3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Richard Heck (XHTML output)
9 * \author Jean-Marc Lasgouttes
10 * \author Angus Leeming
12 * \author André Pönitz
14 * \author Jürgen Vigna
16 * Full author contact details are available in file CREDITS.
21 #include "Paragraph.h"
23 #include "LayoutFile.h"
25 #include "BufferParams.h"
29 #include "InsetList.h"
31 #include "LaTeXFeatures.h"
37 #include "OutputParams.h"
38 #include "output_latex.h"
39 #include "output_xhtml.h"
40 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
43 #include "TextClass.h"
47 #include "WordLangTuple.h"
50 #include "frontends/alert.h"
52 #include "insets/InsetBibitem.h"
53 #include "insets/InsetLabel.h"
54 #include "insets/InsetSpecialChar.h"
56 #include "support/debug.h"
57 #include "support/docstring_list.h"
58 #include "support/ExceptionMessage.h"
59 #include "support/gettext.h"
60 #include "support/lassert.h"
61 #include "support/lstrings.h"
62 #include "support/textutils.h"
68 using namespace lyx::support;
73 /// Inset identifier (above 0x10ffff, for ucs-4)
74 char_type const META_INSET = 0x200001;
78 /////////////////////////////////////////////////////////////////////
82 /////////////////////////////////////////////////////////////////////
84 class SpellResultRange {
86 SpellResultRange(FontSpan range, SpellChecker::Result result)
87 : range_(range), result_(result)
90 FontSpan const & range() const { return range_; }
92 void range(FontSpan const & r) { range_ = r; }
94 SpellChecker::Result result() const { return result_; }
96 void result(SpellChecker::Result r) { result_ = r; }
98 bool inside(pos_type pos) const { return range_.inside(pos); }
100 bool covered(FontSpan const & r) const
102 // 1. first of new range inside current range or
103 // 2. last of new range inside current range or
104 // 3. first of current range inside new range or
105 // 4. last of current range inside new range
106 return range_.inside(r.first) || range_.inside(r.last) ||
107 r.inside(range_.first) || r.inside(range_.last);
110 void shift(pos_type pos, int offset)
112 if (range_.first > pos) {
113 range_.first += offset;
114 range_.last += offset;
115 } else if (range_.last > pos) {
116 range_.last += offset;
121 SpellChecker::Result result_ ;
125 /////////////////////////////////////////////////////////////////////
129 /////////////////////////////////////////////////////////////////////
131 class SpellCheckerState {
133 SpellCheckerState() {
134 needs_refresh_ = true;
135 current_change_number_ = 0;
138 void setRange(FontSpan const fp, SpellChecker::Result state)
141 RangesIterator et = ranges_.end();
142 RangesIterator it = ranges_.begin();
143 for (; it != et; ++it) {
144 if (!it->covered(fp))
145 result.push_back(SpellResultRange(it->range(), it->result()));
146 else if (state == SpellChecker::WORD_OK) {
147 // trim or split the current misspelled range
148 // store misspelled ranges only
149 FontSpan range = it->range();
150 if (fp.first > range.first) {
151 // misspelled area in front of WORD_OK
152 range.last = fp.first - 1;
153 result.push_back(SpellResultRange(range, it->result()));
156 if (fp.last < range.last) {
157 // misspelled area after WORD_OK range
158 range.first = fp.last + 1;
159 result.push_back(SpellResultRange(range, it->result()));
164 if (state != SpellChecker::WORD_OK)
165 ranges_.push_back(SpellResultRange(fp, state));
168 void increasePosAfterPos(pos_type pos)
170 correctRangesAfterPos(pos, 1);
174 void decreasePosAfterPos(pos_type pos)
176 correctRangesAfterPos(pos, -1);
180 void refreshLast(pos_type pos)
182 if (pos < refresh_.last)
186 SpellChecker::Result getState(pos_type pos) const
188 SpellChecker::Result result = SpellChecker::WORD_OK;
189 RangesIterator et = ranges_.end();
190 RangesIterator it = ranges_.begin();
191 for (; it != et; ++it) {
192 if(it->inside(pos)) {
