller_state_.currentChangeNumber();
404 bool ignoreWord(docstring const & word) const ;
406 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
408 pos_type textsize = owner_->size();
409 // check for sane arguments
410 if (to <= from || from >= textsize)
412 FontSpan fp = FontSpan(from, to - 1);
413 speller_state_.setRange(fp, state);
416 void requestSpellCheck(pos_type pos)
419 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
421 speller_state_.needsRefresh(pos);
424 void readySpellCheck()
426 speller_state_.needsRefresh(-1);
429 bool needsSpellCheck() const
431 return speller_state_.needsRefresh();
434 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
436 speller_state_.refreshRange(first, last);
438 last = owner_->size();
441 pos_type endpos = last;
442 owner_->locateWord(first, endpos, WHOLE_WORD);
445 owner_->locateWord(last, endpos, WHOLE_WORD);
450 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
454 while (it != et && it->first < start) {
455 int skip = it->last - it->first + 1;
463 void markMisspelledWords(pos_type const & first, pos_type const & last,
464 SpellChecker::Result result,
465 docstring const & word,
466 SkipPositions const & skips);
468 InsetCode ownerCode() const
470 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
473 /// Which Paragraph owns us?
477 Inset const * inset_owner_;
486 ParagraphParameters params_;
488 /// for recording and looking up changes
492 InsetList insetlist_;
495 pos_type begin_of_body_;
497 typedef docstring TextContainer;
501 typedef set<docstring> Words;
502 typedef map<string, Words> LangWordsMap;
506 Layout const * layout_;
508 SpellCheckerState speller_state_;
514 struct special_phrase {
520 special_phrase const special_phrases[] = {
521 { "LyX", from_ascii("\\LyX{}"), false },
522 { "TeX", from_ascii("\\TeX{}"), true },
523 { "LaTeX2e", from_ascii("\\LaTeXe{}"), true },
524 { "LaTeX", from_ascii("\\LaTeX{}"), true },
527 size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase);
532 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
533 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
539 // Initialization of the counter for the paragraph id's,
541 // FIXME: There should be a more intelligent way to generate and use the
542 // paragraph ids per buffer instead a global static counter for all InsetText
543 // in the running program.
544 static int paragraph_id = -1;
546 Paragraph::Private::Private(Private const & p, Paragraph * owner)
547 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
548 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
549 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
552 id_ = ++paragraph_id;
553 requestSpellCheck(p.text_.size());
557 Paragraph::Private::Private(Private const & p, Paragraph * owner,
558 pos_type beg, pos_type end)
559 : owner_(owner), inset_owner_(p.inset_owner_),
560 params_(p.params_), changes_(p.changes_),
561 insetlist_(p.insetlist_, beg, end),
562 begin_of_body_(p.begin_of_body_), words_(p.words_),
565 id_ = ++paragraph_id;
566 if (beg >= pos_type(p.text_.size()))
568 text_ = p.text_.substr(beg, end - beg);
570 FontList::const_iterator fcit = fontlist_.begin();
571 FontList::const_iterator fend = fontlist_.end();
572 for (; fcit != fend; ++fcit) {
573 if (fcit->pos() < beg)
575 if (fcit->pos() >= end) {
576 // Add last entry in the fontlist_.
577 fontlist_.set(text_.size() - 1, fcit->font());
580 // Add a new entry in the fontlist_.
581 fontlist_.set(fcit->pos() - beg, fcit->font());
583 requestSpellCheck(p.text_.size());
587 void Paragraph::addChangesToToc(DocIterator const & cdit,
588 Buffer const & buf, bool output_active) const
590 d->changes_.addToToc(cdit, buf, output_active);
594 bool Paragraph::isDeleted(pos_type start, pos_type end) const
596 LASSERT(start >= 0 && start <= size(), return false);
597 LASSERT(end > start && end <= size() + 1, return false);
599 return d->changes_.isDeleted(start, end);
603 bool Paragraph::isChanged(pos_type start, pos_type end) const
605 LASSERT(start >= 0 && start <= size(), return false);
606 LASSERT(end > start && end <= size() + 1, return false);
608 return d->changes_.isChanged(start, end);
612 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
614 // keep the logic here in sync with the logic of eraseChars()
618 Change const change = d->changes_.lookup(size());
619 return change.inserted() && change.currentAuthor();
623 void Paragraph::setChange(Change const & change)
625 // beware of the imaginary end-of-par character!
626 d->changes_.set(change, 0, size() + 1);
629 * Propagate the change recursively - but not in case of DELETED!
631 * Imagine that your co-author makes changes in an existing inset. He
632 * sends your document to you and you come to the conclusion that the
633 * inset should go completely. If you erase it, LyX must not delete all
634 * text within the inset. Otherwise, the change tracked insertions of
635 * your co-author get lost and there is no way to restore them later.
637 * Conclusion: An inset's content should remain untouched if you delete it
640 if (!change.deleted()) {
641 for (pos_type pos = 0; pos < size(); ++pos) {
642 if (Inset * inset = getInset(pos))
643 inset->setChange(change);
649 void Paragraph::setChange(pos_type pos, Change const & change)
651 LASSERT(pos >= 0 && pos <= size(), return);
652 d->changes_.set(change, pos);
654 // see comment in setChange(Change const &) above
655 if (!change.deleted() && pos < size())
656 if (Inset * inset = getInset(pos))
657 inset->setChange(change);
661 Change const & Paragraph::lookupChange(pos_type pos) const
663 LBUFERR(pos >= 0 && pos <= size());
664 return d->changes_.lookup(pos);
668 void Paragraph::acceptChanges(pos_type start, pos_type end)
670 LASSERT(start >= 0 && start <= size(), return);
671 LASSERT(end > start && end <= size() + 1, return);
673 for (pos_type pos = start; pos < end; ++pos) {
674 switch (lookupChange(pos).type) {
675 case Change::UNCHANGED:
676 // accept changes in nested inset
677 if (Inset * inset = getInset(pos))
678 inset->acceptChanges();
681 case Change::INSERTED:
682 d->changes_.set(Change(Change::UNCHANGED), pos);
683 // also accept changes in nested inset
684 if (Inset * inset = getInset(pos))
685 inset->acceptChanges();
688 case Change::DELETED:
689 // Suppress access to non-existent
690 // "end-of-paragraph char"
692 eraseChar(pos, false);
703 void Paragraph::rejectChanges(pos_type start, pos_type end)
705 LASSERT(start >= 0 && start <= size(), return);
706 LASSERT(end > start && end <= size() + 1, return);
708 for (pos_type pos = start; pos < end; ++pos) {
709 switch (lookupChange(pos).type) {
710 case Change::UNCHANGED:
711 // reject changes in nested inset
712 if (Inset * inset = getInset(pos))
713 inset->rejectChanges();
716 case Change::INSERTED:
717 // Suppress access to non-existent
718 // "end-of-paragraph char"
720 eraseChar(pos, false);
726 case Change::DELETED:
727 d->changes_.set(Change(Change::UNCHANGED), pos);
729 // Do NOT reject changes within a deleted inset!
730 // There may be insertions of a co-author inside of it!
738 void Paragraph::Private::insertChar(pos_type pos, char_type c,
739 Change const & change)
741 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
744 changes_.insert(change, pos);
746 // This is actually very common when parsing buffers (and
747 // maybe inserting ascii text)
748 if (pos == pos_type(text_.size())) {
749 // when appending characters, no need to update tables
751 // but we want spell checking
752 requestSpellCheck(pos);
756 text_.insert(text_.begin() + pos, c);
758 // Update the font table.
759 fontlist_.increasePosAfterPos(pos);
762 insetlist_.increasePosAfterPos(pos);
764 // Update list of misspelled positions
765 speller_state_.increasePosAfterPos(pos);
769 bool Paragraph::insertInset(pos_type pos, Inset * inset,
770 Font const & font, Change const & change)
772 LASSERT(inset, return false);
773 LASSERT(pos >= 0 && pos <= size(), return false);
775 // Paragraph::insertInset() can be used in cut/copy/paste operation where
776 // d->inset_owner_ is not set yet.
777 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
780 d->insertChar(pos, META_INSET, change);
781 LASSERT(d->text_[pos] == META_INSET, return false);
783 // Add a new entry in the insetlist_.
784 d->insetlist_.insert(inset, pos);
786 // Some insets require run of spell checker
787 requestSpellCheck(pos);
793 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
795 LASSERT(pos >= 0 && pos <= size(), return false);
797 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
800 Change change = d->changes_.lookup(pos);
802 // set the character to DELETED if
803 // a) it was previously unchanged or
804 // b) it was inserted by a co-author
806 if (!change.changed() ||
807 (change.inserted() && !change.currentAuthor())) {
808 setChange(pos, Change(Change::DELETED));
809 // request run of spell checker
810 requestSpellCheck(pos);
814 if (change.deleted())
818 // Don't physically access the imaginary end-of-paragraph character.
819 // eraseChar() can only mark it as DELETED. A physical deletion of
820 // end-of-par must be handled externally.
826 d->changes_.erase(pos);
828 // if it is an inset, delete the inset entry
829 if (d->text_[pos] == META_INSET)
830 d->insetlist_.erase(pos);
832 d->text_.erase(d->text_.begin() + pos);
834 // Update the fontlist_
835 d->fontlist_.erase(pos);
837 // Update the insetlist_
838 d->insetlist_.decreasePosAfterPos(pos);
840 // Update list of misspelled positions
841 d->speller_state_.decreasePosAfterPos(pos);
842 d->speller_state_.refreshLast(size());
848 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
850 LASSERT(start >= 0 && start <= size(), return 0);
851 LASSERT(end >= start && end <= size() + 1, return 0);
854 for (pos_type count = end - start; count; --count) {
855 if (!eraseChar(i, trackChanges))
862 int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
863 char_type next, OutputParams const & runparams)
865 // Writing next here may circumvent a possible font change between
866 // c and next. Since next is only output if it forms a surrogate pair
867 // with c we can ignore this:
868 // A font change inside a surrogate pair does not make sense and is
869 // hopefully impossible to input.
870 // FIXME: change tracking
871 // Is this correct WRT change tracking?
