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 contains(pos_type pos) const { return range_.contains(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 //FIXME: is this the same as !range_.intersect(r).empty() ?
108 return range_.contains(r.first) || range_.contains(r.last) ||
109 r.contains(range_.first) || r.contains(range_.last);
112 void shift(pos_type pos, int offset)
114 if (range_.first > pos) {
115 range_.first += offset;
116 range_.last += offset;
117 } else if (range_.last >= pos) {
118 range_.last += offset;
123 SpellChecker::Result result_ ;
127 /////////////////////////////////////////////////////////////////////
131 /////////////////////////////////////////////////////////////////////
133 class SpellCheckerState {
137 needs_refresh_ = true;
138 current_change_number_ = 0;
141 void setRange(FontSpan const & fp, SpellChecker::Result state)
144 RangesIterator et = ranges_.end();
145 RangesIterator it = ranges_.begin();
146 for (; it != et; ++it) {
147 if (!it->covered(fp))
148 result.push_back(SpellResultRange(it->range(), it->result()));
149 else if (state == SpellChecker::WORD_OK) {
150 // trim or split the current misspelled range
151 // store misspelled ranges only
152 FontSpan range = it->range();
153 if (fp.first > range.first) {
154 // misspelled area in front of WORD_OK
155 range.last = fp.first - 1;
156 result.push_back(SpellResultRange(range, it->result()));
159 if (fp.last < range.last) {
160 // misspelled area after WORD_OK range
161 range.first = fp.last + 1;
162 result.push_back(SpellResultRange(range, it->result()));
167 if (state != SpellChecker::WORD_OK)
168 ranges_.push_back(SpellResultRange(fp, state));
171 void increasePosAfterPos(pos_type pos)
173 correctRangesAfterPos(pos, 1);
177 void decreasePosAfterPos(pos_type pos)
179 correctRangesAfterPos(pos, -1);
183 void refreshLast(pos_type pos)
185 if (pos < refresh_.last)
189 SpellChecker::Result getState(pos_type pos) const
191 SpellChecker::Result result = SpellChecker::WORD_OK;
192 RangesIterator et = ranges_.end();
193 RangesIterator it = ranges_.begin();
194 for (; it != et; ++it) {
195 if(it->contains(pos)) {
202 FontSpan const & getRange(pos_type pos) const
204 /// empty span to indicate mismatch
205 static FontSpan empty_;
206 RangesIterator et = ranges_.end();
207 RangesIterator it = ranges_.begin();
208 for (; it != et; ++it) {
209 if(it->contains(pos)) {
216 bool needsRefresh() const
218 return needs_refresh_;
221 SpellChecker::ChangeNumber currentChangeNumber() const
223 return current_change_number_;
226 void refreshRange(pos_type & first, pos_type & last) const
228 first = refresh_.first;
229 last = refresh_.last;
232 void needsRefresh(pos_type pos)
234 if (needs_refresh_ && pos != -1) {
235 if (pos < refresh_.first)
236 refresh_.first = pos;
237 if (pos > refresh_.last)
239 } else if (pos != -1) {
240 // init request check for neighbour positions too
241 refresh_.first = pos > 0 ? pos - 1 : 0;
242 // no need for special end of paragraph check
243 refresh_.last = pos + 1;
245 needs_refresh_ = pos != -1;
248 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
250 needs_refresh_ = true;
253 current_change_number_ = change_number;
256 typedef vector<SpellResultRange> Ranges;
257 typedef Ranges::const_iterator RangesIterator;
259 /// the area of the paragraph with pending spell check
262 /// spell state cache version number
263 SpellChecker::ChangeNumber current_change_number_;
266 void correctRangesAfterPos(pos_type pos, int offset)
268 RangesIterator et = ranges_.end();
269 Ranges::iterator it = ranges_.begin();
270 for (; it != et; ++it) {
271 it->shift(pos, offset);
277 /////////////////////////////////////////////////////////////////////
279 // Paragraph::Private
281 /////////////////////////////////////////////////////////////////////
283 class Paragraph::Private
285 // Enforce our own "copy" constructor by declaring the standard one and
286 // the assignment operator private without implementing them.
287 Private(Private const &);
288 Private & operator=(Private const &);
291 Private(Paragraph * owner, Layout const & layout);
292 /// "Copy constructor"
293 Private(Private const &, Paragraph * owner);
294 /// Copy constructor from \p beg to \p end
295 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
298 void insertChar(pos_type pos, char_type c, Change const & change);
300 /// Output the surrogate pair formed by \p c and \p next to \p os.
301 /// \return the number of characters written.
302 int latexSurrogatePair(otexstream & os, char_type c, char_type next,
303 OutputParams const &);
305 /// Output a space in appropriate formatting (or a surrogate pair
306 /// if the next character is a combining character).
307 /// \return whether a surrogate pair was output.
308 bool simpleTeXBlanks(OutputParams const &,
311 unsigned int & column,
313 Layout const & style);
315 /// Output consecutive unicode chars, belonging to the same script as
316 /// specified by the latex macro \p ltx, to \p os starting from \p i.
317 /// \return the number of characters written.
318 int writeScriptChars(otexstream & os, docstring const & ltx,
319 Change const &, Encoding const &, pos_type & i);
321 /// This could go to ParagraphParameters if we want to.
322 int startTeXParParams(BufferParams const &, otexstream &,
323 OutputParams const &) const;
325 /// This could go to ParagraphParameters if we want to.
326 bool endTeXParParams(BufferParams const &, otexstream &,
327 OutputParams const &) const;
330 void latexInset(BufferParams const &,
335 Font const & outerfont,
337 Change & running_change,
338 Layout const & style,
340 unsigned int & column);
343 void latexSpecialChar(
345 BufferParams const & bparams,
346 OutputParams const & runparams,
347 Font const & running_font,
348 Change const & running_change,
349 Layout const & style,
352 unsigned int & column);
359 unsigned int & column);
365 unsigned int & column);
368 void validate(LaTeXFeatures & features) const;
370 /// Checks if the paragraph contains only text and no inset or font change.
371 bool onlyText(Buffer const & buf, Font const & outerfont,
372 pos_type initial) const;
374 /// a vector of speller skip positions
375 typedef vector<FontSpan> SkipPositions;
376 typedef SkipPositions::const_iterator SkipPositionsIterator;
378 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
380 Language * getSpellLanguage(pos_type const from) const;
382 Language * locateSpellRange(pos_type & from, pos_type & to,
383 SkipPositions & skips) const;
385 bool hasSpellerChange() const
387 SpellChecker::ChangeNumber speller_change_number = 0;
388 if (theSpellChecker())
389 speller_change_number = theSpellChecker()->changeNumber();
390 return speller_change_number > speller_state_.currentChangeNumber();
393 bool ignoreWord(docstring const & word) const ;
395 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
397 pos_type textsize = owner_->size();
398 // check for sane arguments
399 if (to <= from || from >= textsize)
401 FontSpan fp = FontSpan(from, to - 1);
402 speller_state_.setRange(fp, state);
405 void requestSpellCheck(pos_type pos)
408 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
410 speller_state_.needsRefresh(pos);
413 void readySpellCheck()
415 speller_state_.needsRefresh(-1);
418 bool needsSpellCheck() const
420 return speller_state_.needsRefresh();
423 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
425 speller_state_.refreshRange(first, last);
427 last = owner_->size();
430 pos_type endpos = last;
431 owner_->locateWord(first, endpos, WHOLE_WORD);
434 owner_->locateWord(last, endpos, WHOLE_WORD);
439 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
443 while (it != et && it->first < start) {
444 int skip = it->last - it->first + 1;
452 void markMisspelledWords(pos_type const & first, pos_type const & last,
453 SpellChecker::Result result,
454 docstring const & word,
455 SkipPositions const & skips);
457 InsetCode ownerCode() const
459 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
462 /// Which Paragraph owns us?
466 Inset const * inset_owner_;
475 ParagraphParameters params_;
477 /// for recording and looking up changes
481 InsetList insetlist_;
484 pos_type begin_of_body_;
486 typedef docstring TextContainer;
490 typedef set<docstring> Words;
491 typedef map<string, Words> LangWordsMap;
495 Layout const * layout_;
497 SpellCheckerState speller_state_;
501 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
502 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
508 // Initialization of the counter for the paragraph id's,
510 // FIXME: There should be a more intelligent way to generate and use the
511 // paragraph ids per buffer instead a global static counter for all InsetText
512 // in the running program.
513 static int paragraph_id = -1;
515 Paragraph::Private::Private(Private const & p, Paragraph * owner)
516 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
517 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
518 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
521 id_ = ++paragraph_id;
522 requestSpellCheck(p.text_.size());
526 Paragraph::Private::Private(Private const & p, Paragraph * owner,
527 pos_type beg, pos_type end)
528 : owner_(owner), inset_owner_(p.inset_owner_),
529 params_(p.params_), changes_(p.changes_),
530 insetlist_(p.insetlist_, beg, end),
531 begin_of_body_(p.begin_of_body_), words_(p.words_),
534 id_ = ++paragraph_id;
535 if (beg >= pos_type(p.text_.size()))
537 text_ = p.text_.substr(beg, end - beg);
539 FontList::const_iterator fcit = fontlist_.begin();
540 FontList::const_iterator fend = fontlist_.end();
541 for (; fcit != fend; ++fcit) {
542 if (fcit->pos() < beg)
544 if (fcit->pos() >= end) {
545 // Add last entry in the fontlist_.
546 fontlist_.set(text_.size() - 1, fcit->font());
549 // Add a new entry in the fontlist_.
550 fontlist_.set(fcit->pos() - beg, fcit->font());
552 requestSpellCheck(p.text_.size());
556 void Paragraph::addChangesToToc(DocIterator const & cdit,
557 Buffer const & buf, bool output_active) const
559 d->changes_.addToToc(cdit, buf, output_active);
563 bool Paragraph::isDeleted(pos_type start, pos_type end) const
565 LASSERT(start >= 0 && start <= size(), return false);
566 LASSERT(end > start && end <= size() + 1, return false);
568 return d->changes_.isDeleted(start, end);
572 bool Paragraph::isChanged(pos_type start, pos_type end) const
574 LASSERT(start >= 0 && start <= size(), return false);
575 LASSERT(end > start && end <= size() + 1, return false);
577 return d->changes_.isChanged(start, end);
581 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
583 // keep the logic here in sync with the logic of eraseChars()
587 Change const & change = d->changes_.lookup(size());
588 return change.inserted() && change.currentAuthor();
592 void Paragraph::setChange(Change const & change)
594 // beware of the imaginary end-of-par character!
595 d->changes_.set(change, 0, size() + 1);
598 * Propagate the change recursively - but not in case of DELETED!