199 FontSpan const & getRange(pos_type pos) const
201 /// empty span to indicate mismatch
202 static FontSpan empty_;
203 RangesIterator et = ranges_.end();
204 RangesIterator it = ranges_.begin();
205 for (; it != et; ++it) {
206 if(it->inside(pos)) {
213 bool needsRefresh() const {
214 return needs_refresh_;
217 SpellChecker::ChangeNumber currentChangeNumber() const {
218 return current_change_number_;
221 void refreshRange(pos_type & first, pos_type & last) const {
222 first = refresh_.first;
223 last = refresh_.last;
226 void needsRefresh(pos_type pos) {
227 if (needs_refresh_ && pos != -1) {
228 if (pos < refresh_.first)
229 refresh_.first = pos;
230 if (pos > refresh_.last)
232 } else if (pos != -1) {
233 // init request check for neighbour positions too
234 refresh_.first = pos > 0 ? pos - 1 : 0;
235 // no need for special end of paragraph check
236 refresh_.last = pos + 1;
238 needs_refresh_ = pos != -1;
241 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number) {
242 needs_refresh_ = true;
245 current_change_number_ = change_number;
249 typedef vector<SpellResultRange> Ranges;
250 typedef Ranges::const_iterator RangesIterator;
252 /// the area of the paragraph with pending spell check
255 /// spell state cache version number
256 SpellChecker::ChangeNumber current_change_number_;
259 void correctRangesAfterPos(pos_type pos, int offset)
261 RangesIterator et = ranges_.end();
262 Ranges::iterator it = ranges_.begin();
263 for (; it != et; ++it) {
264 it->shift(pos, offset);
270 /////////////////////////////////////////////////////////////////////
272 // Paragraph::Private
274 /////////////////////////////////////////////////////////////////////
276 class Paragraph::Private
280 Private(Paragraph * owner, Layout const & layout);
281 /// "Copy constructor"
282 Private(Private const &, Paragraph * owner);
283 /// Copy constructor from \p beg to \p end
284 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
287 void insertChar(pos_type pos, char_type c, Change const & change);
289 /// Output the surrogate pair formed by \p c and \p next to \p os.
290 /// \return the number of characters written.
291 int latexSurrogatePair(odocstream & os, char_type c, char_type next,
292 OutputParams const &);
294 /// Output a space in appropriate formatting (or a surrogate pair
295 /// if the next character is a combining character).
296 /// \return whether a surrogate pair was output.
297 bool simpleTeXBlanks(OutputParams const &,
298 odocstream &, TexRow & texrow,
300 unsigned int & column,
302 Layout const & style);
304 /// Output consecutive unicode chars, belonging to the same script as
305 /// specified by the latex macro \p ltx, to \p os starting from \p i.
306 /// \return the number of characters written.
307 int writeScriptChars(odocstream & os, docstring const & ltx,
308 Change const &, Encoding const &, pos_type & i);
310 /// This could go to ParagraphParameters if we want to.
311 int startTeXParParams(BufferParams const &, odocstream &, TexRow &,
312 OutputParams const &) const;
314 /// This could go to ParagraphParameters if we want to.
315 int endTeXParParams(BufferParams const &, odocstream &, TexRow &,
316 OutputParams const &) const;
319 void latexInset(BufferParams const &,
321 TexRow & texrow, OutputParams &,
324 Font const & outerfont,
326 Change & running_change,
327 Layout const & style,
329 unsigned int & column);
332 void latexSpecialChar(
334 OutputParams const & runparams,
335 Font const & running_font,
336 Change const & running_change,
337 Layout const & style,
339 unsigned int & column);
346 unsigned int & column);
348 bool latexSpecialTypewriter(
352 unsigned int & column);
354 bool latexSpecialPhrase(
357 unsigned int & column,
358 OutputParams const & runparams);
361 void validate(LaTeXFeatures & features) const;
363 /// Checks if the paragraph contains only text and no inset or font change.