872 Encoding const & encoding = *(runparams.encoding);
873 docstring latex1 = encoding.latexChar(next).first;
874 if (runparams.inIPA) {
875 string const tipashortcut = Encodings::TIPAShortcut(next);
876 if (!tipashortcut.empty()) {
877 latex1 = from_ascii(tipashortcut);
880 docstring const latex2 = encoding.latexChar(c).first;
881 if (docstring(1, next) == latex1) {
882 // the encoding supports the combination
883 os << latex2 << latex1;
884 return latex1.length() + latex2.length();
885 } else if (runparams.local_font &&
886 runparams.local_font->language()->lang() == "polutonikogreek") {
887 // polutonikogreek only works without the brackets
888 os << latex1 << latex2;
889 return latex1.length() + latex2.length();
891 os << latex1 << '{' << latex2 << '}';
892 return latex1.length() + latex2.length() + 2;
896 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
899 unsigned int & column,
901 Layout const & style)
903 if (style.pass_thru || runparams.pass_thru)
906 if (i + 1 < int(text_.size())) {
907 char_type next = text_[i + 1];
908 if (Encodings::isCombiningChar(next)) {
909 // This space has an accent, so we must always output it.
910 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
915 if (runparams.linelen > 0
916 && column > runparams.linelen
918 && text_[i - 1] != ' '
919 && (i + 1 < int(text_.size()))
920 // same in FreeSpacing mode
921 && !owner_->isFreeSpacing()
922 // In typewriter mode, we want to avoid
923 // ! . ? : at the end of a line
924 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
925 && (text_[i - 1] == '.'
926 || text_[i - 1] == '?'
927 || text_[i - 1] == ':'
928 || text_[i - 1] == '!'))) {
930 os.texrow().start(owner_->id(), i + 1);
932 } else if (style.free_spacing) {
941 int Paragraph::Private::writeScriptChars(otexstream & os,
942 docstring const & ltx,
943 Change const & runningChange,
944 Encoding const & encoding,
947 // FIXME: modifying i here is not very nice...
949 // We only arrive here when a proper language for character text_[i] has
950 // not been specified (i.e., it could not be translated in the current
951 // latex encoding) or its latex translation has been forced, and it
952 // belongs to a known script.
953 // Parameter ltx contains the latex translation of text_[i] as specified
954 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
955 // The latex macro name "textXXX" specifies the script to which text_[i]
956 // belongs and we use it in order to check whether characters from the
957 // same script immediately follow, such that we can collect them in a
958 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
959 // for the first char but only "<spec>" for all subsequent chars.
960 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
961 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
962 string script = to_ascii(ltx.substr(1, brace1 - 1));
965 bool closing_brace = true;
966 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
967 // Correct encoding is being used, so we can avoid \textgreek.
970 closing_brace = false;
972 os << ltx.substr(pos, length);
973 int size = text_.size();
974 while (i + 1 < size) {
975 char_type const next = text_[i + 1];
976 // Stop here if next character belongs to another script
977 // or there is a change in change tracking status.
978 if (!Encodings::isKnownScriptChar(next, script) ||
979 runningChange != owner_->lookupChange(i + 1))
983 FontList::const_iterator cit = fontlist_.begin();
984 FontList::const_iterator end = fontlist_.end();
985 for (; cit != end; ++cit) {
986 if (cit->pos() >= i && !found) {
987 prev_font = cit->font();
990 if (cit->pos() >= i + 1)
993 // Stop here if there is a font attribute or encoding change.
994 if (found && cit != end && prev_font != cit->font())
996 docstring const latex = encoding.latexChar(next).first;
997 docstring::size_type const b1 =
998 latex.find_first_of(from_ascii("{"));
999 docstring::size_type const b2 =
1000 latex.find_last_of(from_ascii("}"));
1001 int const len = b2 - b1 - 1;
1002 os << latex.substr(b1 + 1, len);
1006 if (closing_brace) {
1014 bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
1016 pos_type const len = str.length();
1018 // is the paragraph large enough?
1019 if (pos + len > int(text_.size()))
1022 // does the wanted text start at point?
1023 for (string::size_type i = 0; i < str.length(); ++i) {
1024 // Caution: direct comparison of characters works only
1025 // because str is pure ASCII.
1026 if (str[i] != text_[pos + i])
1030 return fontlist_.hasChangeInRange(pos, len);
1034 void Paragraph::Private::latexInset(BufferParams const & bparams,
1036 OutputParams & runparams,
1037 Font & running_font,
1039 Font const & outerfont,
1041 Change & running_change,
1042 Layout const & style,
1044 unsigned int & column)
1046 Inset * inset = owner_->getInset(i);
1049 if (style.pass_thru) {
1050 odocstringstream ods;
1051 inset->plaintext(ods, runparams);
1056 // FIXME: move this to InsetNewline::latex
1057 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1058 // newlines are handled differently here than
1059 // the default in simpleTeXSpecialChars().
1060 if (!style.newline_allowed) {
1064 column += running_font.latexWriteEndChanges(
1065 os, bparams, runparams,
1066 basefont, basefont);
1070 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1073 basefont = owner_->getLayoutFont(bparams, outerfont);
1074 running_font = basefont;
1076 if (runparams.moving_arg)
1080 os.texrow().start(owner_->id(), i + 1);
1084 if (owner_->isDeleted(i)) {
1085 if( ++runparams.inDeletedInset == 1)
1086 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1089 if (inset->canTrackChanges()) {
1090 column += Changes::latexMarkChange(os, bparams, running_change,
1091 Change(Change::UNCHANGED), runparams);
1092 running_change = Change(Change::UNCHANGED);
1096 odocstream::pos_type const len = os.os().tellp();
1098 if (inset->forceLTR()
1099 && !runparams.use_polyglossia
1100 && running_font.isRightToLeft()
1101 // ERT is an exception, it should be output with no
1102 // decorations at all
1103 && inset->lyxCode() != ERT_CODE) {
1104 if (running_font.language()->lang() == "farsi")
1111 // FIXME: Bug: we can have an empty font change here!
1112 // if there has just been a font change, we are going to close it
1113 // right now, which means stupid latex code like \textsf{}. AFAIK,
1114 // this does not harm dvi output. A minor bug, thus (JMarc)
1116 // Some insets cannot be inside a font change command.
1117 // However, even such insets *can* be placed in \L or \R
1118 // or their equivalents (for RTL language switches), so we don't
1119 // close the language in those cases.
1120 // ArabTeX, though, cannot handle this special behavior, it seems.
1121 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1122 || running_font.language()->lang() == "arabic_arabtex";
1123 if (open_font && !inset->inheritFont()) {
1124 bool closeLanguage = arabtex
1125 || basefont.isRightToLeft() == running_font.isRightToLeft();
1126 unsigned int count = running_font.latexWriteEndChanges(os,
1127 bparams, runparams, basefont, basefont, closeLanguage);
1129 // if any font properties were closed, update the running_font,
1130 // making sure, however, to leave the language as it was
1132 // FIXME: probably a better way to keep track of the old
1133 // language, than copying the entire font?
1134 Font const copy_font(running_font);
1135 basefont = owner_->getLayoutFont(bparams, outerfont);
1136 running_font = basefont;
1138 running_font.setLanguage(copy_font.language());
1139 // leave font open if language is still open
1140 open_font = (running_font.language() == basefont.language());
1142 runparams.local_font = &basefont;
1146 int prev_rows = os.texrow().rows();
1149 runparams.lastid = id_;
1150 runparams.lastpos = i;
1151 inset->latex(os, runparams);
1152 } catch (EncodingException & e) {
1153 // add location information and throw again.
1160 if (running_font.language()->lang() == "farsi")
1166 if (os.texrow().rows() > prev_rows) {
1167 os.texrow().start(owner_->id(), i + 1);
1170 column += (unsigned int)(os.os().tellp() - len);
1173 if (owner_->isDeleted(i))
1174 --runparams.inDeletedInset;
1178 void Paragraph::Private::latexSpecialChar(otexstream & os,
1179 BufferParams const & bparams,
1180 OutputParams const & runparams,
1181 Font const & running_font,
1182 Change const & running_change,
1183 Layout const & style,
1186 unsigned int & column)
1188 // With polyglossia, brackets and stuff need not be reversed
1189 // in RTL scripts (see bug #8251)
1190 char_type const c = (runparams.use_polyglossia) ?
1191 owner_->getUChar(bparams, i) : text_[i];
1193 if (style.pass_thru || runparams.pass_thru) {
1195 Encoding const * const enc = runparams.encoding;
1196 if (enc && !enc->encodable(c))
1197 throw EncodingException(c);
1203 // TIPA uses its own T3 encoding
1204 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1206 // If T1 font encoding is used, use the special
1207 // characters it provides.
1208 // NOTE: some languages reset the font encoding
1210 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1211 && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
1214 // \tt font needs special treatment
1215 if (!runparams.inIPA
1216 && running_font.fontInfo().family() == TYPEWRITER_FAMILY
1217 && latexSpecialTypewriter(c, os, i, column))
1220 // Otherwise, we use what LaTeX provides us.
1223 os << "\\textbackslash{}";
1227 os << "\\textless{}";
1231 os << "\\textgreater{}";
1235 os << "\\textbar{}";
1242 os << "\\char`\\\"{}";
1247 case '%': case '#': case '{':
1255 os << "\\textasciitilde{}";
1260 os << "\\textasciicircum{}";
1267 // avoid being mistaken for optional arguments
1275 // Blanks are printed before font switching.