600 * Imagine that your co-author makes changes in an existing inset. He
601 * sends your document to you and you come to the conclusion that the
602 * inset should go completely. If you erase it, LyX must not delete all
603 * text within the inset. Otherwise, the change tracked insertions of
604 * your co-author get lost and there is no way to restore them later.
606 * Conclusion: An inset's content should remain untouched if you delete it
609 if (!change.deleted()) {
610 for (pos_type pos = 0; pos < size(); ++pos) {
611 if (Inset * inset = getInset(pos))
612 inset->setChange(change);
618 void Paragraph::setChange(pos_type pos, Change const & change)
620 LASSERT(pos >= 0 && pos <= size(), return);
621 d->changes_.set(change, pos);
623 // see comment in setChange(Change const &) above
624 if (!change.deleted() && pos < size())
625 if (Inset * inset = getInset(pos))
626 inset->setChange(change);
630 Change const & Paragraph::lookupChange(pos_type pos) const
632 LBUFERR(pos >= 0 && pos <= size());
633 return d->changes_.lookup(pos);
637 void Paragraph::acceptChanges(pos_type start, pos_type end)
639 LASSERT(start >= 0 && start <= size(), return);
640 LASSERT(end > start && end <= size() + 1, return);
642 for (pos_type pos = start; pos < end; ++pos) {
643 switch (lookupChange(pos).type) {
644 case Change::UNCHANGED:
645 // accept changes in nested inset
646 if (Inset * inset = getInset(pos))
647 inset->acceptChanges();
650 case Change::INSERTED:
651 d->changes_.set(Change(Change::UNCHANGED), pos);
652 // also accept changes in nested inset
653 if (Inset * inset = getInset(pos))
654 inset->acceptChanges();
657 case Change::DELETED:
658 // Suppress access to non-existent
659 // "end-of-paragraph char"
661 eraseChar(pos, false);
672 void Paragraph::rejectChanges(pos_type start, pos_type end)
674 LASSERT(start >= 0 && start <= size(), return);
675 LASSERT(end > start && end <= size() + 1, return);
677 for (pos_type pos = start; pos < end; ++pos) {
678 switch (lookupChange(pos).type) {
679 case Change::UNCHANGED:
680 // reject changes in nested inset
681 if (Inset * inset = getInset(pos))
682 inset->rejectChanges();
685 case Change::INSERTED:
686 // Suppress access to non-existent
687 // "end-of-paragraph char"
689 eraseChar(pos, false);
695 case Change::DELETED:
696 d->changes_.set(Change(Change::UNCHANGED), pos);
698 // Do NOT reject changes within a deleted inset!
699 // There may be insertions of a co-author inside of it!
707 void Paragraph::Private::insertChar(pos_type pos, char_type c,
708 Change const & change)
710 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
713 changes_.insert(change, pos);
715 // This is actually very common when parsing buffers (and
716 // maybe inserting ascii text)
717 if (pos == pos_type(text_.size())) {
718 // when appending characters, no need to update tables
720 // but we want spell checking
721 requestSpellCheck(pos);
725 text_.insert(text_.begin() + pos, c);
727 // Update the font table.
728 fontlist_.increasePosAfterPos(pos);
731 insetlist_.increasePosAfterPos(pos);
733 // Update list of misspelled positions
734 speller_state_.increasePosAfterPos(pos);
738 bool Paragraph::insertInset(pos_type pos, Inset * inset,
739 Font const & font, Change const & change)
741 LASSERT(inset, return false);
742 LASSERT(pos >= 0 && pos <= size(), return false);
744 // Paragraph::insertInset() can be used in cut/copy/paste operation where
745 // d->inset_owner_ is not set yet.
746 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
749 d->insertChar(pos, META_INSET, change);
750 LASSERT(d->text_[pos] == META_INSET, return false);
752 // Add a new entry in the insetlist_.
753 d->insetlist_.insert(inset, pos);
755 // Some insets require run of spell checker
756 requestSpellCheck(pos);
762 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
764 LASSERT(pos >= 0 && pos <= size(), return false);
766 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
769 Change change = d->changes_.lookup(pos);
771 // set the character to DELETED if
772 // a) it was previously unchanged or
773 // b) it was inserted by a co-author
775 if (!change.changed() ||
776 (change.inserted() && !change.currentAuthor())) {
777 setChange(pos, Change(Change::DELETED));
778 // request run of spell checker
779 requestSpellCheck(pos);
783 if (change.deleted())
787 // Don't physically access the imaginary end-of-paragraph character.
788 // eraseChar() can only mark it as DELETED. A physical deletion of
789 // end-of-par must be handled externally.
795 d->changes_.erase(pos);
797 // if it is an inset, delete the inset entry
798 if (d->text_[pos] == META_INSET)
799 d->insetlist_.erase(pos);
801 d->text_.erase(d->text_.begin() + pos);
803 // Update the fontlist_
804 d->fontlist_.erase(pos);
806 // Update the insetlist_
807 d->insetlist_.decreasePosAfterPos(pos);
809 // Update list of misspelled positions
810 d->speller_state_.decreasePosAfterPos(pos);
811 d->speller_state_.refreshLast(size());
817 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
819 LASSERT(start >= 0 && start <= size(), return 0);
820 LASSERT(end >= start && end <= size() + 1, return 0);
823 for (pos_type count = end - start; count; --count) {
824 if (!eraseChar(i, trackChanges))
831 int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
832 char_type next, OutputParams const & runparams)
834 // Writing next here may circumvent a possible font change between
835 // c and next. Since next is only output if it forms a surrogate pair
836 // with c we can ignore this:
837 // A font change inside a surrogate pair does not make sense and is
838 // hopefully impossible to input.
839 // FIXME: change tracking
840 // Is this correct WRT change tracking?
841 Encoding const & encoding = *(runparams.encoding);
842 docstring latex1 = encoding.latexChar(next).first;
843 if (runparams.inIPA) {
844 string const tipashortcut = Encodings::TIPAShortcut(next);
845 if (!tipashortcut.empty()) {
846 latex1 = from_ascii(tipashortcut);
849 docstring const latex2 = encoding.latexChar(c).first;
850 if (docstring(1, next) == latex1) {
851 // the encoding supports the combination
852 os << latex2 << latex1;
853 return latex1.length() + latex2.length();
854 } else if (runparams.local_font &&
855 runparams.local_font->language()->lang() == "polutonikogreek") {
856 // polutonikogreek only works without the brackets
857 os << latex1 << latex2;
858 return latex1.length() + latex2.length();
860 os << latex1 << '{' << latex2 << '}';
861 return latex1.length() + latex2.length() + 2;
865 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
868 unsigned int & column,
870 Layout const & style)
872 if (style.pass_thru || runparams.pass_thru)
875 if (i + 1 < int(text_.size())) {
876 char_type next = text_[i + 1];
877 if (Encodings::isCombiningChar(next)) {
878 // This space has an accent, so we must always output it.
879 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
884 if (runparams.linelen > 0
885 && column > runparams.linelen
887 && text_[i - 1] != ' '
888 && (i + 1 < int(text_.size()))
889 // same in FreeSpacing mode
890 && !owner_->isFreeSpacing()
891 // In typewriter mode, we want to avoid
892 // ! . ? : at the end of a line
893 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
894 && (text_[i - 1] == '.'
895 || text_[i - 1] == '?'
896 || text_[i - 1] == ':'
897 || text_[i - 1] == '!'))) {
899 os.texrow().start(owner_->id(), i + 1);
901 } else if (style.free_spacing) {
910 int Paragraph::Private::writeScriptChars(otexstream & os,
911 docstring const & ltx,
912 Change const & runningChange,
913 Encoding const & encoding,
916 // FIXME: modifying i here is not very nice...
918 // We only arrive here when a proper language for character text_[i] has
919 // not been specified (i.e., it could not be translated in the current
920 // latex encoding) or its latex translation has been forced, and it
921 // belongs to a known script.
922 // Parameter ltx contains the latex translation of text_[i] as specified
923 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
924 // The latex macro name "textXXX" specifies the script to which text_[i]
925 // belongs and we use it in order to check whether characters from the
926 // same script immediately follow, such that we can collect them in a
927 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
928 // for the first char but only "<spec>" for all subsequent chars.
929 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
930 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
931 string script = to_ascii(ltx.substr(1, brace1 - 1));
934 bool closing_brace = true;
935 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
936 // Correct encoding is being used, so we can avoid \textgreek.
939 closing_brace = false;
941 os << ltx.substr(pos, length);
942 int size = text_.size();
943 while (i + 1 < size) {
944 char_type const next = text_[i + 1];
945 // Stop here if next character belongs to another script
946 // or there is a change in change tracking status.
947 if (!Encodings::isKnownScriptChar(next, script) ||
948 runningChange != owner_->lookupChange(i + 1))
952 FontList::const_iterator cit = fontlist_.begin();
953 FontList::const_iterator end = fontlist_.end();
954 for (; cit != end; ++cit) {
955 if (cit->pos() >= i && !found) {
956 prev_font = cit->font();
959 if (cit->pos() >= i + 1)
962 // Stop here if there is a font attribute or encoding change.
963 if (found && cit != end && prev_font != cit->font())
965 docstring const latex = encoding.latexChar(next).first;
966 docstring::size_type const b1 =
967 latex.find_first_of(from_ascii("{"));
968 docstring::size_type const b2 =
969 latex.find_last_of(from_ascii("}"));
970 int const len = b2 - b1 - 1;
971 os << latex.substr(b1 + 1, len);
983 void Paragraph::Private::latexInset(BufferParams const & bparams,
985 OutputParams & runparams,
988 Font const & outerfont,
990 Change & running_change,
991 Layout const & style,
993 unsigned int & column)
995 Inset * inset = owner_->getInset(i);
998 if (style.pass_thru) {
999 odocstringstream ods;
1000 inset->plaintext(ods, runparams);
1005 // FIXME: move this to InsetNewline::latex
1006 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1007 // newlines are handled differently here than
1008 // the default in simpleTeXSpecialChars().
1009 if (!style.newline_allowed) {
1013 column += running_font.latexWriteEndChanges(
1014 os, bparams, runparams,
1015 basefont, basefont);
1019 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1022 basefont = owner_->getLayoutFont(bparams, outerfont);
1023 running_font = basefont;
1025 if (runparams.moving_arg)
1029 os.texrow().start(owner_->id(), i + 1);
1033 if (owner_->isDeleted(i)) {
1034 if( ++runparams.inDeletedInset == 1)
1035 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1038 if (inset->canTrackChanges()) {
1039 column += Changes::latexMarkChange(os, bparams, running_change,
1040 Change(Change::UNCHANGED), runparams);
1041 running_change = Change(Change::UNCHANGED);
1045 odocstream::pos_type const len = os.os().tellp();
1047 if (inset->forceLTR()
1048 && !runparams.use_polyglossia
1049 && running_font.isRightToLeft()
1050 // ERT is an exception, it should be output with no
1051 // decorations at all
1052 && inset->lyxCode() != ERT_CODE) {
1053 if (running_font.language()->lang() == "farsi")
1060 // FIXME: Bug: we can have an empty font change here!