364 bool onlyText(Buffer const & buf, Font const & outerfont,
365 pos_type initial) const;
367 /// match a string against a particular point in the paragraph
368 bool isTextAt(string const & str, pos_type pos) const;
370 /// a vector of speller skip positions
371 typedef vector<FontSpan> SkipPositions;
372 typedef SkipPositions::const_iterator SkipPositionsIterator;
374 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
376 Language * getSpellLanguage(pos_type const from) const;
378 Language * locateSpellRange(pos_type & from, pos_type & to,
379 SkipPositions & skips) const;
381 bool hasSpellerChange() const {
382 SpellChecker::ChangeNumber speller_change_number = 0;
383 if (theSpellChecker())
384 speller_change_number = theSpellChecker()->changeNumber();
385 return speller_change_number > speller_state_.currentChangeNumber();
388 bool ignoreWord(docstring const & word) const ;
390 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
392 pos_type textsize = owner_->size();
393 // check for sane arguments
394 if (to < from || from >= textsize)
396 FontSpan fp = FontSpan(from, to);
397 // don't mark end of paragraph
398 if (fp.last >= textsize)
399 fp.last = textsize - 1;
400 speller_state_.setRange(fp, state);
403 void requestSpellCheck(pos_type pos) {
404 speller_state_.needsRefresh(pos);
407 void readySpellCheck() {
408 speller_state_.needsRefresh(-1);
411 bool needsSpellCheck() const
413 return speller_state_.needsRefresh();
416 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
418 speller_state_.refreshRange(first, last);
420 last = owner_->size();
423 pos_type endpos = last;
424 owner_->locateWord(first, endpos, WHOLE_WORD);
427 owner_->locateWord(last, endpos, WHOLE_WORD);
432 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
436 while (it != et && it->first < start) {
437 int skip = it->last - it->first + 1;
445 void markMisspelledWords(pos_type const & first, pos_type const & last,
446 SpellChecker::Result result,
447 docstring const & word,
448 SkipPositions const & skips);
450 InsetCode ownerCode() const
452 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
455 /// Which Paragraph owns us?
459 Inset const * inset_owner_;
468 ParagraphParameters params_;
470 /// for recording and looking up changes
474 InsetList insetlist_;
477 pos_type begin_of_body_;
479 typedef docstring TextContainer;
483 typedef set<docstring> Words;
484 typedef map<Language, Words> LangWordsMap;
488 Layout const * layout_;
490 SpellCheckerState speller_state_;
496 struct special_phrase {
502 special_phrase const special_phrases[] = {
503 { "LyX", from_ascii("\\LyX{}"), false },
504 { "TeX", from_ascii("\\TeX{}"), true },
505 { "LaTeX2e", from_ascii("\\LaTeXe{}"), true },
506 { "LaTeX", from_ascii("\\LaTeX{}"), true },
509 size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase);
514 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
515 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
521 // Initialization of the counter for the paragraph id's,
523 // FIXME: There should be a more intelligent way to generate and use the
524 // paragraph ids per buffer instead a global static counter for all InsetText
525 // in the running program.
526 static int paragraph_id = -1;
528 Paragraph::Private::Private(Private const & p, Paragraph * owner)
529 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
530 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
531 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
534 id_ = ++paragraph_id;
535 requestSpellCheck(p.text_.size());
539 Paragraph::Private::Private(Private const & p, Paragraph * owner,
540 pos_type beg, pos_type end)
541 : owner_(owner), inset_owner_(p.inset_owner_),
542 params_(p.params_), changes_(p.changes_),
543 insetlist_(p.insetlist_, beg, end),
544 begin_of_body_(p.begin_of_body_), words_(p.words_),
547 id_ = ++paragraph_id;
548 if (beg >= pos_type(p.text_.size()))
550 text_ = p.text_.substr(beg, end - beg);
552 FontList::const_iterator fcit = fontlist_.begin();
553 FontList::const_iterator fend = fontlist_.end();
554 for (; fcit != fend; ++fcit) {
555 if (fcit->pos() < beg)
557 if (fcit->pos() >= end) {
558 // Add last entry in the fontlist_.
559 fontlist_.set(text_.size() - 1, fcit->font());
562 // Add a new entry in the fontlist_.
563 fontlist_.set(fcit->pos() - beg, fcit->font());
565 requestSpellCheck(p.text_.size());
569 void Paragraph::addChangesToToc(DocIterator const & cdit,
570 Buffer const & buf) const
572 d->changes_.addToToc(cdit, buf);
576 bool Paragraph::isDeleted(pos_type start, pos_type end) const
578 LASSERT(start >= 0 && start <= size(), /**/);
579 LASSERT(end > start && end <= size() + 1, /**/);
581 return d->changes_.isDeleted(start, end);
585 bool Paragraph::isChanged(pos_type start, pos_type end) const
587 LASSERT(start >= 0 && start <= size(), /**/);
588 LASSERT(end > start && end <= size() + 1, /**/);
590 return d->changes_.isChanged(start, end);
594 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
596 // keep the logic here in sync with the logic of eraseChars()
600 Change const change = d->changes_.lookup(size());
601 return change.inserted() && change.currentAuthor();
605 void Paragraph::setChange(Change const & change)
607 // beware of the imaginary end-of-par character!