1276 // Sure? I am not! (try nice-latex)
1277 // I am sure it's correct. LyX might be smarter
1278 // in the future, but for now, nothing wrong is
1284 if (latexSpecialPhrase(os, i, end_pos, column, runparams))
1290 Encoding const & encoding = *(runparams.encoding);
1291 char_type next = '\0';
1292 if (i + 1 < int(text_.size())) {
1293 next = text_[i + 1];
1294 if (Encodings::isCombiningChar(next)) {
1295 column += latexSurrogatePair(os, c, next, runparams) - 1;
1301 pair<docstring, bool> latex = encoding.latexChar(c);
1302 docstring nextlatex;
1303 bool nexttipas = false;
1304 string nexttipashortcut;
1305 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1306 nextlatex = encoding.latexChar(next).first;
1307 if (runparams.inIPA) {
1308 nexttipashortcut = Encodings::TIPAShortcut(next);
1309 nexttipas = !nexttipashortcut.empty();
1313 if (runparams.inIPA) {
1314 string const tipashortcut = Encodings::TIPAShortcut(c);
1315 if (!tipashortcut.empty()) {
1316 latex.first = from_ascii(tipashortcut);
1317 latex.second = false;
1321 if (Encodings::isKnownScriptChar(c, script)
1322 && prefixIs(latex.first, from_ascii("\\" + script)))
1323 column += writeScriptChars(os, latex.first,
1324 running_change, encoding, i) - 1;
1325 else if (latex.second
1326 && ((!prefixIs(nextlatex, '\\')
1327 && !prefixIs(nextlatex, '{')
1328 && !prefixIs(nextlatex, '}'))
1330 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1332 // Prevent eating of a following
1333 // space or command corruption by
1334 // following characters
1335 if (next == ' ' || next == '\0') {
1336 column += latex.first.length() + 1;
1337 os << latex.first << "{}";
1339 column += latex.first.length();
1340 os << latex.first << " ";
1343 column += latex.first.length() - 1;
1351 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1352 pos_type i, unsigned int & column)
1358 // In T1 encoding, these characters exist
1359 // but we should avoid ligatures
1360 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1362 os << "\\textcompwordmark{}";
1369 // soul.sty breaks with \char`\"
1370 os << "\\textquotedbl{}";
1379 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1380 pos_type /*i*/, unsigned int & column)
1390 os << "\\textvertline{}";
1399 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, otexstream & os,
1400 pos_type i, unsigned int & column)
1404 // within \ttfamily, "--" is merged to "-" (no endash)
1405 // so we avoid this rather irritating ligature
1406 if (i + 1 < int(text_.size()) && text_[i + 1] == '-') {
1413 // everything else has to be checked separately
1414 // (depending on the encoding)
1422 /// If [start_pos, end_pos) does not include entirely the special phrase, then
1423 /// do not apply the macro transformation.
1424 bool Paragraph::Private::latexSpecialPhrase(otexstream & os, pos_type & i, pos_type end_pos,
1425 unsigned int & column, OutputParams const & runparams)
1427 // FIXME: if we have "LaTeX" with a font
1428 // change in the middle (before the 'T', then
1429 // the "TeX" part is still special cased.
1430 // Really we should only operate this on
1431 // "words" for some definition of word
1433 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1434 if (!isTextAt(special_phrases[pnr].phrase, i)
1435 || (end_pos != -1 && i + int(special_phrases[pnr].phrase.size()) > end_pos))
1437 if (runparams.moving_arg)
1439 os << special_phrases[pnr].macro;
1440 i += special_phrases[pnr].phrase.length() - 1;
1441 column += special_phrases[pnr].macro.length() - 1;
1448 void Paragraph::Private::validate(LaTeXFeatures & features) const
1450 if (layout_->inpreamble && inset_owner_) {
1451 bool const is_command = layout_->latextype == LATEX_COMMAND;
1452 Buffer const & buf = inset_owner_->buffer();
1453 BufferParams const & bp = features.runparams().is_child
1454 ? buf.masterParams() : buf.params();
1457 // Using a string stream here circumvents the encoding
1458 // switching machinery of odocstream. Therefore the
1459 // output is wrong if this paragraph contains content
1460 // that needs to switch encoding.
1461 odocstringstream ods;
1462 otexstream os(ods, texrow);
1464 os << '\\' << from_ascii(layout_->latexname());
1465 // we have to provide all the optional arguments here, even though
1466 // the last one is the only one we care about.
1467 // Separate handling of optional argument inset.
1468 if (!layout_->latexargs().empty()) {
1469 OutputParams rp = features.runparams();
1470 rp.local_font = &owner_->getFirstFontSettings(bp);
1471 latexArgInsets(*owner_, os, rp, layout_->latexargs());
1473 os << from_ascii(layout_->latexparam());
1475 docstring::size_type const length = ods.str().length();
1476 // this will output "{" at the beginning, but not at the end
1477 owner_->latex(bp, f, os, features.runparams(), 0, -1, true);
1478 if (ods.str().length() > length) {
1481 if (!layout_->postcommandargs().empty()) {
1482 OutputParams rp = features.runparams();
1483 rp.local_font = &owner_->getFirstFontSettings(bp);
1484 latexArgInsets(*owner_, os, rp, layout_->postcommandargs(), "post:");
1487 string const snippet = to_utf8(ods.str());
1488 features.addPreambleSnippet(snippet);
1492 if (features.runparams().flavor == OutputParams::HTML
1493 && layout_->htmltitle()) {
1494 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1497 // check the params.
1498 if (!params_.spacing().isDefault())
1499 features.require("setspace");
1502 features.useLayout(layout_->name());
1505 fontlist_.validate(features);
1507 // then the indentation
1508 if (!params_.leftIndent().zero())
1509 features.require("ParagraphLeftIndent");
1512 InsetList::const_iterator icit = insetlist_.begin();
1513 InsetList::const_iterator iend = insetlist_.end();
1514 for (; icit != iend; ++icit) {
1516 icit->inset->validate(features);
1517 if (layout_->needprotect &&
1518 icit->inset->lyxCode() == FOOT_CODE)
1519 features.require("NeedLyXFootnoteCode");
1523 // then the contents
1524 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1525 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1526 if (!special_phrases[pnr].builtin
1527 && isTextAt(special_phrases[pnr].phrase, i)) {
1528 features.require(special_phrases[pnr].phrase);
1532 BufferEncodings::validate(text_[i], features);
1536 /////////////////////////////////////////////////////////////////////
1540 /////////////////////////////////////////////////////////////////////
1543 Layout const emptyParagraphLayout;
1546 Paragraph::Paragraph()
1547 : d(new Paragraph::Private(this, emptyParagraphLayout))
1554 Paragraph::Paragraph(Paragraph const & par)
1555 : itemdepth(par.itemdepth),
1556 d(new Paragraph::Private(*par.d, this))
1562 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1563 : itemdepth(par.itemdepth),
1564 d(new Paragraph::Private(*par.d, this, beg, end))
1570 Paragraph & Paragraph::operator=(Paragraph const & par)
1572 // needed as we will destroy the private part before copying it
1574 itemdepth = par.itemdepth;
1578 d = new Private(*par.d, this);
1585 Paragraph::~Paragraph()
1594 // this shall be called just before every "os << ..." action.
1595 void flushString(ostream & os, docstring & s)
1604 void Paragraph::write(ostream & os, BufferParams const & bparams,
1605 depth_type & dth) const
1607 // The beginning or end of a deeper (i.e. nested) area?
1608 if (dth != d->params_.depth()) {
1609 if (d->params_.depth() > dth) {
1610 while (d->params_.depth() > dth) {
1611 os << "\n\\begin_deeper";
1615 while (d->params_.depth() < dth) {
1616 os << "\n\\end_deeper";
1622 // First write the layout
1623 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1625 d->params_.write(os);
1627 Font font1(inherit_font, bparams.language);
1629 Change running_change = Change(Change::UNCHANGED);
1631 // this string is used as a buffer to avoid repetitive calls
1632 // to to_utf8(), which turn out to be expensive (JMarc)
1633 docstring write_buffer;
1636 for (pos_type i = 0; i <= size(); ++i) {
1638 Change const change = lookupChange(i);
1639 if (change != running_change)
1640 flushString(os, write_buffer);
1641 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1642 running_change = change;
1647 // Write font changes
1648 Font font2 = getFontSettings(bparams, i);
1649 if (font2 != font1) {
1650 flushString(os, write_buffer);
1651 font2.lyxWriteChanges(font1, os);
1656 char_type const c = d->text_[i];
1659 if (Inset const * inset = getInset(i)) {
1660 flushString(os, write_buffer);
1661 if (inset->directWrite()) {
1662 // international char, let it write
1663 // code directly so it's shorter in
1669 os << "\\begin_inset ";
1671 os << "\n\\end_inset\n\n";
1677 flushString(os, write_buffer);
1678 os << "\n\\backslash\n";
1682 flushString(os, write_buffer);
1683 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1690 if ((column > 70 && c == ' ')
1692 flushString(os, write_buffer);
1696 // this check is to amend a bug. LyX sometimes
1697 // inserts '\0' this could cause problems.
1699 write_buffer.push_back(c);
1701 LYXERR0("NUL char in structure.");
1707 flushString(os, write_buffer);
1708 os << "\n\\end_layout\n";
1712 void Paragraph::validate(LaTeXFeatures & features) const
1714 d->validate(features);
1718 void Paragraph::insert(pos_type start, docstring const & str,
1719 Font const & font, Change const & change)
1721 for (size_t i = 0, n = str.size(); i != n ; ++i)
1722 insertChar(start + i, str[i], font, change);
1726 void Paragraph::appendChar(char_type c, Font const & font,
1727 Change const & change)
1730 d->changes_.insert(change, d->text_.size());
1731 // when appending characters, no need to update tables
1732 d->text_.push_back(c);
1733 setFont(d->text_.size() - 1, font);
1734 d->requestSpellCheck(d->text_.size() - 1);
1738 void Paragraph::appendString(docstring const & s, Font const & font,
1739 Change const & change)
1741 pos_type end = s.size();
1742 size_t oldsize = d->text_.size();
1743 size_t newsize = oldsize + end;
1744 size_t capacity = d->text_.capacity();
1745 if (newsize >= capacity)
1746 d->text_.reserve(max(capacity + 100, newsize));
1748 // when appending characters, no need to update tables
1751 // FIXME: Optimize this!
1752 for (size_t i = oldsize; i != newsize; ++i) {
1754 d->changes_.insert(change, i);
1755 d->requestSpellCheck(i);
1757 d->fontlist_.set(oldsize, font);
1758 d->fontlist_.set(newsize - 1, font);
1762 void Paragraph::insertChar(pos_type pos, char_type c,
1765 d->insertChar(pos, c, Change(trackChanges ?
1766 Change::INSERTED : Change::UNCHANGED));
1770 void Paragraph::insertChar(pos_type pos, char_type c,
1771 Font const & font, bool trackChanges)
1773 d->insertChar(pos, c, Change(trackChanges ?
1774 Change::INSERTED : Change::UNCHANGED));
1779 void Paragraph::insertChar(pos_type pos, char_type c,
1780 Font const & font, Change const & change)
1782 d->insertChar(pos, c, change);
1787 void Paragraph::resetFonts(Font const & font)
1789 d->fontlist_.clear();
1790 d->fontlist_.set(0, font);
1791 d->fontlist_.set(d->text_.size() - 1, font);
1794 // Gets uninstantiated font setting at position.
1795 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1799 LYXERR0("pos: " << pos << " size: " << size());
1803 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1804 if (cit != d->fontlist_.end())
1807 if (pos == size() && !empty())
1808 return getFontSettings(bparams, pos - 1);
1810 // Optimisation: avoid a full font instantiation if there is no
1811 // language change from previous call.