1061 // if there has just been a font change, we are going to close it
1062 // right now, which means stupid latex code like \textsf{}. AFAIK,
1063 // this does not harm dvi output. A minor bug, thus (JMarc)
1065 // Some insets cannot be inside a font change command.
1066 // However, even such insets *can* be placed in \L or \R
1067 // or their equivalents (for RTL language switches), so we don't
1068 // close the language in those cases.
1069 // ArabTeX, though, cannot handle this special behavior, it seems.
1070 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1071 || running_font.language()->lang() == "arabic_arabtex";
1072 if (open_font && !inset->inheritFont()) {
1073 bool closeLanguage = arabtex
1074 || basefont.isRightToLeft() == running_font.isRightToLeft();
1075 unsigned int count = running_font.latexWriteEndChanges(os,
1076 bparams, runparams, basefont, basefont, closeLanguage);
1078 // if any font properties were closed, update the running_font,
1079 // making sure, however, to leave the language as it was
1081 // FIXME: probably a better way to keep track of the old
1082 // language, than copying the entire font?
1083 Font const copy_font(running_font);
1084 basefont = owner_->getLayoutFont(bparams, outerfont);
1085 running_font = basefont;
1087 running_font.setLanguage(copy_font.language());
1088 // leave font open if language is still open
1089 open_font = (running_font.language() == basefont.language());
1091 runparams.local_font = &basefont;
1095 int prev_rows = os.texrow().rows();
1098 runparams.lastid = id_;
1099 runparams.lastpos = i;
1100 inset->latex(os, runparams);
1101 } catch (EncodingException & e) {
1102 // add location information and throw again.
1109 if (running_font.language()->lang() == "farsi")
1115 if (os.texrow().rows() > prev_rows) {
1116 os.texrow().start(owner_->id(), i + 1);
1119 column += (unsigned int)(os.os().tellp() - len);
1122 if (owner_->isDeleted(i))
1123 --runparams.inDeletedInset;
1127 void Paragraph::Private::latexSpecialChar(otexstream & os,
1128 BufferParams const & bparams,
1129 OutputParams const & runparams,
1130 Font const & running_font,
1131 Change const & running_change,
1132 Layout const & style,
1135 unsigned int & column)
1137 // With polyglossia, brackets and stuff need not be reversed
1138 // in RTL scripts (see bug #8251)
1139 char_type const c = (runparams.use_polyglossia) ?
1140 owner_->getUChar(bparams, i) : text_[i];
1142 if (style.pass_thru || runparams.pass_thru) {
1144 Encoding const * const enc = runparams.encoding;
1145 if (enc && !enc->encodable(c))
1146 throw EncodingException(c);
1152 // TIPA uses its own T3 encoding
1153 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1155 // If T1 font encoding is used, use the special
1156 // characters it provides.
1157 // NOTE: Some languages reset the font encoding internally.
1158 // If we are using such a language, we do not output
1159 // special T1 chars.
1160 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1161 && bparams.font_encoding() == "T1" && latexSpecialT1(c, os, i, column))
1164 // Otherwise, we use what LaTeX provides us.
1167 os << "\\textbackslash{}";
1171 os << "\\textless{}";
1175 os << "\\textgreater{}";
1179 os << "\\textbar{}";
1184 if (i + 1 < end_pos && text_[i+1] == '-') {
1185 // Prevent "--" becoming an endash and "---" becoming
1187 // Within \ttfamily, "--" is merged to "-" (no endash)
1188 // so we avoid this rather irritating ligature as well
1194 os << "\\char`\\\"{}";
1199 case '%': case '#': case '{':
1207 os << "\\textasciitilde{}";
1212 os << "\\textasciicircum{}";
1219 // avoid being mistaken for optional arguments
1227 // Blanks are printed before font switching.
1228 // Sure? I am not! (try nice-latex)
1229 // I am sure it's correct. LyX might be smarter
1230 // in the future, but for now, nothing wrong is
1238 Encoding const & encoding = *(runparams.encoding);
1239 char_type next = '\0';
1240 if (i + 1 < int(text_.size())) {
1241 next = text_[i + 1];
1242 if (Encodings::isCombiningChar(next)) {
1243 column += latexSurrogatePair(os, c, next, runparams) - 1;
1249 pair<docstring, bool> latex = encoding.latexChar(c);
1250 docstring nextlatex;
1251 bool nexttipas = false;
1252 string nexttipashortcut;
1253 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1254 nextlatex = encoding.latexChar(next).first;
1255 if (runparams.inIPA) {
1256 nexttipashortcut = Encodings::TIPAShortcut(next);
1257 nexttipas = !nexttipashortcut.empty();
1261 if (runparams.inIPA) {
1262 string const tipashortcut = Encodings::TIPAShortcut(c);
1263 if (!tipashortcut.empty()) {
1264 latex.first = from_ascii(tipashortcut);
1265 latex.second = false;
1269 if (Encodings::isKnownScriptChar(c, script)
1270 && prefixIs(latex.first, from_ascii("\\" + script)))
1271 column += writeScriptChars(os, latex.first,
1272 running_change, encoding, i) - 1;
1273 else if (latex.second
1274 && ((!prefixIs(nextlatex, '\\')
1275 && !prefixIs(nextlatex, '{')
1276 && !prefixIs(nextlatex, '}'))
1278 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1280 // Prevent eating of a following
1281 // space or command corruption by
1282 // following characters
1283 if (next == ' ' || next == '\0') {
1284 column += latex.first.length() + 1;
1285 os << latex.first << "{}";
1287 column += latex.first.length();
1288 os << latex.first << " ";
1291 column += latex.first.length() - 1;
1299 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1300 pos_type i, unsigned int & column)
1306 // In T1 encoding, these characters exist
1307 // but we should avoid ligatures
1308 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1310 os << "\\textcompwordmark{}";
1317 // soul.sty breaks with \char`\"
1318 os << "\\textquotedbl{}";
1327 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1328 pos_type /*i*/, unsigned int & column)
1338 os << "\\textvertline{}";
1347 void Paragraph::Private::validate(LaTeXFeatures & features) const
1349 if (layout_->inpreamble && inset_owner_) {
1350 bool const is_command = layout_->latextype == LATEX_COMMAND;
1351 Buffer const & buf = inset_owner_->buffer();
1352 BufferParams const & bp = features.runparams().is_child
1353 ? buf.masterParams() : buf.params();
1356 // Using a string stream here circumvents the encoding
1357 // switching machinery of odocstream. Therefore the
1358 // output is wrong if this paragraph contains content
1359 // that needs to switch encoding.
1360 odocstringstream ods;
1361 otexstream os(ods, texrow);
1363 os << '\\' << from_ascii(layout_->latexname());
1364 // we have to provide all the optional arguments here, even though
1365 // the last one is the only one we care about.
1366 // Separate handling of optional argument inset.
1367 if (!layout_->latexargs().empty()) {
1368 OutputParams rp = features.runparams();
1369 rp.local_font = &owner_->getFirstFontSettings(bp);
1370 latexArgInsets(*owner_, os, rp, layout_->latexargs());
1372 os << from_ascii(layout_->latexparam());
1374 docstring::size_type const length = ods.str().length();
1375 // this will output "{" at the beginning, but not at the end
1376 owner_->latex(bp, f, os, features.runparams(), 0, -1, true);
1377 if (ods.str().length() > length) {
1380 if (!layout_->postcommandargs().empty()) {
1381 OutputParams rp = features.runparams();
1382 rp.local_font = &owner_->getFirstFontSettings(bp);
1383 latexArgInsets(*owner_, os, rp, layout_->postcommandargs(), "post:");
1386 string const snippet = to_utf8(ods.str());
1387 features.addPreambleSnippet(snippet);
1391 if (features.runparams().flavor == OutputParams::HTML
1392 && layout_->htmltitle()) {
1393 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1396 // check the params.
1397 if (!params_.spacing().isDefault())
1398 features.require("setspace");
1401 features.useLayout(layout_->name());
1404 fontlist_.validate(features);
1406 // then the indentation
1407 if (!params_.leftIndent().zero())
1408 features.require("ParagraphLeftIndent");
1411 InsetList::const_iterator icit = insetlist_.begin();
1412 InsetList::const_iterator iend = insetlist_.end();
1413 for (; icit != iend; ++icit) {
1415 icit->inset->validate(features);
1416 if (layout_->needprotect &&
1417 icit->inset->lyxCode() == FOOT_CODE)
1418 features.require("footmisc");
1422 // then the contents
1423 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1424 BufferEncodings::validate(text_[i], features);
1428 /////////////////////////////////////////////////////////////////////
1432 /////////////////////////////////////////////////////////////////////
1435 Layout const emptyParagraphLayout;
1438 Paragraph::Paragraph()
1439 : d(new Paragraph::Private(this, emptyParagraphLayout))
1446 Paragraph::Paragraph(Paragraph const & par)
1447 : itemdepth(par.itemdepth),
1448 d(new Paragraph::Private(*par.d, this))
1454 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1455 : itemdepth(par.itemdepth),
1456 d(new Paragraph::Private(*par.d, this, beg, end))
1462 Paragraph & Paragraph::operator=(Paragraph const & par)
1464 // needed as we will destroy the private part before copying it
1466 itemdepth = par.itemdepth;
1470 d = new Private(*par.d, this);
1477 Paragraph::~Paragraph()
1486 // this shall be called just before every "os << ..." action.
1487 void flushString(ostream & os, docstring & s)
1496 void Paragraph::write(ostream & os, BufferParams const & bparams,
1497 depth_type & dth) const
1499 // The beginning or end of a deeper (i.e. nested) area?
1500 if (dth != d->params_.depth()) {
1501 if (d->params_.depth() > dth) {
1502 while (d->params_.depth() > dth) {
1503 os << "\n\\begin_deeper";
1507 while (d->params_.depth() < dth) {
1508 os << "\n\\end_deeper";
1514 // First write the layout
1515 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1517 d->params_.write(os);
1519 Font font1(inherit_font, bparams.language);
1521 Change running_change = Change(Change::UNCHANGED);
1523 // this string is used as a buffer to avoid repetitive calls
1524 // to to_utf8(), which turn out to be expensive (JMarc)
1525 docstring write_buffer;
1528 for (pos_type i = 0; i <= size(); ++i) {
1530 Change const & change = lookupChange(i);
1531 if (change != running_change)
1532 flushString(os, write_buffer);
1533 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1534 running_change = change;
1539 // Write font changes
1540 Font font2 = getFontSettings(bparams, i);
1541 if (font2 != font1) {
1542 flushString(os, write_buffer);
1543 font2.lyxWriteChanges(font1, os);
1548 char_type const c = d->text_[i];
1551 if (Inset const * inset = getInset(i)) {
1552 flushString(os, write_buffer);
1553 if (inset->directWrite()) {
1554 // international char, let it write
1555 // code directly so it's shorter in
1561 os << "\\begin_inset ";
1563 os << "\n\\end_inset\n\n";
1566 // FIXME This can be removed again once the mystery
1567 // crash has been resolved.