608 d->changes_.set(change, 0, size() + 1);
611 * Propagate the change recursively - but not in case of DELETED!
613 * Imagine that your co-author makes changes in an existing inset. He
614 * sends your document to you and you come to the conclusion that the
615 * inset should go completely. If you erase it, LyX must not delete all
616 * text within the inset. Otherwise, the change tracked insertions of
617 * your co-author get lost and there is no way to restore them later.
619 * Conclusion: An inset's content should remain untouched if you delete it
622 if (!change.deleted()) {
623 for (pos_type pos = 0; pos < size(); ++pos) {
624 if (Inset * inset = getInset(pos))
625 inset->setChange(change);
631 void Paragraph::setChange(pos_type pos, Change const & change)
633 LASSERT(pos >= 0 && pos <= size(), /**/);
634 d->changes_.set(change, pos);
636 // see comment in setChange(Change const &) above
637 if (!change.deleted() && pos < size())
638 if (Inset * inset = getInset(pos))
639 inset->setChange(change);
643 Change const & Paragraph::lookupChange(pos_type pos) const
645 LASSERT(pos >= 0 && pos <= size(), /**/);
646 return d->changes_.lookup(pos);
650 void Paragraph::acceptChanges(pos_type start, pos_type end)
652 LASSERT(start >= 0 && start <= size(), /**/);
653 LASSERT(end > start && end <= size() + 1, /**/);
655 for (pos_type pos = start; pos < end; ++pos) {
656 switch (lookupChange(pos).type) {
657 case Change::UNCHANGED:
658 // accept changes in nested inset
659 if (Inset * inset = getInset(pos))
660 inset->acceptChanges();
663 case Change::INSERTED:
664 d->changes_.set(Change(Change::UNCHANGED), pos);
665 // also accept changes in nested inset
666 if (Inset * inset = getInset(pos))
667 inset->acceptChanges();
670 case Change::DELETED:
671 // Suppress access to non-existent
672 // "end-of-paragraph char"
674 eraseChar(pos, false);
685 void Paragraph::rejectChanges(pos_type start, pos_type end)
687 LASSERT(start >= 0 && start <= size(), /**/);
688 LASSERT(end > start && end <= size() + 1, /**/);
690 for (pos_type pos = start; pos < end; ++pos) {
691 switch (lookupChange(pos).type) {
692 case Change::UNCHANGED:
693 // reject changes in nested inset
694 if (Inset * inset = getInset(pos))
695 inset->rejectChanges();
698 case Change::INSERTED:
699 // Suppress access to non-existent
700 // "end-of-paragraph char"
702 eraseChar(pos, false);
708 case Change::DELETED:
709 d->changes_.set(Change(Change::UNCHANGED), pos);
711 // Do NOT reject changes within a deleted inset!
712 // There may be insertions of a co-author inside of it!
720 void Paragraph::Private::insertChar(pos_type pos, char_type c,
721 Change const & change)
723 LASSERT(pos >= 0 && pos <= int(text_.size()), /**/);
726 changes_.insert(change, pos);
728 // This is actually very common when parsing buffers (and
729 // maybe inserting ascii text)
730 if (pos == pos_type(text_.size())) {
731 // when appending characters, no need to update tables
733 // but we want spell checking
734 requestSpellCheck(pos);
738 text_.insert(text_.begin() + pos, c);
740 // Update the font table.
741 fontlist_.increasePosAfterPos(pos);
744 insetlist_.increasePosAfterPos(pos);
746 // Update list of misspelled positions
747 speller_state_.increasePosAfterPos(pos);
751 bool Paragraph::insertInset(pos_type pos, Inset * inset,
752 Change const & change)
754 LASSERT(inset, /**/);
755 LASSERT(pos >= 0 && pos <= size(), /**/);
757 // Paragraph::insertInset() can be used in cut/copy/paste operation where
758 // d->inset_owner_ is not set yet.
759 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
762 d->insertChar(pos, META_INSET, change);
763 LASSERT(d->text_[pos] == META_INSET, /**/);
765 // Add a new entry in the insetlist_.