1812 static Font previous_font;
1813 static Language const * previous_lang = 0;
1814 Language const * lang = getParLanguage(bparams);
1815 if (lang != previous_lang) {
1816 previous_lang = lang;
1817 previous_font = Font(inherit_font, lang);
1819 return previous_font;
1823 FontSpan Paragraph::fontSpan(pos_type pos) const
1825 LBUFERR(pos < size());
1828 FontList::const_iterator cit = d->fontlist_.begin();
1829 FontList::const_iterator end = d->fontlist_.end();
1830 for (; cit != end; ++cit) {
1831 if (cit->pos() >= pos) {
1832 if (pos >= beginOfBody())
1833 return FontSpan(max(start, beginOfBody()),
1836 return FontSpan(start,
1837 min(beginOfBody() - 1,
1840 start = cit->pos() + 1;
1843 // This should not happen, but if so, we take no chances.
1844 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1845 LASSERT(false, return FontSpan(pos, pos));
1849 // Gets uninstantiated font setting at position 0
1850 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1852 if (!empty() && !d->fontlist_.empty())
1853 return d->fontlist_.begin()->font();
1855 // Optimisation: avoid a full font instantiation if there is no
1856 // language change from previous call.
1857 static Font previous_font;
1858 static Language const * previous_lang = 0;
1859 if (bparams.language != previous_lang) {
1860 previous_lang = bparams.language;
1861 previous_font = Font(inherit_font, bparams.language);
1864 return previous_font;
1868 // Gets the fully instantiated font at a given position in a paragraph
1869 // This is basically the same function as Text::GetFont() in text2.cpp.
1870 // The difference is that this one is used for generating the LaTeX file,
1871 // and thus cosmetic "improvements" are disallowed: This has to deliver
1872 // the true picture of the buffer. (Asger)
1873 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1874 Font const & outerfont) const
1878 Font font = getFontSettings(bparams, pos);
1880 pos_type const body_pos = beginOfBody();
1881 FontInfo & fi = font.fontInfo();
1883 fi.realize(d->layout_->labelfont);
1885 fi.realize(d->layout_->font);
1887 fi.realize(outerfont.fontInfo());
1888 fi.realize(bparams.getFont().fontInfo());
1894 Font const Paragraph::getLabelFont
1895 (BufferParams const & bparams, Font const & outerfont) const
1897 FontInfo tmpfont = d->layout_->labelfont;
1898 tmpfont.realize(outerfont.fontInfo());
1899 tmpfont.realize(bparams.getFont().fontInfo());
1900 return Font(tmpfont, getParLanguage(bparams));
1904 Font const Paragraph::getLayoutFont
1905 (BufferParams const & bparams, Font const & outerfont) const
1907 FontInfo tmpfont = d->layout_->font;
1908 tmpfont.realize(outerfont.fontInfo());
1909 tmpfont.realize(bparams.getFont().fontInfo());
1910 return Font(tmpfont, getParLanguage(bparams));
1914 /// Returns the height of the highest font in range
1915 FontSize Paragraph::highestFontInRange
1916 (pos_type startpos, pos_type endpos, FontSize def_size) const
1918 return d->fontlist_.highestInRange(startpos, endpos, def_size);
1922 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1924 char_type c = d->text_[pos];
1925 if (!lyxrc.rtl_support || !getFontSettings(bparams, pos).isRightToLeft())
1928 // FIXME: The arabic special casing is due to the difference of arabic
1929 // round brackets input introduced in r18599. Check if this should be
1930 // unified with Hebrew or at least if all bracket types should be
1931 // handled the same (file format change in either case).
1932 string const & lang = getFontSettings(bparams, pos).language()->lang();
1933 bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
1938 uc = arabic ? c : ')';
1941 uc = arabic ? c : '(';
1967 void Paragraph::setFont(pos_type pos, Font const & font)
1969 LASSERT(pos <= size(), return);
1971 // First, reduce font against layout/label font
1972 // Update: The setCharFont() routine in text2.cpp already
1973 // reduces font, so we don't need to do that here. (Asger)
1975 d->fontlist_.set(pos, font);
1979 void Paragraph::makeSameLayout(Paragraph const & par)
1981 d->layout_ = par.d->layout_;
1982 d->params_ = par.d->params_;
1986 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1988 if (isFreeSpacing())
1994 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1995 if (eraseChar(pos, trackChanges))
2001 return count > 0 || pos > 0;
2005 bool Paragraph::hasSameLayout(Paragraph const & par) const
2007 return par.d->layout_ == d->layout_
2008 && d->params_.sameLayout(par.d->params_);
2012 depth_type Paragraph::getDepth() const
2014 return d->params_.depth();
2018 depth_type Paragraph::getMaxDepthAfter() const
2020 if (d->layout_->isEnvironment())
2021 return d->params_.depth() + 1;
2023 return d->params_.depth();
2027 char Paragraph::getAlign() const
2029 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2030 return d->layout_->align;
2032 return d->params_.align();
2036 docstring const & Paragraph::labelString() const
2038 return d->params_.labelString();
2042 // the next two functions are for the manual labels
2043 docstring const Paragraph::getLabelWidthString() const
2045 if (d->layout_->margintype == MARGIN_MANUAL
2046 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2047 return d->params_.labelWidthString();
2049 return _("Senseless with this layout!");
2053 void Paragraph::setLabelWidthString(docstring const & s)
2055 d->params_.labelWidthString(s);
2059 docstring Paragraph::expandLabel(Layout const & layout,
2060 BufferParams const & bparams) const
2062 return expandParagraphLabel(layout, bparams, true);
2066 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2067 BufferParams const & bparams) const
2069 return expandParagraphLabel(layout, bparams, false);
2073 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2074 BufferParams const & bparams, bool process_appendix) const
2076 DocumentClass const & tclass = bparams.documentClass();
2077 string const & lang = getParLanguage(bparams)->code();
2078 bool const in_appendix = process_appendix && d->params_.appendix();
2079 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2081 if (fmt.empty() && !layout.counter.empty())
2082 return tclass.counters().theCounter(layout.counter, lang);
2084 // handle 'inherited level parts' in 'fmt',
2085 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2086 size_t const i = fmt.find('@', 0);
2087 if (i != docstring::npos) {
2088 size_t const j = fmt.find('@', i + 1);
2089 if (j != docstring::npos) {
2090 docstring parent(fmt, i + 1, j - i - 1);
2091 docstring label = from_ascii("??");
2092 if (tclass.hasLayout(parent))
2093 docstring label = expandParagraphLabel(tclass[parent], bparams,
2095 fmt = docstring(fmt, 0, i) + label
2096 + docstring(fmt, j + 1, docstring::npos);
2100 return tclass.counters().counterLabel(fmt, lang);
2104 void Paragraph::applyLayout(Layout const & new_layout)
2106 d->layout_ = &new_layout;
2107 LyXAlignment const oldAlign = d->params_.align();
2109 if (!(oldAlign & d->layout_->alignpossible)) {
2110 frontend::Alert::warning(_("Alignment not permitted"),
2111 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2112 d->params_.align(LYX_ALIGN_LAYOUT);
2117 pos_type Paragraph::beginOfBody() const
2119 return d->begin_of_body_;
2123 void Paragraph::setBeginOfBody()
2125 if (d->layout_->labeltype != LABEL_MANUAL) {
2126 d->begin_of_body_ = 0;
2130 // Unroll the first two cycles of the loop
2131 // and remember the previous character to
2132 // remove unnecessary getChar() calls
2134 pos_type end = size();
2135 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2137 char_type previous_char = 0;
2140 previous_char = d->text_[i];
2141 if (!(isNewline(i) || isEnvSeparator(i))) {
2143 while (i < end && previous_char != ' ') {
2145 if (isNewline(i) || isEnvSeparator(i))
2148 previous_char = temp;
2154 d->begin_of_body_ = i;
2158 bool Paragraph::allowParagraphCustomization() const
2160 return inInset().allowParagraphCustomization();
2164 bool Paragraph::usePlainLayout() const
2166 return inInset().usePlainLayout();
2170 bool Paragraph::isPassThru() const
2172 return inInset().isPassThru() || d->layout_->pass_thru;
2177 // paragraphs inside floats need different alignment tags to avoid
2180 bool noTrivlistCentering(InsetCode code)
2182 return code == FLOAT_CODE
2183 || code == WRAP_CODE
2184 || code == CELL_CODE;
2188 string correction(string const & orig)
2190 if (orig == "flushleft")
2191 return "raggedright";
2192 if (orig == "flushright")
2193 return "raggedleft";
2194 if (orig == "center")
2200 string const corrected_env(string const & suffix, string const & env,
2201 InsetCode code, bool const lastpar)
2203 string output = suffix + "{";
2204 if (noTrivlistCentering(code)) {
2206 // the last paragraph in non-trivlist-aligned
2207 // context is special (to avoid unwanted whitespace)
2208 if (suffix == "\\begin")
2209 return "\\" + correction(env) + "{}";
2212 output += correction(env);
2216 if (suffix == "\\begin")
2222 void adjust_column(string const & str, int & column)
2224 if (!contains(str, "\n"))
2225 column += str.size();
2228 column = rsplit(str, tmp, '\n').size();
2235 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2236 otexstream & os, OutputParams const & runparams) const
2240 if (params_.noindent() && !layout_->pass_thru
2241 && (layout_->toggle_indent != ITOGGLE_NEVER)) {
2242 os << "\\noindent ";
2246 LyXAlignment const curAlign = params_.align();
2248 if (curAlign == layout_->align)
2252 case LYX_ALIGN_NONE:
2253 case LYX_ALIGN_BLOCK:
2254 case LYX_ALIGN_LAYOUT:
2255 case LYX_ALIGN_SPECIAL:
2256 case LYX_ALIGN_DECIMAL:
2258 case LYX_ALIGN_LEFT:
2259 case LYX_ALIGN_RIGHT:
2260 case LYX_ALIGN_CENTER:
2261 if (runparams.moving_arg) {
2268 string const begin_tag = "\\begin";
2269 InsetCode code = ownerCode();
2270 bool const lastpar = runparams.isLastPar;
2273 case LYX_ALIGN_NONE:
2274 case LYX_ALIGN_BLOCK:
2275 case LYX_ALIGN_LAYOUT:
2276 case LYX_ALIGN_SPECIAL:
2277 case LYX_ALIGN_DECIMAL:
2279 case LYX_ALIGN_LEFT: {
2281 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2282 output = corrected_env(begin_tag, "flushleft", code, lastpar);
2284 output = corrected_env(begin_tag, "flushright", code, lastpar);
2285 os << from_ascii(output);
2286 adjust_column(output, column);
2288 } case LYX_ALIGN_RIGHT: {
2290 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2291 output = corrected_env(begin_tag, "flushright", code, lastpar);
2293 output = corrected_env(begin_tag, "flushleft", code, lastpar);
2294 os << from_ascii(output);
2295 adjust_column(output, column);
2297 } case LYX_ALIGN_CENTER: {
2299 output = corrected_env(begin_tag, "center", code, lastpar);
2300 os << from_ascii(output);
2301 adjust_column(output, column);
2310 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2311 otexstream & os, OutputParams const & runparams) const
2313 LyXAlignment const curAlign = params_.align();
2315 if (curAlign == layout_->align)
2319 case LYX_ALIGN_NONE:
2320 case LYX_ALIGN_BLOCK:
2321 case LYX_ALIGN_LAYOUT:
2322 case LYX_ALIGN_SPECIAL:
2323 case LYX_ALIGN_DECIMAL:
2325 case LYX_ALIGN_LEFT:
2326 case LYX_ALIGN_RIGHT:
2327 case LYX_ALIGN_CENTER:
2328 if (runparams.moving_arg)
2334 string const end_tag = "\n\\par\\end";
2335 InsetCode code = ownerCode();
2336 bool const lastpar = runparams.isLastPar;
2339 case LYX_ALIGN_NONE:
2340 case LYX_ALIGN_BLOCK:
2341 case LYX_ALIGN_LAYOUT:
2342 case LYX_ALIGN_SPECIAL:
2343 case LYX_ALIGN_DECIMAL:
2345 case LYX_ALIGN_LEFT: {
2346 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2347 output = corrected_env(end_tag, "flushleft", code, lastpar);
2349 output = corrected_env(end_tag, "flushright", code, lastpar);
2350 os << from_ascii(output);
2352 } case LYX_ALIGN_RIGHT: {
2353 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2354 output = corrected_env(end_tag, "flushright", code, lastpar);
2356 output = corrected_env(end_tag, "flushleft", code, lastpar);
2357 os << from_ascii(output);
2359 } case LYX_ALIGN_CENTER: {
2360 output = corrected_env(end_tag, "center", code, lastpar);
2361 os << from_ascii(output);
2366 return !output.empty() || lastpar;
2370 // This one spits out the text of the paragraph
2371 void Paragraph::latex(BufferParams const & bparams,
2372 Font const & outerfont,
2374 OutputParams const & runparams,
2375 int start_pos, int end_pos, bool force) const
2377 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2379 // FIXME This check should not be needed. Perhaps issue an
2380 // error if it triggers.