1572 flushString(os, write_buffer);
1573 os << "\n\\backslash\n";
1577 flushString(os, write_buffer);
1578 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1585 if ((column > 70 && c == ' ')
1587 flushString(os, write_buffer);
1591 // this check is to amend a bug. LyX sometimes
1592 // inserts '\0' this could cause problems.
1594 write_buffer.push_back(c);
1596 LYXERR0("NUL char in structure.");
1602 flushString(os, write_buffer);
1603 os << "\n\\end_layout\n";
1604 // FIXME This can be removed again once the mystery
1605 // crash has been resolved.
1610 void Paragraph::validate(LaTeXFeatures & features) const
1612 d->validate(features);
1616 void Paragraph::insert(pos_type start, docstring const & str,
1617 Font const & font, Change const & change)
1619 for (size_t i = 0, n = str.size(); i != n ; ++i)
1620 insertChar(start + i, str[i], font, change);
1624 void Paragraph::appendChar(char_type c, Font const & font,
1625 Change const & change)
1628 d->changes_.insert(change, d->text_.size());
1629 // when appending characters, no need to update tables
1630 d->text_.push_back(c);
1631 setFont(d->text_.size() - 1, font);
1632 d->requestSpellCheck(d->text_.size() - 1);
1636 void Paragraph::appendString(docstring const & s, Font const & font,
1637 Change const & change)
1639 pos_type end = s.size();
1640 size_t oldsize = d->text_.size();
1641 size_t newsize = oldsize + end;
1642 size_t capacity = d->text_.capacity();
1643 if (newsize >= capacity)
1644 d->text_.reserve(max(capacity + 100, newsize));
1646 // when appending characters, no need to update tables
1649 // FIXME: Optimize this!
1650 for (size_t i = oldsize; i != newsize; ++i) {
1652 d->changes_.insert(change, i);
1653 d->requestSpellCheck(i);
1655 d->fontlist_.set(oldsize, font);
1656 d->fontlist_.set(newsize - 1, font);
1660 void Paragraph::insertChar(pos_type pos, char_type c,
1663 d->insertChar(pos, c, Change(trackChanges ?
1664 Change::INSERTED : Change::UNCHANGED));
1668 void Paragraph::insertChar(pos_type pos, char_type c,
1669 Font const & font, bool trackChanges)
1671 d->insertChar(pos, c, Change(trackChanges ?
1672 Change::INSERTED : Change::UNCHANGED));
1677 void Paragraph::insertChar(pos_type pos, char_type c,
1678 Font const & font, Change const & change)
1680 d->insertChar(pos, c, change);
1685 void Paragraph::resetFonts(Font const & font)
1687 d->fontlist_.clear();
1688 d->fontlist_.set(0, font);
1689 d->fontlist_.set(d->text_.size() - 1, font);
1692 // Gets uninstantiated font setting at position.
1693 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1697 LYXERR0("pos: " << pos << " size: " << size());
1701 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1702 if (cit != d->fontlist_.end())
1705 if (pos == size() && !empty())
1706 return getFontSettings(bparams, pos - 1);
1708 // Optimisation: avoid a full font instantiation if there is no
1709 // language change from previous call.
1710 static Font previous_font;
1711 static Language const * previous_lang = 0;
1712 Language const * lang = getParLanguage(bparams);
1713 if (lang != previous_lang) {
1714 previous_lang = lang;
1715 previous_font = Font(inherit_font, lang);
1717 return previous_font;
1721 FontSpan Paragraph::fontSpan(pos_type pos) const
1723 LBUFERR(pos < size());
1726 FontList::const_iterator cit = d->fontlist_.begin();
1727 FontList::const_iterator end = d->fontlist_.end();
1728 for (; cit != end; ++cit) {
1729 if (cit->pos() >= pos) {
1730 if (pos >= beginOfBody())
1731 return FontSpan(max(start, beginOfBody()),
1734 return FontSpan(start,
1735 min(beginOfBody() - 1,
1738 start = cit->pos() + 1;
1741 // This should not happen, but if so, we take no chances.
1742 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1743 LASSERT(false, return FontSpan(pos, pos));
1747 // Gets uninstantiated font setting at position 0
1748 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1750 if (!empty() && !d->fontlist_.empty())
1751 return d->fontlist_.begin()->font();
1753 // Optimisation: avoid a full font instantiation if there is no
1754 // language change from previous call.
1755 static Font previous_font;
1756 static Language const * previous_lang = 0;
1757 if (bparams.language != previous_lang) {
1758 previous_lang = bparams.language;
1759 previous_font = Font(inherit_font, bparams.language);
1762 return previous_font;
1766 // Gets the fully instantiated font at a given position in a paragraph
1767 // This is basically the same function as Text::GetFont() in text2.cpp.
1768 // The difference is that this one is used for generating the LaTeX file,
1769 // and thus cosmetic "improvements" are disallowed: This has to deliver
1770 // the true picture of the buffer. (Asger)
1771 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1772 Font const & outerfont) const
1776 Font font = getFontSettings(bparams, pos);
1778 pos_type const body_pos = beginOfBody();
1779 FontInfo & fi = font.fontInfo();
1781 fi.realize(d->layout_->labelfont);
1783 fi.realize(d->layout_->font);
1785 fi.realize(outerfont.fontInfo());
1786 fi.realize(bparams.getFont().fontInfo());
1792 Font const Paragraph::getLabelFont
1793 (BufferParams const & bparams, Font const & outerfont) const
1795 FontInfo tmpfont = d->layout_->labelfont;
1796 tmpfont.realize(outerfont.fontInfo());
1797 tmpfont.realize(bparams.getFont().fontInfo());
1798 return Font(tmpfont, getParLanguage(bparams));
1802 Font const Paragraph::getLayoutFont
1803 (BufferParams const & bparams, Font const & outerfont) const
1805 FontInfo tmpfont = d->layout_->font;
1806 tmpfont.realize(outerfont.fontInfo());
1807 tmpfont.realize(bparams.getFont().fontInfo());
1808 return Font(tmpfont, getParLanguage(bparams));
1812 /// Returns the height of the highest font in range
1813 FontSize Paragraph::highestFontInRange
1814 (pos_type startpos, pos_type endpos, FontSize def_size) const
1816 return d->fontlist_.highestInRange(startpos, endpos, def_size);
1820 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1822 char_type c = d->text_[pos];
1823 if (!getFontSettings(bparams, pos).isRightToLeft())
1826 // FIXME: The arabic special casing is due to the difference of arabic
1827 // round brackets input introduced in r18599. Check if this should be
1828 // unified with Hebrew or at least if all bracket types should be
1829 // handled the same (file format change in either case).
1830 string const & lang = getFontSettings(bparams, pos).language()->lang();
1831 bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
1836 uc = arabic ? c : ')';
1839 uc = arabic ? c : '(';
1865 void Paragraph::setFont(pos_type pos, Font const & font)
1867 LASSERT(pos <= size(), return);
1869 // First, reduce font against layout/label font
1870 // Update: The setCharFont() routine in text2.cpp already
1871 // reduces font, so we don't need to do that here. (Asger)
1873 d->fontlist_.set(pos, font);
1877 void Paragraph::makeSameLayout(Paragraph const & par)
1879 d->layout_ = par.d->layout_;
1880 d->params_ = par.d->params_;
1884 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1886 if (isFreeSpacing())
1892 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1893 if (eraseChar(pos, trackChanges))
1899 return count > 0 || pos > 0;
1903 bool Paragraph::hasSameLayout(Paragraph const & par) const
1905 return par.d->layout_ == d->layout_
1906 && d->params_.sameLayout(par.d->params_);
1910 depth_type Paragraph::getDepth() const
1912 return d->params_.depth();
1916 depth_type Paragraph::getMaxDepthAfter() const
1918 if (d->layout_->isEnvironment())
1919 return d->params_.depth() + 1;
1921 return d->params_.depth();
1925 char Paragraph::getAlign() const
1927 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1928 return d->layout_->align;
1930 return d->params_.align();
1934 docstring const & Paragraph::labelString() const
1936 return d->params_.labelString();
1940 // the next two functions are for the manual labels
1941 docstring const Paragraph::getLabelWidthString() const
1943 if (d->layout_->margintype == MARGIN_MANUAL
1944 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
1945 return d->params_.labelWidthString();
1947 return _("Senseless with this layout!");
1951 void Paragraph::setLabelWidthString(docstring const & s)
1953 d->params_.labelWidthString(s);
1957 docstring Paragraph::expandLabel(Layout const & layout,
1958 BufferParams const & bparams) const
1960 return expandParagraphLabel(layout, bparams, true);
1964 docstring Paragraph::expandDocBookLabel(Layout const & layout,
1965 BufferParams const & bparams) const
1967 return expandParagraphLabel(layout, bparams, false);
1971 docstring Paragraph::expandParagraphLabel(Layout const & layout,
1972 BufferParams const & bparams, bool process_appendix) const
1974 DocumentClass const & tclass = bparams.documentClass();
1975 string const & lang = getParLanguage(bparams)->code();
1976 bool const in_appendix = process_appendix && d->params_.appendix();
1977 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
1979 if (fmt.empty() && !layout.counter.empty())
1980 return tclass.counters().theCounter(layout.counter, lang);
1982 // handle 'inherited level parts' in 'fmt',
1983 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
1984 size_t const i = fmt.find('@', 0);
1985 if (i != docstring::npos) {
1986 size_t const j = fmt.find('@', i + 1);
1987 if (j != docstring::npos) {
1988 docstring parent(fmt, i + 1, j - i - 1);
1989 docstring label = from_ascii("??");
1990 if (tclass.hasLayout(parent))
1991 docstring label = expandParagraphLabel(tclass[parent], bparams,
1993 fmt = docstring(fmt, 0, i) + label
1994 + docstring(fmt, j + 1, docstring::npos);
1998 return tclass.counters().counterLabel(fmt, lang);
2002 void Paragraph::applyLayout(Layout const & new_layout)
2004 d->layout_ = &new_layout;
2005 LyXAlignment const oldAlign = d->params_.align();
2007 if (!(oldAlign & d->layout_->alignpossible)) {
2008 frontend::Alert::warning(_("Alignment not permitted"),
2009 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2010 d->params_.align(LYX_ALIGN_LAYOUT);
2015 pos_type Paragraph::beginOfBody() const
2017 return d->begin_of_body_;