2381 Layout const & style = inInset().forcePlainLayout() ?
2382 bparams.documentClass().plainLayout() : *d->layout_;
2384 if (!force && style.inpreamble)
2387 bool const allowcust = allowParagraphCustomization();
2389 // Current base font for all inherited font changes, without any
2390 // change caused by an individual character, except for the language:
2391 // It is set to the language of the first character.
2392 // As long as we are in the label, this font is the base font of the
2393 // label. Before the first body character it is set to the base font
2397 // Maybe we have to create a optional argument.
2398 pos_type body_pos = beginOfBody();
2399 unsigned int column = 0;
2402 // the optional argument is kept in curly brackets in
2403 // case it contains a ']'
2404 // This is not strictly needed, but if this is changed it
2405 // would be a file format change, and tex2lyx would need
2406 // to be adjusted, since it unconditionally removes the
2407 // braces when it parses \item.
2410 basefont = getLabelFont(bparams, outerfont);
2412 basefont = getLayoutFont(bparams, outerfont);
2415 // Which font is currently active?
2416 Font running_font(basefont);
2417 // Do we have an open font change?
2418 bool open_font = false;
2420 Change runningChange = Change(Change::UNCHANGED);
2422 Encoding const * const prev_encoding = runparams.encoding;
2424 os.texrow().start(id(), 0);
2426 // if the paragraph is empty, the loop will not be entered at all
2428 if (style.isCommand()) {
2432 if (!style.leftdelim().empty()) {
2433 os << style.leftdelim();
2434 column += style.leftdelim().size();
2437 column += d->startTeXParParams(bparams, os, runparams);
2440 for (pos_type i = 0; i < size(); ++i) {
2441 // First char in paragraph or after label?
2442 if (i == body_pos) {
2445 column += running_font.latexWriteEndChanges(
2446 os, bparams, runparams,
2447 basefont, basefont);
2450 basefont = getLayoutFont(bparams, outerfont);
2451 running_font = basefont;
2453 column += Changes::latexMarkChange(os, bparams,
2454 runningChange, Change(Change::UNCHANGED),
2456 runningChange = Change(Change::UNCHANGED);
2461 if (style.isCommand()) {
2466 if (!style.leftdelim().empty()) {
2467 os << style.leftdelim();
2468 column += style.leftdelim().size();
2472 column += d->startTeXParParams(bparams, os,
2476 Change const & change = runparams.inDeletedInset
2477 ? runparams.changeOfDeletedInset : lookupChange(i);
2479 if (bparams.output_changes && runningChange != change) {
2481 column += running_font.latexWriteEndChanges(
2482 os, bparams, runparams, basefont, basefont);
2485 basefont = getLayoutFont(bparams, outerfont);
2486 running_font = basefont;
2488 column += Changes::latexMarkChange(os, bparams, runningChange,
2490 runningChange = change;
2493 // do not output text which is marked deleted
2494 // if change tracking output is disabled
2495 if (!bparams.output_changes && change.deleted()) {
2501 // Fully instantiated font
2502 Font const font = getFont(bparams, i, outerfont);
2504 Font const last_font = running_font;
2506 // Do we need to close the previous font?
2508 (font != running_font ||
2509 font.language() != running_font.language()))
2511 column += running_font.latexWriteEndChanges(
2512 os, bparams, runparams, basefont,
2513 (i == body_pos-1) ? basefont : font);
2514 running_font = basefont;
2518 string const running_lang = runparams.use_polyglossia ?
2519 running_font.language()->polyglossia() : running_font.language()->babel();
2520 // close babel's font environment before opening CJK.
2521 string const lang_end_command = runparams.use_polyglossia ?
2522 "\\end{$$lang}" : lyxrc.language_command_end;
2523 if (!running_lang.empty() &&
2524 font.language()->encoding()->package() == Encoding::CJK) {
2525 string end_tag = subst(lang_end_command,
2528 os << from_ascii(end_tag);
2529 column += end_tag.length();
2532 // Switch file encoding if necessary (and allowed)
2533 if (!runparams.pass_thru && !style.pass_thru &&
2534 runparams.encoding->package() != Encoding::none &&
2535 font.language()->encoding()->package() != Encoding::none) {
2536 pair<bool, int> const enc_switch =
2537 switchEncoding(os.os(), bparams, runparams,
2538 *(font.language()->encoding()));
2539 if (enc_switch.first) {
2540 column += enc_switch.second;
2541 runparams.encoding = font.language()->encoding();
2545 char_type const c = d->text_[i];
2547 // Do we need to change font?
2548 if ((font != running_font ||
2549 font.language() != running_font.language()) &&
2552 odocstringstream ods;
2553 column += font.latexWriteStartChanges(ods, bparams,
2554 runparams, basefont,
2556 running_font = font;
2558 docstring fontchange = ods.str();
2559 // check whether the fontchange ends with a \\textcolor
2560 // modifier and the text starts with a space (bug 4473)
2561 docstring const last_modifier = rsplit(fontchange, '\\');
2562 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2563 os << fontchange << from_ascii("{}");
2564 // check if the fontchange ends with a trailing blank
2565 // (like "\small " (see bug 3382)
2566 else if (suffixIs(fontchange, ' ') && c == ' ')
2567 os << fontchange.substr(0, fontchange.size() - 1)
2568 << from_ascii("{}");
2573 // FIXME: think about end_pos implementation...
2574 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2575 // FIXME: integrate this case in latexSpecialChar
2576 // Do not print the separation of the optional argument
2577 // if style.pass_thru is false. This works because
2578 // latexSpecialChar ignores spaces if
2579 // style.pass_thru is false.
2580 if (i != body_pos - 1) {
2581 if (d->simpleTeXBlanks(runparams, os,
2582 i, column, font, style)) {
2583 // A surrogate pair was output. We
2584 // must not call latexSpecialChar
2585 // in this iteration, since it would output
2586 // the combining character again.
2593 OutputParams rp = runparams;
2594 rp.free_spacing = style.free_spacing;
2595 rp.local_font = &font;
2596 rp.intitle = style.intitle;
2598 // Two major modes: LaTeX or plain
2599 // Handle here those cases common to both modes
2600 // and then split to handle the two modes separately.
2601 if (c == META_INSET) {
2602 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2603 d->latexInset(bparams, os, rp, running_font,
2604 basefont, outerfont, open_font,
2605 runningChange, style, i, column);
2608 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2610 d->latexSpecialChar(os, bparams, rp, running_font, runningChange,
2611 style, i, end_pos, column);
2612 } catch (EncodingException & e) {
2613 if (runparams.dryrun) {
2614 os << "<" << _("LyX Warning: ")
2615 << _("uncodable character") << " '";
2619 // add location information and throw again.
2628 // Set the encoding to that returned from latexSpecialChar (see
2629 // comment for encoding member in OutputParams.h)
2630 runparams.encoding = rp.encoding;
2633 // If we have an open font definition, we have to close it
2635 #ifdef FIXED_LANGUAGE_END_DETECTION
2637 running_font.latexWriteEndChanges(os, bparams,
2638 runparams, basefont,
2639 next_->getFont(bparams, 0, outerfont));
2641 running_font.latexWriteEndChanges(os, bparams,
2642 runparams, basefont, basefont);
2645 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2646 //FIXME: there as we start another \selectlanguage with the next paragraph if
2647 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2648 running_font.latexWriteEndChanges(os, bparams, runparams,
2649 basefont, basefont);
2653 column += Changes::latexMarkChange(os, bparams, runningChange,
2654 Change(Change::UNCHANGED), runparams);
2656 // Needed if there is an optional argument but no contents.