2021 void Paragraph::setBeginOfBody()
2023 if (d->layout_->labeltype != LABEL_MANUAL) {
2024 d->begin_of_body_ = 0;
2028 // Unroll the first two cycles of the loop
2029 // and remember the previous character to
2030 // remove unnecessary getChar() calls
2032 pos_type end = size();
2033 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2035 char_type previous_char = 0;
2038 previous_char = d->text_[i];
2039 if (!(isNewline(i) || isEnvSeparator(i))) {
2041 while (i < end && previous_char != ' ') {
2043 if (isNewline(i) || isEnvSeparator(i))
2046 previous_char = temp;
2052 d->begin_of_body_ = i;
2056 bool Paragraph::allowParagraphCustomization() const
2058 return inInset().allowParagraphCustomization();
2062 bool Paragraph::usePlainLayout() const
2064 return inInset().usePlainLayout();
2068 bool Paragraph::isPassThru() const
2070 return inInset().isPassThru() || d->layout_->pass_thru;
2075 // paragraphs inside floats need different alignment tags to avoid
2078 bool noTrivlistCentering(InsetCode code)
2080 return code == FLOAT_CODE
2081 || code == WRAP_CODE
2082 || code == CELL_CODE;
2086 string correction(string const & orig)
2088 if (orig == "flushleft")
2089 return "raggedright";
2090 if (orig == "flushright")
2091 return "raggedleft";
2092 if (orig == "center")
2098 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2099 InsetCode code, bool const lastpar, int & col)
2101 string macro = suffix + "{";
2102 if (noTrivlistCentering(code)) {
2104 // the last paragraph in non-trivlist-aligned
2105 // context is special (to avoid unwanted whitespace)
2106 if (suffix == "\\begin") {
2107 macro = "\\" + correction(env) + "{}";
2108 os << from_ascii(macro);
2109 col += macro.size();
2114 macro += correction(env);
2118 if (suffix == "\\par\\end") {
2122 os << from_ascii(macro);
2123 col += macro.size();
2124 if (suffix == "\\begin") {
2134 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2135 otexstream & os, OutputParams const & runparams) const
2139 if (params_.noindent() && !layout_->pass_thru
2140 && (layout_->toggle_indent != ITOGGLE_NEVER)) {
2141 os << "\\noindent ";
2145 LyXAlignment const curAlign = params_.align();
2147 if (curAlign == layout_->align)
2151 case LYX_ALIGN_NONE:
2152 case LYX_ALIGN_BLOCK:
2153 case LYX_ALIGN_LAYOUT:
2154 case LYX_ALIGN_SPECIAL:
2155 case LYX_ALIGN_DECIMAL:
2157 case LYX_ALIGN_LEFT:
2158 case LYX_ALIGN_RIGHT:
2159 case LYX_ALIGN_CENTER:
2160 if (runparams.moving_arg) {
2167 string const begin_tag = "\\begin";
2168 InsetCode code = ownerCode();
2169 bool const lastpar = runparams.isLastPar;
2172 case LYX_ALIGN_NONE:
2173 case LYX_ALIGN_BLOCK:
2174 case LYX_ALIGN_LAYOUT:
2175 case LYX_ALIGN_SPECIAL:
2176 case LYX_ALIGN_DECIMAL:
2178 case LYX_ALIGN_LEFT: {
2179 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2180 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2182 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2184 } case LYX_ALIGN_RIGHT: {
2186 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2187 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2189 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2191 } case LYX_ALIGN_CENTER: {
2192 corrected_env(os, begin_tag, "center", code, lastpar, column);
2201 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2202 otexstream & os, OutputParams const & runparams) const
2204 LyXAlignment const curAlign = params_.align();
2206 if (curAlign == layout_->align)
2210 case LYX_ALIGN_NONE:
2211 case LYX_ALIGN_BLOCK:
2212 case LYX_ALIGN_LAYOUT:
2213 case LYX_ALIGN_SPECIAL:
2214 case LYX_ALIGN_DECIMAL:
2216 case LYX_ALIGN_LEFT:
2217 case LYX_ALIGN_RIGHT:
2218 case LYX_ALIGN_CENTER:
2219 if (runparams.moving_arg)
2224 bool output = false;
2226 string const end_tag = "\\par\\end";
2227 InsetCode code = ownerCode();
2228 bool const lastpar = runparams.isLastPar;
2231 case LYX_ALIGN_NONE:
2232 case LYX_ALIGN_BLOCK:
2233 case LYX_ALIGN_LAYOUT:
2234 case LYX_ALIGN_SPECIAL:
2235 case LYX_ALIGN_DECIMAL:
2237 case LYX_ALIGN_LEFT: {
2238 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2239 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2241 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2243 } case LYX_ALIGN_RIGHT: {
2244 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2245 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2247 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2249 } case LYX_ALIGN_CENTER: {
2250 corrected_env(os, end_tag, "center", code, lastpar, col);
2255 return output || lastpar;
2259 // This one spits out the text of the paragraph
2260 void Paragraph::latex(BufferParams const & bparams,
2261 Font const & outerfont,
2263 OutputParams const & runparams,
2264 int start_pos, int end_pos, bool force) const
2266 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2268 // FIXME This check should not be needed. Perhaps issue an
2269 // error if it triggers.
2270 Layout const & style = inInset().forcePlainLayout() ?
2271 bparams.documentClass().plainLayout() : *d->layout_;
2273 if (!force && style.inpreamble)
2276 bool const allowcust = allowParagraphCustomization();
2278 // Current base font for all inherited font changes, without any
2279 // change caused by an individual character, except for the language:
2280 // It is set to the language of the first character.
2281 // As long as we are in the label, this font is the base font of the
2282 // label. Before the first body character it is set to the base font
2286 // Maybe we have to create a optional argument.
2287 pos_type body_pos = beginOfBody();
2288 unsigned int column = 0;
2291 // the optional argument is kept in curly brackets in
2292 // case it contains a ']'
2293 // This is not strictly needed, but if this is changed it
2294 // would be a file format change, and tex2lyx would need
2295 // to be adjusted, since it unconditionally removes the
2296 // braces when it parses \item.
2299 basefont = getLabelFont(bparams, outerfont);
2301 basefont = getLayoutFont(bparams, outerfont);
2304 // Which font is currently active?
2305 Font running_font(basefont);
2306 // Do we have an open font change?
2307 bool open_font = false;
2309 Change runningChange = Change(Change::UNCHANGED);
2311 Encoding const * const prev_encoding = runparams.encoding;
2313 os.texrow().start(id(), 0);
2315 // if the paragraph is empty, the loop will not be entered at all
2317 if (style.isCommand()) {
2321 if (!style.leftdelim().empty()) {
2322 os << style.leftdelim();
2323 column += style.leftdelim().size();
2326 column += d->startTeXParParams(bparams, os, runparams);
2329 for (pos_type i = 0; i < size(); ++i) {
2330 // First char in paragraph or after label?
2331 if (i == body_pos) {
2334 column += running_font.latexWriteEndChanges(
2335 os, bparams, runparams,
2336 basefont, basefont);
2339 basefont = getLayoutFont(bparams, outerfont);
2340 running_font = basefont;
2342 column += Changes::latexMarkChange(os, bparams,
2343 runningChange, Change(Change::UNCHANGED),
2345 runningChange = Change(Change::UNCHANGED);
2350 if (style.isCommand()) {
2355 if (!style.leftdelim().empty()) {
2356 os << style.leftdelim();
2357 column += style.leftdelim().size();
2361 column += d->startTeXParParams(bparams, os,
2365 Change const & change = runparams.inDeletedInset
2366 ? runparams.changeOfDeletedInset : lookupChange(i);
2368 if (bparams.output_changes && runningChange != change) {
2370 column += running_font.latexWriteEndChanges(
2371 os, bparams, runparams, basefont, basefont);
2374 basefont = getLayoutFont(bparams, outerfont);
2375 running_font = basefont;
2377 column += Changes::latexMarkChange(os, bparams, runningChange,
2379 runningChange = change;
2382 // do not output text which is marked deleted
2383 // if change tracking output is disabled
2384 if (!bparams.output_changes && change.deleted()) {
2390 // Fully instantiated font
2391 Font const font = getFont(bparams, i, outerfont);
2393 Font const last_font = running_font;
2395 // Do we need to close the previous font?
2397 (font != running_font ||
2398 font.language() != running_font.language()))
2400 column += running_font.latexWriteEndChanges(
2401 os, bparams, runparams, basefont,
2402 (i == body_pos-1) ? basefont : font);
2403 running_font = basefont;
2407 string const running_lang = runparams.use_polyglossia ?
2408 running_font.language()->polyglossia() : running_font.language()->babel();
2409 // close babel's font environment before opening CJK.
2410 string const lang_end_command = runparams.use_polyglossia ?
2411 "\\end{$$lang}" : lyxrc.language_command_end;
2412 if (!running_lang.empty() &&
2413 font.language()->encoding()->package() == Encoding::CJK) {
2414 string end_tag = subst(lang_end_command,
2417 os << from_ascii(end_tag);
2418 column += end_tag.length();
2421 // Switch file encoding if necessary (and allowed)
2422 if (!runparams.pass_thru && !style.pass_thru &&
2423 runparams.encoding->package() != Encoding::none &&
2424 font.language()->encoding()->package() != Encoding::none) {
2425 pair<bool, int> const enc_switch =
2426 switchEncoding(os.os(), bparams, runparams,
2427 *(font.language()->encoding()));
2428 if (enc_switch.first) {
2429 column += enc_switch.second;
2430 runparams.encoding = font.language()->encoding();
2434 char_type const c = d->text_[i];
2436 // Do we need to change font?
2437 if ((font != running_font ||
2438 font.language() != running_font.language()) &&
2441 odocstringstream ods;
2442 column += font.latexWriteStartChanges(ods, bparams,
2443 runparams, basefont,
2445 running_font = font;
2447 docstring fontchange = ods.str();
2448 // check whether the fontchange ends with a \\textcolor
2449 // modifier and the text starts with a space (bug 4473)
2450 docstring const last_modifier = rsplit(fontchange, '\\');
2451 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2452 os << fontchange << from_ascii("{}");
2453 // check if the fontchange ends with a trailing blank
2454 // (like "\small " (see bug 3382)
2455 else if (suffixIs(fontchange, ' ') && c == ' ')
2456 os << fontchange.substr(0, fontchange.size() - 1)
2457 << from_ascii("{}");
2462 // FIXME: think about end_pos implementation...