2657 if (body_pos > 0 && body_pos == size()) {
2661 if (!style.rightdelim().empty()) {
2662 os << style.rightdelim();
2663 column += style.rightdelim().size();
2666 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2667 && runparams.encoding != prev_encoding) {
2668 runparams.encoding = prev_encoding;
2669 if (!runparams.isFullUnicode())
2670 os << setEncoding(prev_encoding->iconvName());
2673 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2677 bool Paragraph::emptyTag() const
2679 for (pos_type i = 0; i < size(); ++i) {
2680 if (Inset const * inset = getInset(i)) {
2681 InsetCode lyx_code = inset->lyxCode();
2682 // FIXME testing like that is wrong. What is
2684 if (lyx_code != TOC_CODE &&
2685 lyx_code != INCLUDE_CODE &&
2686 lyx_code != GRAPHICS_CODE &&
2687 lyx_code != ERT_CODE &&
2688 lyx_code != LISTINGS_CODE &&
2689 lyx_code != FLOAT_CODE &&
2690 lyx_code != TABULAR_CODE) {
2694 char_type c = d->text_[i];
2695 if (c != ' ' && c != '\t')
2703 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2706 for (pos_type i = 0; i < size(); ++i) {
2707 if (Inset const * inset = getInset(i)) {
2708 InsetCode lyx_code = inset->lyxCode();
2709 if (lyx_code == LABEL_CODE) {
2710 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2711 docstring const & id = il->getParam("name");
2712 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2720 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2724 for (i = 0; i < size(); ++i) {
2725 if (Inset const * inset = getInset(i)) {
2726 inset->docbook(os, runparams);
2728 char_type c = d->text_[i];
2731 os << sgml::escapeChar(c);
2738 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2742 for (i = 0; i < size(); ++i) {
2743 if (Inset const * inset = getInset(i)) {
2744 inset->xhtml(xs, runparams);
2746 char_type c = d->text_[i];
2756 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2759 pos_type size = text_.size();
2760 for (pos_type i = initial; i < size; ++i) {
2761 Font font = owner_->getFont(buf.params(), i, outerfont);
2762 if (text_[i] == META_INSET)
2764 if (i != initial && font != font_old)
2773 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2775 OutputParams const & runparams,
2776 Font const & outerfont,
2777 pos_type initial) const
2779 bool emph_flag = false;
2781 Layout const & style = *d->layout_;
2783 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2785 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2788 // parsing main loop
2789 for (pos_type i = initial; i < size(); ++i) {
2790 Font font = getFont(buf.params(), i, outerfont);
2792 // handle <emphasis> tag
2793 if (font_old.emph() != font.fontInfo().emph()) {
2794 if (font.fontInfo().emph() == FONT_ON) {
2797 } else if (i != initial) {
2798 os << "</emphasis>";
2803 if (Inset const * inset = getInset(i)) {
2804 inset->docbook(os, runparams);
2806 char_type c = d->text_[i];
2808 if (style.pass_thru)
2811 os << sgml::escapeChar(c);
2813 font_old = font.fontInfo();
2817 os << "</emphasis>";
2820 if (style.free_spacing)
2822 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2828 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2829 vector<html::EndFontTag> & tagsToClose,
2830 bool & flag, FontState curstate, html::FontTypes type)
2832 if (curstate == FONT_ON) {
2833 tagsToOpen.push_back(html::FontTag(type));
2836 tagsToClose.push_back(html::EndFontTag(type));
2843 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2845 OutputParams const & runparams,
2846 Font const & outerfont,
2847 pos_type initial) const
2851 // track whether we have opened these tags
2852 bool emph_flag = false;
2853 bool bold_flag = false;
2854 bool noun_flag = false;
2855 bool ubar_flag = false;
2856 bool dbar_flag = false;
2857 bool sout_flag = false;
2858 bool wave_flag = false;
2860 bool shap_flag = false;
2862 bool faml_flag = false;
2864 bool size_flag = false;
2866 Layout const & style = *d->layout_;
2868 xs.startParagraph(allowEmpty());
2871 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2873 FontShape curr_fs = INHERIT_SHAPE;
2874 FontFamily curr_fam = INHERIT_FAMILY;
2875 FontSize curr_size = FONT_SIZE_INHERIT;
2877 string const default_family =
2878 buf.masterBuffer()->params().fonts_default_family;
2880 vector<html::FontTag> tagsToOpen;
2881 vector<html::EndFontTag> tagsToClose;
2883 // parsing main loop
2884 for (pos_type i = initial; i < size(); ++i) {
2885 // let's not show deleted material in the output
2889 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
2892 FontState curstate = font.fontInfo().emph();
2893 if (font_old.emph() != curstate)
2894 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
2897 curstate = font.fontInfo().noun();
2898 if (font_old.noun() != curstate)
2899 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
2902 curstate = font.fontInfo().underbar();
2903 if (font_old.underbar() != curstate)
2904 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
2907 curstate = font.fontInfo().strikeout();
2908 if (font_old.strikeout() != curstate)
2909 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
2912 curstate = font.fontInfo().uuline();
2913 if (font_old.uuline() != curstate)
2914 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
2917 curstate = font.fontInfo().uwave();
2918 if (font_old.uwave() != curstate)
2919 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
2922 // a little hackish, but allows us to reuse what we have.
2923 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
2924 if (font_old.series() != font.fontInfo().series())
2925 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
2928 curr_fs = font.fontInfo().shape();
2929 FontShape old_fs = font_old.shape();
2930 if (old_fs != curr_fs) {
2934 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
2937 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
2939 case SMALLCAPS_SHAPE:
2940 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
2946 // the other tags are for internal use
2954 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
2958 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
2961 case SMALLCAPS_SHAPE:
2962 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
2969 // the other tags are for internal use
2976 curr_fam = font.fontInfo().family();
2977 FontFamily old_fam = font_old.family();
2978 if (old_fam != curr_fam) {
2982 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
2985 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
2987 case TYPEWRITER_FAMILY:
2988 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
2990 case INHERIT_FAMILY:
2993 // the other tags are for internal use
3001 // we will treat a "default" font family as roman, since we have
3002 // no other idea what to do.
3003 if (default_family != "rmdefault" && default_family != "default") {
3004 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3009 if (default_family != "sfdefault") {
3010 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3014 case TYPEWRITER_FAMILY:
3015 if (default_family != "ttdefault") {
3016 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3020 case INHERIT_FAMILY:
3023 // the other tags are for internal use
3030 curr_size = font.fontInfo().size();
3031 FontSize old_size = font_old.size();
3032 if (old_size != curr_size) {
3035 case FONT_SIZE_TINY:
3036 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3038 case FONT_SIZE_SCRIPT:
3039 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3041 case FONT_SIZE_FOOTNOTE:
3042 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3044 case FONT_SIZE_SMALL:
3045 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3047 case FONT_SIZE_LARGE:
3048 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3050 case FONT_SIZE_LARGER:
3051 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3053 case FONT_SIZE_LARGEST:
3054 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3056 case FONT_SIZE_HUGE:
3057 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3059 case FONT_SIZE_HUGER:
3060 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3062 case FONT_SIZE_INCREASE:
3063 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3065 case FONT_SIZE_DECREASE:
3066 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3068 case FONT_SIZE_INHERIT:
3069 case FONT_SIZE_NORMAL:
3072 // the other tags are for internal use
3078 switch (curr_size) {
3079 case FONT_SIZE_TINY:
3080 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3083 case FONT_SIZE_SCRIPT:
3084 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3087 case FONT_SIZE_FOOTNOTE:
3088 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3091 case FONT_SIZE_SMALL:
3092 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3095 case FONT_SIZE_LARGE:
3096 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3099 case FONT_SIZE_LARGER:
3100 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3103 case FONT_SIZE_LARGEST:
3104 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3107 case FONT_SIZE_HUGE:
3108 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3111 case FONT_SIZE_HUGER:
3112 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3115 case FONT_SIZE_INCREASE:
3116 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3119 case FONT_SIZE_DECREASE:
3120 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3123 case FONT_SIZE_NORMAL:
3124 case FONT_SIZE_INHERIT:
3127 // the other tags are for internal use
3134 // Other such tags? What about the other text ranges?
3136 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3137 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3138 for (; cit != cen; ++cit)
3141 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3142 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3143 for (; sit != sen; ++sit)
3146 tagsToClose.clear();
3149 Inset const * inset = getInset(i);
3151 if (!runparams.for_toc || inset->isInToc()) {
3152 OutputParams np = runparams;
3153 np.local_font = &font;
3154 if (!inset->getLayout().htmlisblock())
3155 np.html_in_par = true;
3156 retval += inset->xhtml(xs, np);
3159 char_type c = getUChar(buf.masterBuffer()->params(), i);
3161 if (style.pass_thru || runparams.pass_thru)
3163 else if (c == '-') {
3166 if (j < size() && d->text_[j] == '-') {
3168 if (j < size() && d->text_[j] == '-') {
3169 str += from_ascii("—");
3172 str += from_ascii("–");
3178 // We don't want to escape the entities. Note that
3179 // it is safe to do this, since str can otherwise
3180 // only be "-". E.g., it can't be "<".
3181 xs << XHTMLStream::ESCAPE_NONE << str;
3185 font_old = font.fontInfo();
3194 bool Paragraph::isHfill(pos_type pos) const
3196 Inset const * inset = getInset(pos);
3197 return inset && (inset->lyxCode() == SPACE_CODE &&
3198 inset->isStretchableSpace());
3202 bool Paragraph::isNewline(pos_type pos) const
3204 Inset const * inset = getInset(pos);
3205 return inset && inset->lyxCode() == NEWLINE_CODE;
3209 bool Paragraph::isEnvSeparator(pos_type pos) const
3211 Inset const * inset = getInset(pos);
3212 return inset && inset->lyxCode() == SEPARATOR_CODE;
3216 bool Paragraph::isLineSeparator(pos_type pos) const
3218 char_type const c = d->text_[pos];
3219 if (isLineSeparatorChar(c))
3221 Inset const * inset = getInset(pos);
3222 return inset && inset->isLineSeparator();
3226 bool Paragraph::isWordSeparator(pos_type pos) const
3230 if (Inset const * inset = getInset(pos))
3231 return !inset->isLetter();
3232 // if we have a hard hyphen (no en- or emdash) or apostrophe
3233 // we pass this to the spell checker
3234 // FIXME: this method is subject to change, visit
3235 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3236 // to get an impression how complex this is.