2463 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2464 // FIXME: integrate this case in latexSpecialChar
2465 // Do not print the separation of the optional argument
2466 // if style.pass_thru is false. This works because
2467 // latexSpecialChar ignores spaces if
2468 // style.pass_thru is false.
2469 if (i != body_pos - 1) {
2470 if (d->simpleTeXBlanks(runparams, os,
2471 i, column, font, style)) {
2472 // A surrogate pair was output. We
2473 // must not call latexSpecialChar
2474 // in this iteration, since it would output
2475 // the combining character again.
2482 OutputParams rp = runparams;
2483 rp.free_spacing = style.free_spacing;
2484 rp.local_font = &font;
2485 rp.intitle = style.intitle;
2487 // Two major modes: LaTeX or plain
2488 // Handle here those cases common to both modes
2489 // and then split to handle the two modes separately.
2490 if (c == META_INSET) {
2491 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2492 d->latexInset(bparams, os, rp, running_font,
2493 basefont, outerfont, open_font,
2494 runningChange, style, i, column);
2497 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2499 d->latexSpecialChar(os, bparams, rp, running_font, runningChange,
2500 style, i, end_pos, column);
2501 } catch (EncodingException & e) {
2502 if (runparams.dryrun) {
2503 os << "<" << _("LyX Warning: ")
2504 << _("uncodable character") << " '";
2508 // add location information and throw again.
2517 // Set the encoding to that returned from latexSpecialChar (see
2518 // comment for encoding member in OutputParams.h)
2519 runparams.encoding = rp.encoding;
2522 // If we have an open font definition, we have to close it
2524 #ifdef FIXED_LANGUAGE_END_DETECTION
2526 running_font.latexWriteEndChanges(os, bparams,
2527 runparams, basefont,
2528 next_->getFont(bparams, 0, outerfont));
2530 running_font.latexWriteEndChanges(os, bparams,
2531 runparams, basefont, basefont);
2534 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2535 //FIXME: there as we start another \selectlanguage with the next paragraph if
2536 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2537 running_font.latexWriteEndChanges(os, bparams, runparams,
2538 basefont, basefont);
2542 column += Changes::latexMarkChange(os, bparams, runningChange,
2543 Change(Change::UNCHANGED), runparams);
2545 // Needed if there is an optional argument but no contents.
2546 if (body_pos > 0 && body_pos == size()) {
2550 if (!style.rightdelim().empty()) {
2551 os << style.rightdelim();
2552 column += style.rightdelim().size();
2555 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2556 && runparams.encoding != prev_encoding) {
2557 runparams.encoding = prev_encoding;
2558 if (!runparams.isFullUnicode())
2559 os << setEncoding(prev_encoding->iconvName());
2562 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2566 bool Paragraph::emptyTag() const
2568 for (pos_type i = 0; i < size(); ++i) {
2569 if (Inset const * inset = getInset(i)) {
2570 InsetCode lyx_code = inset->lyxCode();
2571 // FIXME testing like that is wrong. What is
2573 if (lyx_code != TOC_CODE &&
2574 lyx_code != INCLUDE_CODE &&
2575 lyx_code != GRAPHICS_CODE &&
2576 lyx_code != ERT_CODE &&
2577 lyx_code != LISTINGS_CODE &&
2578 lyx_code != FLOAT_CODE &&
2579 lyx_code != TABULAR_CODE) {
2583 char_type c = d->text_[i];
2584 if (c != ' ' && c != '\t')
2592 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2595 for (pos_type i = 0; i < size(); ++i) {
2596 if (Inset const * inset = getInset(i)) {
2597 InsetCode lyx_code = inset->lyxCode();
2598 if (lyx_code == LABEL_CODE) {
2599 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2600 docstring const & id = il->getParam("name");
2601 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2609 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2613 for (i = 0; i < size(); ++i) {
2614 if (Inset const * inset = getInset(i)) {
2615 inset->docbook(os, runparams);
2617 char_type c = d->text_[i];
2620 os << sgml::escapeChar(c);
2627 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2631 for (i = 0; i < size(); ++i) {
2632 if (Inset const * inset = getInset(i)) {
2633 inset->xhtml(xs, runparams);
2635 char_type c = d->text_[i];
2645 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2648 pos_type size = text_.size();
2649 for (pos_type i = initial; i < size; ++i) {
2650 Font font = owner_->getFont(buf.params(), i, outerfont);
2651 if (text_[i] == META_INSET)
2653 if (i != initial && font != font_old)
2662 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2664 OutputParams const & runparams,
2665 Font const & outerfont,
2666 pos_type initial) const
2668 bool emph_flag = false;
2670 Layout const & style = *d->layout_;
2672 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2674 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2677 // parsing main loop
2678 for (pos_type i = initial; i < size(); ++i) {
2679 Font font = getFont(buf.params(), i, outerfont);
2681 // handle <emphasis> tag
2682 if (font_old.emph() != font.fontInfo().emph()) {
2683 if (font.fontInfo().emph() == FONT_ON) {
2686 } else if (i != initial) {
2687 os << "</emphasis>";
2692 if (Inset const * inset = getInset(i)) {
2693 inset->docbook(os, runparams);
2695 char_type c = d->text_[i];
2697 if (style.pass_thru)
2700 os << sgml::escapeChar(c);
2702 font_old = font.fontInfo();
2706 os << "</emphasis>";
2709 if (style.free_spacing)
2711 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2717 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2718 vector<html::EndFontTag> & tagsToClose,
2719 bool & flag, FontState curstate, html::FontTypes type)
2721 if (curstate == FONT_ON) {
2722 tagsToOpen.push_back(html::FontTag(type));
2725 tagsToClose.push_back(html::EndFontTag(type));
2732 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2734 OutputParams const & runparams,
2735 Font const & outerfont,
2736 pos_type initial) const
2740 // track whether we have opened these tags
2741 bool emph_flag = false;
2742 bool bold_flag = false;
2743 bool noun_flag = false;
2744 bool ubar_flag = false;
2745 bool dbar_flag = false;
2746 bool sout_flag = false;
2747 bool wave_flag = false;
2749 bool shap_flag = false;
2751 bool faml_flag = false;
2753 bool size_flag = false;
2755 Layout const & style = *d->layout_;
2757 xs.startParagraph(allowEmpty());
2760 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2762 FontShape curr_fs = INHERIT_SHAPE;
2763 FontFamily curr_fam = INHERIT_FAMILY;
2764 FontSize curr_size = FONT_SIZE_INHERIT;
2766 string const default_family =
2767 buf.masterBuffer()->params().fonts_default_family;
2769 vector<html::FontTag> tagsToOpen;
2770 vector<html::EndFontTag> tagsToClose;
2772 // parsing main loop
2773 for (pos_type i = initial; i < size(); ++i) {
2774 // let's not show deleted material in the output
2778 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
2781 FontState curstate = font.fontInfo().emph();
2782 if (font_old.emph() != curstate)
2783 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
2786 curstate = font.fontInfo().noun();
2787 if (font_old.noun() != curstate)
2788 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
2791 curstate = font.fontInfo().underbar();
2792 if (font_old.underbar() != curstate)
2793 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
2796 curstate = font.fontInfo().strikeout();
2797 if (font_old.strikeout() != curstate)
2798 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
2801 curstate = font.fontInfo().uuline();
2802 if (font_old.uuline() != curstate)
2803 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
2806 curstate = font.fontInfo().uwave();
2807 if (font_old.uwave() != curstate)
2808 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
2811 // a little hackish, but allows us to reuse what we have.
2812 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
2813 if (font_old.series() != font.fontInfo().series())
2814 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
2817 curr_fs = font.fontInfo().shape();
2818 FontShape old_fs = font_old.shape();
2819 if (old_fs != curr_fs) {
2823 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
2826 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
2828 case SMALLCAPS_SHAPE:
2829 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
2835 // the other tags are for internal use
2843 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
2847 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
2850 case SMALLCAPS_SHAPE:
2851 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
2858 // the other tags are for internal use
2865 curr_fam = font.fontInfo().family();
2866 FontFamily old_fam = font_old.family();
2867 if (old_fam != curr_fam) {
2871 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
2874 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
2876 case TYPEWRITER_FAMILY:
2877 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
2879 case INHERIT_FAMILY:
2882 // the other tags are for internal use
2890 // we will treat a "default" font family as roman, since we have
2891 // no other idea what to do.
2892 if (default_family != "rmdefault" && default_family != "default") {
2893 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
2898 if (default_family != "sfdefault") {
2899 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
2903 case TYPEWRITER_FAMILY:
2904 if (default_family != "ttdefault") {
2905 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
2909 case INHERIT_FAMILY:
2912 // the other tags are for internal use
2919 curr_size = font.fontInfo().size();
2920 FontSize old_size = font_old.size();
2921 if (old_size != curr_size) {
2924 case FONT_SIZE_TINY:
2925 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
2927 case FONT_SIZE_SCRIPT:
2928 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
2930 case FONT_SIZE_FOOTNOTE:
2931 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
2933 case FONT_SIZE_SMALL:
2934 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
2936 case FONT_SIZE_LARGE:
2937 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
2939 case FONT_SIZE_LARGER:
2940 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
2942 case FONT_SIZE_LARGEST:
2943 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
2945 case FONT_SIZE_HUGE:
2946 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
2948 case FONT_SIZE_HUGER:
2949 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
2951 case FONT_SIZE_INCREASE:
2952 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
2954 case FONT_SIZE_DECREASE:
2955 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
2957 case FONT_SIZE_INHERIT:
2958 case FONT_SIZE_NORMAL:
2961 // the other tags are for internal use
2967 switch (curr_size) {
2968 case FONT_SIZE_TINY:
2969 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
2972 case FONT_SIZE_SCRIPT:
2973 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
2976 case FONT_SIZE_FOOTNOTE:
2977 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
2980 case FONT_SIZE_SMALL:
2981 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
2984 case FONT_SIZE_LARGE:
2985 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
2988 case FONT_SIZE_LARGER:
2989 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
2992 case FONT_SIZE_LARGEST:
2993 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
2996 case FONT_SIZE_HUGE:
2997 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3000 case FONT_SIZE_HUGER:
3001 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3004 case FONT_SIZE_INCREASE:
3005 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3008 case FONT_SIZE_DECREASE:
3009 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3012 case FONT_SIZE_NORMAL:
3013 case FONT_SIZE_INHERIT:
3016 // the other tags are for internal use
3023 // Other such tags? What about the other text ranges?
3025 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3026 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3027 for (; cit != cen; ++cit)
3030 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3031 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3032 for (; sit != sen; ++sit)
3035 tagsToClose.clear();
3038 Inset const * inset = getInset(i);
3040 if (!runparams.for_toc || inset->isInToc()) {
3041 OutputParams np = runparams;
3042 np.local_font = &font;
3043 if (!inset->getLayout().htmlisblock())
3044 np.html_in_par = true;
3045 retval += inset->xhtml(xs, np);
3048 char_type c = getUChar(buf.masterBuffer()->params(), i);
3051 font_old = font.fontInfo();
3060 bool Paragraph::isHfill(pos_type pos) const
3062 Inset const * inset = getInset(pos);
3063 return inset && inset->isHfill();
3067 bool Paragraph::isNewline(pos_type pos) const
3069 Inset const * inset = getInset(pos);
3070 return inset && inset->lyxCode() == NEWLINE_CODE;
3074 bool Paragraph::isEnvSeparator(pos_type pos) const
3076 Inset const * inset = getInset(pos);
3077 return inset && inset->lyxCode() == SEPARATOR_CODE;
3081 bool Paragraph::isLineSeparator(pos_type pos) const
3083 char_type const c = d->text_[pos];
3084 if (isLineSeparatorChar(c))
3086 Inset const * inset = getInset(pos);
3087 return inset && inset->isLineSeparator();
3091 bool Paragraph::isWordSeparator(pos_type pos) const
3095 if (Inset const * inset = getInset(pos))
3096 return !inset->isLetter();
3097 // if we have a hard hyphen (no en- or emdash) or apostrophe
3098 // we pass this to the spell checker
3099 // FIXME: this method is subject to change, visit
3100 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3101 // to get an impression how complex this is.
3102 if (isHardHyphenOrApostrophe(pos))
3104 char_type const c = d->text_[pos];
3105 // We want to pass the escape chars to the spellchecker
3106 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3107 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3111 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3113 pos_type const psize = size();
3116 char_type const c = d->text_[pos];
3117 if (c != '-' && c != '\'')
3119 int nextpos = pos + 1;
3120 int prevpos = pos > 0 ? pos - 1 : 0;
3121 if ((nextpos == psize || isSpace(nextpos))
3122 && (pos == 0 || isSpace(prevpos)))
3128 bool Paragraph::isSameSpellRange(pos_type pos1, pos_type pos2) const
3131 || d->speller_state_.getRange(pos1) == d->speller_state_.getRange(pos2);
3135 bool Paragraph::isChar(pos_type pos) const
3137 if (Inset const * inset = getInset(pos))
3138 return inset->isChar();
3139 char_type const c = d->text_[pos];
3140 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3144 bool Paragraph::isSpace(pos_type pos) const
3146 if (Inset const * inset = getInset(pos))
3147 return inset->isSpace();
3148 char_type const c = d->text_[pos];
3149 return lyx::isSpace(c);
3154 Paragraph::getParLanguage(BufferParams const & bparams) const
3157 return getFirstFontSettings(bparams).language();
3158 // FIXME: we should check the prev par as well (Lgb)
3159 return bparams.language;
3163 bool Paragraph::isRTL(BufferParams const & bparams) const
3165 return getParLanguage(bparams)->rightToLeft()
3166 && !inInset().getLayout().forceLTR();
3170 void Paragraph::changeLanguage(BufferParams const & bparams,
3171 Language const * from, Language const * to)
3173 // change language including dummy font change at the end
3174 for (pos_type i = 0; i <= size(); ++i) {
3175 Font font = getFontSettings(bparams, i);
3176 if (font.language() == from) {
3177 font.setLanguage(to);
3179 d->requestSpellCheck(i);
3185 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3187 Language const * doc_language = bparams.language;
3188 FontList::const_iterator cit = d->fontlist_.begin();
3189 FontList::const_iterator end = d->fontlist_.end();
3191 for (; cit != end; ++cit)
3192 if (cit->font().language() != ignore_language &&
3193 cit->font().language() != latex_language &&
3194 cit->font().language() != doc_language)
3200 void Paragraph::getLanguages(std::set<Language const *> & languages) const
3202 FontList::const_iterator cit = d->fontlist_.begin();
3203 FontList::const_iterator end = d->fontlist_.end();
3205 for (; cit != end; ++cit) {
3206 Language const * lang = cit->font().language();
3207 if (lang != ignore_language &&
3208 lang != latex_language)
3209 languages.insert(lang);
3214 docstring Paragraph::asString(int options) const
3216 return asString(0, size(), options);
3220 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3222 odocstringstream os;
3225 && options & AS_STR_LABEL
3226 && !d->params_.labelString().empty())
3227 os << d->params_.labelString() << ' ';
3229 for (pos_type i = beg; i < end; ++i) {
3230 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3232 char_type const c = d->text_[i];
3233 if (isPrintable(c) || c == '\t'
3234 || (c == '\n' && (options & AS_STR_NEWLINES)))
3236 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3237 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3238 LASSERT(runparams != 0, return docstring());
3239 getInset(i)->plaintext(os, *runparams);
3241 getInset(i)->toString(os);
3250 void Paragraph::forOutliner(docstring & os, size_t maxlen) const
3252 if (!d->params_.labelString().empty())
3253 os += d->params_.labelString() + ' ';
3254 for (pos_type i = 0; i < size() && os.length() < maxlen; ++i) {
3257 char_type const c = d->text_[i];
3260 else if (c == '\t' || c == '\n')
3262 else if (c == META_INSET)
3263 getInset(i)->forOutliner(os, maxlen);
3268 void Paragraph::setInsetOwner(Inset const * inset)
3270 d->inset_owner_ = inset;
3274 int Paragraph::id() const
3280 void Paragraph::setId(int id)
3286 Layout const & Paragraph::layout() const
3292 void Paragraph::setLayout(Layout const & layout)
3294 d->layout_ = &layout;
3298 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3300 setLayout(tc.defaultLayout());
3304 void Paragraph::setPlainLayout(DocumentClass const & tc)
3306 setLayout(tc.plainLayout());
3310 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3312 if (usePlainLayout())
3313 setPlainLayout(tclass);
3315 setDefaultLayout(tclass);
3319 Inset const & Paragraph::inInset() const
3321 LBUFERR(d->inset_owner_);
3322 return *d->inset_owner_;
3326 ParagraphParameters & Paragraph::params()
3332 ParagraphParameters const & Paragraph::params() const
3338 bool Paragraph::isFreeSpacing() const
3340 if (d->layout_->free_spacing)
3342 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3346 bool Paragraph::allowEmpty() const
3348 if (d->layout_->keepempty)
3350 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3354 bool Paragraph::brokenBiblio() const
3356 // there is a problem if there is no bibitem at position 0 or
3357 // if there is another bibitem in the paragraph.
3358 return d->layout_->labeltype == LABEL_BIBLIO
3359 && (d->insetlist_.find(BIBITEM_CODE) != 0
3360 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3364 int Paragraph::fixBiblio(Buffer const & buffer)
3366 // FIXME: What about the case where paragraph is not BIBLIO
3367 // but there is an InsetBibitem?
3368 // FIXME: when there was already an inset at 0, the return value is 1,
3369 // which does not tell whether another inset has been remove; the
3370 // cursor cannot be correctly updated.
3372 if (d->layout_->labeltype != LABEL_BIBLIO)
3375 bool const track_changes = buffer.params().track_changes;
3376 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3377 bool const hasbibitem0 = bibitem_pos == 0;
3380 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3381 // There was an InsetBibitem at pos 0, and no other one => OK
3382 if (bibitem_pos == -1)
3384 // there is a bibitem at the 0 position, but since
3385 // there is a second one, we copy the second on the
3386 // first. We're assuming there are at most two of
3387 // these, which there should be.
3388 // FIXME: why does it make sense to do that rather
3389 // than keep the first? (JMarc)
3390 Inset * inset = releaseInset(bibitem_pos);
3391 d->insetlist_.begin()->inset = inset;
3392 return -bibitem_pos;
3395 // We need to create an inset at the beginning
3397 if (bibitem_pos > 0) {
3398 // there was one somewhere in the paragraph, let's move it
3399 inset = d->insetlist_.release(bibitem_pos);
3400 eraseChar(bibitem_pos, track_changes);
3403 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3404 InsetCommandParams(BIBITEM_CODE));
3406 Font font(inherit_font, buffer.params().language);
3407 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3408 : Change::UNCHANGED));
3414 void Paragraph::checkAuthors(AuthorList const & authorList)
3416 d->changes_.checkAuthors(authorList);
3420 bool Paragraph::isChanged(pos_type pos) const
3422 return lookupChange(pos).changed();
3426 bool Paragraph::isInserted(pos_type pos) const
3428 return lookupChange(pos).inserted();
3432 bool Paragraph::isDeleted(pos_type pos) const
3434 return lookupChange(pos).deleted();
3438 InsetList const & Paragraph::insetList() const
3440 return d->insetlist_;
3444 void Paragraph::setBuffer(Buffer & b)
3446 d->insetlist_.setBuffer(b);
3450 Inset * Paragraph::releaseInset(pos_type pos)
3452 Inset * inset = d->insetlist_.release(pos);
3453 /// does not honour change tracking!
3454 eraseChar(pos, false);
3459 Inset * Paragraph::getInset(pos_type pos)
3461 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3462 ? d->insetlist_.get(pos) : 0;
3466 Inset const * Paragraph::getInset(pos_type pos) const
3468 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3469 ? d->insetlist_.get(pos) : 0;
3473 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3474 pos_type & right, TextCase action)
3476 // process sequences of modified characters; in change
3477 // tracking mode, this approach results in much better
3478 // usability than changing case on a char-by-char basis
3479 // We also need to track the current font, since font
3480 // changes within sequences can occur.
3481 vector<pair<char_type, Font> > changes;
3483 bool const trackChanges = bparams.track_changes;
3485 bool capitalize = true;
3487 for (; pos < right; ++pos) {
3488 char_type oldChar = d->text_[pos];
3489 char_type newChar = oldChar;
3491 // ignore insets and don't play with deleted text!
3492 if (oldChar != META_INSET && !isDeleted(pos)) {
3494 case text_lowercase:
3495 newChar = lowercase(oldChar);
3497 case text_capitalization:
3499 newChar = uppercase(oldChar);
3503 case text_uppercase:
3504 newChar = uppercase(oldChar);
3509 if (isWordSeparator(pos) || isDeleted(pos)) {
3510 // permit capitalization again
3514 if (oldChar != newChar) {
3515 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3516 if (pos != right - 1)
3518 // step behind the changing area
3522 int erasePos = pos - changes.size();
3523 for (size_t i = 0; i < changes.size(); i++) {
3524 insertChar(pos, changes[i].first,
3527 if (!eraseChar(erasePos, trackChanges)) {
3530 ++right; // expand selection
3538 int Paragraph::find(docstring const & str, bool cs, bool mw,
3539 pos_type start_pos, bool del) const
3541 pos_type pos = start_pos;
3542 int const strsize = str.length();
3544 pos_type const parsize = d->text_.size();
3545 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3546 // Ignore "invisible" letters such as ligature breaks
3547 // and hyphenation chars while searching
3548 while (pos < parsize - 1 && isInset(pos)) {
3549 odocstringstream os;
3550 getInset(pos)->toString(os);
3551 if (!getInset(pos)->isLetter() || !os.str().empty())
3555 if (cs && str[i] != d->text_[pos])
3557 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3559 if (!del && isDeleted(pos))
3566 // if necessary, check whether string matches word
3568 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3571 && !isWordSeparator(pos))
3575 return pos - start_pos;
3579 char_type Paragraph::getChar(pos_type pos) const
3581 return d->text_[pos];
3585 pos_type Paragraph::size() const
3587 return d->text_.size();
3591 bool Paragraph::empty() const
3593 return d->text_.empty();
3597 bool Paragraph::isInset(pos_type pos) const
3599 return d->text_[pos] == META_INSET;
3603 bool Paragraph::isSeparator(pos_type pos) const
3605 //FIXME: Are we sure this can be the only separator?