3237 if (isHardHyphenOrApostrophe(pos))
3239 char_type const c = d->text_[pos];
3240 // We want to pass the escape chars to the spellchecker
3241 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3242 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3246 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3248 pos_type const psize = size();
3251 char_type const c = d->text_[pos];
3252 if (c != '-' && c != '\'')
3254 int nextpos = pos + 1;
3255 int prevpos = pos > 0 ? pos - 1 : 0;
3256 if ((nextpos == psize || isSpace(nextpos))
3257 && (pos == 0 || isSpace(prevpos)))
3260 || ((nextpos == psize || d->text_[nextpos] != '-')
3261 && (pos == 0 || d->text_[prevpos] != '-'));
3265 bool Paragraph::isSameSpellRange(pos_type pos1, pos_type pos2) const
3268 || d->speller_state_.getRange(pos1) == d->speller_state_.getRange(pos2);
3272 bool Paragraph::isChar(pos_type pos) const
3274 if (Inset const * inset = getInset(pos))
3275 return inset->isChar();
3276 char_type const c = d->text_[pos];
3277 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3281 bool Paragraph::isSpace(pos_type pos) const
3283 if (Inset const * inset = getInset(pos))
3284 return inset->isSpace();
3285 char_type const c = d->text_[pos];
3286 return lyx::isSpace(c);
3291 Paragraph::getParLanguage(BufferParams const & bparams) const
3294 return getFirstFontSettings(bparams).language();
3295 // FIXME: we should check the prev par as well (Lgb)
3296 return bparams.language;
3300 bool Paragraph::isRTL(BufferParams const & bparams) const
3302 return lyxrc.rtl_support
3303 && getParLanguage(bparams)->rightToLeft()
3304 && !inInset().getLayout().forceLTR();
3308 void Paragraph::changeLanguage(BufferParams const & bparams,
3309 Language const * from, Language const * to)
3311 // change language including dummy font change at the end
3312 for (pos_type i = 0; i <= size(); ++i) {
3313 Font font = getFontSettings(bparams, i);
3314 if (font.language() == from) {
3315 font.setLanguage(to);
3317 d->requestSpellCheck(i);
3323 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3325 Language const * doc_language = bparams.language;
3326 FontList::const_iterator cit = d->fontlist_.begin();
3327 FontList::const_iterator end = d->fontlist_.end();
3329 for (; cit != end; ++cit)
3330 if (cit->font().language() != ignore_language &&
3331 cit->font().language() != latex_language &&
3332 cit->font().language() != doc_language)
3338 void Paragraph::getLanguages(std::set<Language const *> & languages) const
3340 FontList::const_iterator cit = d->fontlist_.begin();
3341 FontList::const_iterator end = d->fontlist_.end();
3343 for (; cit != end; ++cit) {
3344 Language const * lang = cit->font().language();
3345 if (lang != ignore_language &&
3346 lang != latex_language)
3347 languages.insert(lang);
3352 docstring Paragraph::asString(int options) const
3354 return asString(0, size(), options);
3358 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3360 odocstringstream os;
3363 && options & AS_STR_LABEL
3364 && !d->params_.labelString().empty())
3365 os << d->params_.labelString() << ' ';
3367 for (pos_type i = beg; i < end; ++i) {
3368 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3370 char_type const c = d->text_[i];
3371 if (isPrintable(c) || c == '\t'
3372 || (c == '\n' && (options & AS_STR_NEWLINES)))
3374 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3375 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3376 LASSERT(runparams != 0, return docstring());
3377 getInset(i)->plaintext(os, *runparams);
3379 getInset(i)->toString(os);
3380 if (getInset(i)->asInsetMath())
3390 void Paragraph::forOutliner(docstring & os, size_t maxlen) const
3392 if (!d->params_.labelString().empty())
3393 os += d->params_.labelString() + ' ';
3394 for (pos_type i = 0; i < size() && os.length() < maxlen; ++i) {
3397 char_type const c = d->text_[i];
3400 else if (c == '\t' || c == '\n')
3402 else if (c == META_INSET)
3403 getInset(i)->forOutliner(os, maxlen);
3408 void Paragraph::setInsetOwner(Inset const * inset)
3410 d->inset_owner_ = inset;
3414 int Paragraph::id() const
3420 void Paragraph::setId(int id)
3426 Layout const & Paragraph::layout() const
3432 void Paragraph::setLayout(Layout const & layout)
3434 d->layout_ = &layout;
3438 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3440 setLayout(tc.defaultLayout());
3444 void Paragraph::setPlainLayout(DocumentClass const & tc)
3446 setLayout(tc.plainLayout());
3450 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3452 if (usePlainLayout())
3453 setPlainLayout(tclass);
3455 setDefaultLayout(tclass);
3459 Inset const & Paragraph::inInset() const
3461 LBUFERR(d->inset_owner_);
3462 return *d->inset_owner_;
3466 ParagraphParameters & Paragraph::params()
3472 ParagraphParameters const & Paragraph::params() const
3478 bool Paragraph::isFreeSpacing() const
3480 if (d->layout_->free_spacing)
3482 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3486 bool Paragraph::allowEmpty() const
3488 if (d->layout_->keepempty)
3490 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3494 char_type Paragraph::transformChar(char_type c, pos_type pos) const
3496 if (!Encodings::isArabicChar(c))
3499 char_type prev_char = ' ';
3500 char_type next_char = ' ';
3502 for (pos_type i = pos - 1; i >= 0; --i) {
3503 char_type const par_char = d->text_[i];
3504 if (!Encodings::isArabicComposeChar(par_char)) {
3505 prev_char = par_char;
3510 for (pos_type i = pos + 1, end = size(); i < end; ++i) {
3511 char_type const par_char = d->text_[i];
3512 if (!Encodings::isArabicComposeChar(par_char)) {
3513 next_char = par_char;
3518 if (Encodings::isArabicChar(next_char)) {
3519 if (Encodings::isArabicChar(prev_char) &&
3520 !Encodings::isArabicSpecialChar(prev_char))
3521 return Encodings::transformChar(c, Encodings::FORM_MEDIAL);
3523 return Encodings::transformChar(c, Encodings::FORM_INITIAL);
3525 if (Encodings::isArabicChar(prev_char) &&
3526 !Encodings::isArabicSpecialChar(prev_char))
3527 return Encodings::transformChar(c, Encodings::FORM_FINAL);
3529 return Encodings::transformChar(c, Encodings::FORM_ISOLATED);
3534 bool Paragraph::brokenBiblio() const
3536 // there is a problem if there is no bibitem at position 0 or
3537 // if there is another bibitem in the paragraph.
3538 return d->layout_->labeltype == LABEL_BIBLIO
3539 && (d->insetlist_.find(BIBITEM_CODE) != 0
3540 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3544 int Paragraph::fixBiblio(Buffer const & buffer)
3546 // FIXME: What about the case where paragraph is not BIBLIO
3547 // but there is an InsetBibitem?
3548 // FIXME: when there was already an inset at 0, the return value is 1,
3549 // which does not tell whether another inset has been remove; the
3550 // cursor cannot be correctly updated.
3552 if (d->layout_->labeltype != LABEL_BIBLIO)
3555 bool const track_changes = buffer.params().track_changes;
3556 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3557 bool const hasbibitem0 = bibitem_pos == 0;
3560 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3561 // There was an InsetBibitem at pos 0, and no other one => OK
3562 if (bibitem_pos == -1)
3564 // there is a bibitem at the 0 position, but since
3565 // there is a second one, we copy the second on the
3566 // first. We're assuming there are at most two of
3567 // these, which there should be.
3568 // FIXME: why does it make sense to do that rather
3569 // than keep the first? (JMarc)
3570 Inset * inset = releaseInset(bibitem_pos);
3571 d->insetlist_.begin()->inset = inset;
3572 return -bibitem_pos;
3575 // We need to create an inset at the beginning
3577 if (bibitem_pos > 0) {
3578 // there was one somewhere in the paragraph, let's move it
3579 inset = d->insetlist_.release(bibitem_pos);
3580 eraseChar(bibitem_pos, track_changes);
3583 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3584 InsetCommandParams(BIBITEM_CODE));
3586 Font font(inherit_font, buffer.params().language);
3587 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3588 : Change::UNCHANGED));
3594 void Paragraph::checkAuthors(AuthorList const & authorList)
3596 d->changes_.checkAuthors(authorList);
3600 bool Paragraph::isChanged(pos_type pos) const
3602 return lookupChange(pos).changed();
3606 bool Paragraph::isInserted(pos_type pos) const
3608 return lookupChange(pos).inserted();
3612 bool Paragraph::isDeleted(pos_type pos) const
3614 return lookupChange(pos).deleted();
3618 InsetList const & Paragraph::insetList() const
3620 return d->insetlist_;
3624 void Paragraph::setBuffer(Buffer & b)
3626 d->insetlist_.setBuffer(b);
3630 Inset * Paragraph::releaseInset(pos_type pos)
3632 Inset * inset = d->insetlist_.release(pos);
3633 /// does not honour change tracking!
3634 eraseChar(pos, false);
3639 Inset * Paragraph::getInset(pos_type pos)
3641 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3642 ? d->insetlist_.get(pos) : 0;
3646 Inset const * Paragraph::getInset(pos_type pos) const
3648 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3649 ? d->insetlist_.get(pos) : 0;
3653 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3654 pos_type & right, TextCase action)
3656 // process sequences of modified characters; in change
3657 // tracking mode, this approach results in much better
3658 // usability than changing case on a char-by-char basis
3659 // We also need to track the current font, since font
3660 // changes within sequences can occur.
3661 vector<pair<char_type, Font> > changes;
3663 bool const trackChanges = bparams.track_changes;
3665 bool capitalize = true;
3667 for (; pos < right; ++pos) {
3668 char_type oldChar = d->text_[pos];
3669 char_type newChar = oldChar;
3671 // ignore insets and don't play with deleted text!