3606 return d->text_[pos] == ' ';
3610 void Paragraph::deregisterWords()
3612 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3613 Private::LangWordsMap::const_iterator ite = d->words_.end();
3614 for (; itl != ite; ++itl) {
3615 WordList * wl = theWordList(itl->first);
3616 Private::Words::const_iterator it = (itl->second).begin();
3617 Private::Words::const_iterator et = (itl->second).end();
3618 for (; it != et; ++it)
3625 void Paragraph::locateWord(pos_type & from, pos_type & to,
3626 word_location const loc) const
3629 case WHOLE_WORD_STRICT:
3630 if (from == 0 || from == size()
3631 || isWordSeparator(from)
3632 || isWordSeparator(from - 1)) {
3636 // no break here, we go to the next
3639 // If we are already at the beginning of a word, do nothing
3640 if (!from || isWordSeparator(from - 1))
3642 // no break here, we go to the next
3645 // always move the cursor to the beginning of previous word
3646 while (from && !isWordSeparator(from - 1))
3650 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3653 // no need to move the 'from' cursor
3657 while (to < size() && !isWordSeparator(to))
3662 void Paragraph::collectWords()
3664 for (pos_type pos = 0; pos < size(); ++pos) {
3665 if (isWordSeparator(pos))
3667 pos_type from = pos;
3668 locateWord(from, pos, WHOLE_WORD);
3669 // Work around MSVC warning: The statement
3670 // if (pos < from + lyxrc.completion_minlength)
3671 // triggers a signed vs. unsigned warning.
3672 // I don't know why this happens, it could be a MSVC bug, or
3673 // related to LLP64 (windows) vs. LP64 (unix) programming
3674 // model, or the C++ standard might be ambigous in the section
3675 // defining the "usual arithmetic conversions". However, using
3676 // a temporary variable is safe and works on all compilers.
3677 pos_type const endpos = from + lyxrc.completion_minlength;
3680 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
3681 if (cit == d->fontlist_.end())
3683 Language const * lang = cit->font().language();
3684 docstring const word = asString(from, pos, AS_STR_NONE);
3685 d->words_[lang->lang()].insert(word);
3690 void Paragraph::registerWords()
3692 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3693 Private::LangWordsMap::const_iterator ite = d->words_.end();
3694 for (; itl != ite; ++itl) {
3695 WordList * wl = theWordList(itl->first);
3696 Private::Words::const_iterator it = (itl->second).begin();
3697 Private::Words::const_iterator et = (itl->second).end();
3698 for (; it != et; ++it)
3704 void Paragraph::updateWords()
3712 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3714 SkipPositionsIterator begin = skips.begin();
3715 SkipPositions::iterator end = skips.end();
3716 if (pos > 0 && begin < end) {
3718 if (end->last == pos - 1) {
3723 skips.insert(end, FontSpan(pos, pos));
3727 Language * Paragraph::Private::locateSpellRange(
3728 pos_type & from, pos_type & to,
3729 SkipPositions & skips) const
3731 // skip leading white space
3732 while (from < to && owner_->isWordSeparator(from))
3734 // don't check empty range
3737 // get current language
3738 Language * lang = getSpellLanguage(from);
3739 pos_type last = from;
3740 bool samelang = true;
3741 bool sameinset = true;
3742 while (last < to && samelang && sameinset) {
3743 // hop to end of word
3744 while (last < to && !owner_->isWordSeparator(last)) {
3745 if (owner_->getInset(last)) {
3746 appendSkipPosition(skips, last);
3747 } else if (owner_->isDeleted(last)) {
3748 appendSkipPosition(skips, last);
3752 // hop to next word while checking for insets
3753 while (sameinset && last < to && owner_->isWordSeparator(last)) {
3754 if (Inset const * inset = owner_->getInset(last))
3755 sameinset = inset->isChar() && inset->isLetter();
3756 if (sameinset && owner_->isDeleted(last)) {
3757 appendSkipPosition(skips, last);
3762 if (sameinset && last < to) {
3763 // now check for language change
3764 samelang = lang == getSpellLanguage(last);
3767 // if language change detected backstep is needed
3775 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
3778 const_cast<Language *>(owner_->getFontSettings(
3779 inset_owner_->buffer().params(), from).language());
3780 if (lang == inset_owner_->buffer().params().language
3781 && !lyxrc.spellchecker_alt_lang.empty()) {
3783 string const lang_variety =
3784 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3785 lang->setCode(lang_code);
3786 lang->setVariety(lang_variety);
3792 void Paragraph::requestSpellCheck(pos_type pos)
3794 d->requestSpellCheck(pos);
3798 bool Paragraph::needsSpellCheck() const
3800 SpellChecker::ChangeNumber speller_change_number = 0;
3801 if (theSpellChecker())
3802 speller_change_number = theSpellChecker()->changeNumber();
3803 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
3804 d->speller_state_.needsCompleteRefresh(speller_change_number);
3806 return d->needsSpellCheck();
3810 bool Paragraph::Private::ignoreWord(docstring const & word) const
3812 // Ignore words with digits
3813 // FIXME: make this customizable
3814 // (note that some checkers ignore words with digits by default)
3815 docstring::const_iterator cit = word.begin();
3816 docstring::const_iterator const end = word.end();
3817 for (; cit != end; ++cit) {
3818 if (isNumber((*cit)))
3825 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
3826 WordLangTuple & wl, docstring_list & suggestions,
3827 bool do_suggestion, bool check_learned) const
3829 SpellChecker::Result result = SpellChecker::WORD_OK;
3830 SpellChecker * speller = theSpellChecker();
3834 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
3837 locateWord(from, to, WHOLE_WORD);
3838 if (from == to || from >= size())
3841 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3842 Language * lang = d->getSpellLanguage(from);
3844 wl = WordLangTuple(word, lang);
3849 if (needsSpellCheck() || check_learned) {
3851 if (!d->ignoreWord(word)) {
3852 bool const trailing_dot = to < size() && d->text_[to] == '.';
3853 result = speller->check(wl);
3854 if (SpellChecker::misspelled(result) && trailing_dot) {
3855 wl = WordLangTuple(word.append(from_ascii(".")), lang);
3856 result = speller->check(wl);
3857 if (!SpellChecker::misspelled(result)) {
3858 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
3860 from << ".." << to << "]");
3862 // spell check with dot appended failed too
3863 // restore original word/lang value
3864 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3865 wl = WordLangTuple(word, lang);
3869 if (!SpellChecker::misspelled(result)) {
3870 // area up to the begin of the next word is not misspelled
3871 while (end < size() && isWordSeparator(end))
3874 d->setMisspelled(from, end, result);
3876 result = d->speller_state_.getState(from);
3880 suggestions.clear();
3882 if (SpellChecker::misspelled(result)) {
3883 LYXERR(Debug::GUI, "misspelled word: \"" <<
3885 from << ".." << to << "]");
3887 speller->suggest(wl, suggestions);
3893 void Paragraph::Private::markMisspelledWords(
3894 pos_type const & first, pos_type const & last,
3895 SpellChecker::Result result,
3896 docstring const & word,
3897 SkipPositions const & skips)
3899 if (!SpellChecker::misspelled(result)) {
3900 setMisspelled(first, last, SpellChecker::WORD_OK);
3904 SpellChecker * speller = theSpellChecker();
3905 // locate and enumerate the error positions
3906 int nerrors = speller->numMisspelledWords();
3908 SkipPositionsIterator it = skips.begin();
3909 SkipPositionsIterator et = skips.end();
3910 for (int index = 0; index < nerrors; ++index) {
3913 speller->misspelledWord(index, wstart, wlen);
3914 /// should not happen if speller supports range checks
3915 if (!wlen) continue;
3916 docstring const misspelled = word.substr(wstart, wlen);
3917 wstart += first + numskipped;
3918 if (snext < wstart) {
3919 /// mark the range of correct spelling
3920 numskipped += countSkips(it, et, wstart);
3921 setMisspelled(snext,
3922 wstart - 1, SpellChecker::WORD_OK);
3924 snext = wstart + wlen;
3925 numskipped += countSkips(it, et, snext);
3926 /// mark the range of misspelling
3927 setMisspelled(wstart, snext, result);
3928 LYXERR(Debug::GUI, "misspelled word: \"" <<
3929 misspelled << "\" [" <<
3930 wstart << ".." << (snext-1) << "]");
3933 if (snext <= last) {
3934 /// mark the range of correct spelling at end
3935 setMisspelled(snext, last, SpellChecker::WORD_OK);
3940 void Paragraph::spellCheck() const
3942 SpellChecker * speller = theSpellChecker();
3943 if (!speller || empty() ||!needsSpellCheck())
3947 d->rangeOfSpellCheck(start, endpos);
3948 if (speller->canCheckParagraph()) {
3949 // loop until we leave the range
3950 for (pos_type first = start; first < endpos; ) {
3951 pos_type last = endpos;
3952 Private::SkipPositions skips;
3953 Language * lang = d->locateSpellRange(first, last, skips);
3954 if (first >= endpos)
3956 // start the spell checker on the unit of meaning
3957 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
3958 WordLangTuple wl = WordLangTuple(word, lang);
3959 SpellChecker::Result result = word.size() ?
3960 speller->check(wl) : SpellChecker::WORD_OK;
3961 d->markMisspelledWords(first, last, result, word, skips);
3965 static docstring_list suggestions;
3966 pos_type to = endpos;
3967 while (start < endpos) {
3969 spellCheck(start, to, wl, suggestions, false);
3973 d->readySpellCheck();
3977 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
3979 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
3980 if (result || pos <= 0 || pos > size())
3982 if (check_boundary && (pos == size() || isWordSeparator(pos)))
3983 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
3988 string Paragraph::magicLabel() const
3991 ss << "magicparlabel-" << id();