3672 if (oldChar != META_INSET && !isDeleted(pos)) {
3674 case text_lowercase:
3675 newChar = lowercase(oldChar);
3677 case text_capitalization:
3679 newChar = uppercase(oldChar);
3683 case text_uppercase:
3684 newChar = uppercase(oldChar);
3689 if (isWordSeparator(pos) || isDeleted(pos)) {
3690 // permit capitalization again
3694 if (oldChar != newChar) {
3695 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3696 if (pos != right - 1)
3698 // step behind the changing area
3702 int erasePos = pos - changes.size();
3703 for (size_t i = 0; i < changes.size(); i++) {
3704 insertChar(pos, changes[i].first,
3707 if (!eraseChar(erasePos, trackChanges)) {
3710 ++right; // expand selection
3718 int Paragraph::find(docstring const & str, bool cs, bool mw,
3719 pos_type start_pos, bool del) const
3721 pos_type pos = start_pos;
3722 int const strsize = str.length();
3724 pos_type const parsize = d->text_.size();
3725 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3726 // Ignore "invisible" letters such as ligature breaks
3727 // and hyphenation chars while searching
3728 while (pos < parsize - 1 && isInset(pos)) {
3729 odocstringstream os;
3730 getInset(pos)->toString(os);
3731 if (!getInset(pos)->isLetter() || !os.str().empty())
3735 if (cs && str[i] != d->text_[pos])
3737 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3739 if (!del && isDeleted(pos))
3746 // if necessary, check whether string matches word
3748 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3751 && !isWordSeparator(pos))
3755 return pos - start_pos;
3759 char_type Paragraph::getChar(pos_type pos) const
3761 return d->text_[pos];
3765 pos_type Paragraph::size() const
3767 return d->text_.size();
3771 bool Paragraph::empty() const
3773 return d->text_.empty();
3777 bool Paragraph::isInset(pos_type pos) const
3779 return d->text_[pos] == META_INSET;
3783 bool Paragraph::isSeparator(pos_type pos) const
3785 //FIXME: Are we sure this can be the only separator?
3786 return d->text_[pos] == ' ';
3790 void Paragraph::deregisterWords()
3792 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3793 Private::LangWordsMap::const_iterator ite = d->words_.end();
3794 for (; itl != ite; ++itl) {
3795 WordList * wl = theWordList(itl->first);
3796 Private::Words::const_iterator it = (itl->second).begin();
3797 Private::Words::const_iterator et = (itl->second).end();
3798 for (; it != et; ++it)
3805 void Paragraph::locateWord(pos_type & from, pos_type & to,
3806 word_location const loc) const
3809 case WHOLE_WORD_STRICT:
3810 if (from == 0 || from == size()
3811 || isWordSeparator(from)
3812 || isWordSeparator(from - 1)) {
3816 // no break here, we go to the next
3819 // If we are already at the beginning of a word, do nothing
3820 if (!from || isWordSeparator(from - 1))
3822 // no break here, we go to the next
3825 // always move the cursor to the beginning of previous word
3826 while (from && !isWordSeparator(from - 1))
3830 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3833 // no need to move the 'from' cursor
3837 while (to < size() && !isWordSeparator(to))
3842 void Paragraph::collectWords()
3844 for (pos_type pos = 0; pos < size(); ++pos) {
3845 if (isWordSeparator(pos))
3847 pos_type from = pos;
3848 locateWord(from, pos, WHOLE_WORD);
3849 if (pos < from + lyxrc.completion_minlength)
3851 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
3852 if (cit == d->fontlist_.end())
3854 Language const * lang = cit->font().language();
3855 docstring const word = asString(from, pos, AS_STR_NONE);
3856 d->words_[lang->lang()].insert(word);
3861 void Paragraph::registerWords()
3863 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3864 Private::LangWordsMap::const_iterator ite = d->words_.end();
3865 for (; itl != ite; ++itl) {
3866 WordList * wl = theWordList(itl->first);
3867 Private::Words::const_iterator it = (itl->second).begin();
3868 Private::Words::const_iterator et = (itl->second).end();
3869 for (; it != et; ++it)
3875 void Paragraph::updateWords()
3883 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3885 SkipPositionsIterator begin = skips.begin();
3886 SkipPositions::iterator end = skips.end();
3887 if (pos > 0 && begin < end) {
3889 if (end->last == pos - 1) {
3894 skips.insert(end, FontSpan(pos, pos));
3898 Language * Paragraph::Private::locateSpellRange(
3899 pos_type & from, pos_type & to,
3900 SkipPositions & skips) const
3902 // skip leading white space
3903 while (from < to && owner_->isWordSeparator(from))
3905 // don't check empty range
3908 // get current language
3909 Language * lang = getSpellLanguage(from);
3910 pos_type last = from;
3911 bool samelang = true;
3912 bool sameinset = true;
3913 while (last < to && samelang && sameinset) {
3914 // hop to end of word
3915 while (last < to && !owner_->isWordSeparator(last)) {
3916 if (owner_->getInset(last)) {
3917 appendSkipPosition(skips, last);
3918 } else if (owner_->isDeleted(last)) {
3919 appendSkipPosition(skips, last);
3923 // hop to next word while checking for insets
3924 while (sameinset && last < to && owner_->isWordSeparator(last)) {
3925 if (Inset const * inset = owner_->getInset(last))
3926 sameinset = inset->isChar() && inset->isLetter();
3927 if (sameinset && owner_->isDeleted(last)) {
3928 appendSkipPosition(skips, last);
3933 if (sameinset && last < to) {
3934 // now check for language change
3935 samelang = lang == getSpellLanguage(last);
3938 // if language change detected backstep is needed
3946 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
3949 const_cast<Language *>(owner_->getFontSettings(
3950 inset_owner_->buffer().params(), from).language());
3951 if (lang == inset_owner_->buffer().params().language
3952 && !lyxrc.spellchecker_alt_lang.empty()) {
3954 string const lang_variety =
3955 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3956 lang->setCode(lang_code);
3957 lang->setVariety(lang_variety);
3963 void Paragraph::requestSpellCheck(pos_type pos)
3965 d->requestSpellCheck(pos);
3969 bool Paragraph::needsSpellCheck() const
3971 SpellChecker::ChangeNumber speller_change_number = 0;
3972 if (theSpellChecker())
3973 speller_change_number = theSpellChecker()->changeNumber();
3974 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
3975 d->speller_state_.needsCompleteRefresh(speller_change_number);
3977 return d->needsSpellCheck();
3981 bool Paragraph::Private::ignoreWord(docstring const & word) const
3983 // Ignore words with digits
3984 // FIXME: make this customizable
3985 // (note that some checkers ignore words with digits by default)
3986 docstring::const_iterator cit = word.begin();
3987 docstring::const_iterator const end = word.end();
3988 for (; cit != end; ++cit) {
3989 if (isNumber((*cit)))
3996 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
3997 WordLangTuple & wl, docstring_list & suggestions,
3998 bool do_suggestion, bool check_learned) const
4000 SpellChecker::Result result = SpellChecker::WORD_OK;
4001 SpellChecker * speller = theSpellChecker();
4005 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4008 locateWord(from, to, WHOLE_WORD);
4009 if (from == to || from >= size())
4012 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4013 Language * lang = d->getSpellLanguage(from);
4015 wl = WordLangTuple(word, lang);
4020 if (needsSpellCheck() || check_learned) {
4022 if (!d->ignoreWord(word)) {
4023 bool const trailing_dot = to < size() && d->text_[to] == '.';
4024 result = speller->check(wl);
4025 if (SpellChecker::misspelled(result) && trailing_dot) {
4026 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4027 result = speller->check(wl);
4028 if (!SpellChecker::misspelled(result)) {
4029 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4031 from << ".." << to << "]");
4033 // spell check with dot appended failed too
4034 // restore original word/lang value
4035 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4036 wl = WordLangTuple(word, lang);
4040 if (!SpellChecker::misspelled(result)) {
4041 // area up to the begin of the next word is not misspelled
4042 while (end < size() && isWordSeparator(end))
4045 d->setMisspelled(from, end, result);
4047 result = d->speller_state_.getState(from);
4051 suggestions.clear();
4053 if (SpellChecker::misspelled(result)) {
4054 LYXERR(Debug::GUI, "misspelled word: \"" <<
4056 from << ".." << to << "]");
4058 speller->suggest(wl, suggestions);
4064 void Paragraph::Private::markMisspelledWords(
4065 pos_type const & first, pos_type const & last,
4066 SpellChecker::Result result,
4067 docstring const & word,
4068 SkipPositions const & skips)
4070 if (!SpellChecker::misspelled(result)) {
4071 setMisspelled(first, last, SpellChecker::WORD_OK);
4075 SpellChecker * speller = theSpellChecker();
4076 // locate and enumerate the error positions
4077 int nerrors = speller->numMisspelledWords();
4079 SkipPositionsIterator it = skips.begin();
4080 SkipPositionsIterator et = skips.end();
4081 for (int index = 0; index < nerrors; ++index) {
4084 speller->misspelledWord(index, wstart, wlen);
4085 /// should not happen if speller supports range checks
4086 if (!wlen) continue;
4087 docstring const misspelled = word.substr(wstart, wlen);
4088 wstart += first + numskipped;
4089 if (snext < wstart) {
4090 /// mark the range of correct spelling
4091 numskipped += countSkips(it, et, wstart);
4092 setMisspelled(snext,
4093 wstart - 1, SpellChecker::WORD_OK);
4095 snext = wstart + wlen;
4096 numskipped += countSkips(it, et, snext);
4097 /// mark the range of misspelling
4098 setMisspelled(wstart, snext, result);
4099 LYXERR(Debug::GUI, "misspelled word: \"" <<
4100 misspelled << "\" [" <<
4101 wstart << ".." << (snext-1) << "]");
4104 if (snext <= last) {
4105 /// mark the range of correct spelling at end
4106 setMisspelled(snext, last, SpellChecker::WORD_OK);
4111 void Paragraph::spellCheck() const
4113 SpellChecker * speller = theSpellChecker();
4114 if (!speller || empty() ||!needsSpellCheck())
4118 d->rangeOfSpellCheck(start, endpos);
4119 if (speller->canCheckParagraph()) {
4120 // loop until we leave the range
4121 for (pos_type first = start; first < endpos; ) {
4122 pos_type last = endpos;
4123 Private::SkipPositions skips;
4124 Language * lang = d->locateSpellRange(first, last, skips);
4125 if (first >= endpos)
4127 // start the spell checker on the unit of meaning
4128 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4129 WordLangTuple wl = WordLangTuple(word, lang);
4130 SpellChecker::Result result = word.size() ?
4131 speller->check(wl) : SpellChecker::WORD_OK;
4132 d->markMisspelledWords(first, last, result, word, skips);
4136 static docstring_list suggestions;
4137 pos_type to = endpos;
4138 while (start < endpos) {
4140 spellCheck(start, to, wl, suggestions, false);
4144 d->readySpellCheck();
4148 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4150 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4151 if (result || pos <= 0 || pos > size())
4153 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4154 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4159 string Paragraph::magicLabel() const
4162 ss << "magicparlabel-" << id();