766 d->insetlist_.insert(inset, pos);
768 // Some insets require run of spell checker
769 requestSpellCheck(pos);
774 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
776 LASSERT(pos >= 0 && pos <= size(), return false);
778 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
781 Change change = d->changes_.lookup(pos);
783 // set the character to DELETED if
784 // a) it was previously unchanged or
785 // b) it was inserted by a co-author
787 if (!change.changed() ||
788 (change.inserted() && !change.currentAuthor())) {
789 setChange(pos, Change(Change::DELETED));
790 // request run of spell checker
791 requestSpellCheck(pos);
795 if (change.deleted())
799 // Don't physically access the imaginary end-of-paragraph character.
800 // eraseChar() can only mark it as DELETED. A physical deletion of
801 // end-of-par must be handled externally.
807 d->changes_.erase(pos);
809 // if it is an inset, delete the inset entry
810 if (d->text_[pos] == META_INSET)
811 d->insetlist_.erase(pos);
813 d->text_.erase(d->text_.begin() + pos);
815 // Update the fontlist_
816 d->fontlist_.erase(pos);
818 // Update the insetlist_
819 d->insetlist_.decreasePosAfterPos(pos);
821 // Update list of misspelled positions
822 d->speller_state_.decreasePosAfterPos(pos);
823 d->speller_state_.refreshLast(size());
829 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
831 LASSERT(start >= 0 && start <= size(), /**/);
832 LASSERT(end >= start && end <= size() + 1, /**/);
835 for (pos_type count = end - start; count; --count) {
836 if (!eraseChar(i, trackChanges))
843 int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
844 char_type next, OutputParams const & runparams)
846 // Writing next here may circumvent a possible font change between
847 // c and next. Since next is only output if it forms a surrogate pair
848 // with c we can ignore this:
849 // A font change inside a surrogate pair does not make sense and is
850 // hopefully impossible to input.
851 // FIXME: change tracking
852 // Is this correct WRT change tracking?
853 Encoding const & encoding = *(runparams.encoding);
854 docstring const latex1 = encoding.latexChar(next);
855 docstring const latex2 = encoding.latexChar(c);
856 if (docstring(1, next) == latex1) {
857 // the encoding supports the combination
858 os << latex2 << latex1;
859 return latex1.length() + latex2.length();
860 } else if (runparams.local_font &&
861 runparams.local_font->language()->lang() == "polutonikogreek") {
862 // polutonikogreek only works without the brackets
863 os << latex1 << latex2;
864 return latex1.length() + latex2.length();
866 os << latex1 << '{' << latex2 << '}';
867 return latex1.length() + latex2.length() + 2;
871 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
872 odocstream & os, TexRow & texrow,
874 unsigned int & column,
876 Layout const & style)
878 if (style.pass_thru || runparams.pass_thru)
881 if (i + 1 < int(text_.size())) {
882 char_type next = text_[i + 1];
883 if (Encodings::isCombiningChar(next)) {
884 // This space has an accent, so we must always output it.
885 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
890 if (runparams.linelen > 0
891 && column > runparams.linelen
893 && text_[i - 1] != ' '
894 && (i + 1 < int(text_.size()))
895 // same in FreeSpacing mode
896 && !owner_->isFreeSpacing()
897 // In typewriter mode, we want to avoid
898 // ! . ? : at the end of a line
899 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
900 && (text_[i - 1] == '.'
901 || text_[i - 1] == '?'
902 || text_[i - 1] == ':'
903 || text_[i - 1] == '!'))) {
906 texrow.start(owner_->id(), i + 1);
908 } else if (style.free_spacing) {
917 int Paragraph::Private::writeScriptChars(odocstream & os,
918 docstring const & ltx,
919 Change const & runningChange,
920 Encoding const & encoding,
923 // FIXME: modifying i here is not very nice...
925 // We only arrive here when a proper language for character text_[i] has
926 // not been specified (i.e., it could not be translated in the current
927 // latex encoding) or its latex translation has been forced, and it
928 // belongs to a known script.
929 // Parameter ltx contains the latex translation of text_[i] as specified
930 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
931 // The latex macro name "textXXX" specifies the script to which text_[i]
932 // belongs and we use it in order to check whether characters from the
933 // same script immediately follow, such that we can collect them in a
934 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
935 // for the first char but only "<spec>" for all subsequent chars.
936 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
937 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
938 string script = to_ascii(ltx.substr(1, brace1 - 1));
941 bool closing_brace = true;
942 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
943 // Correct encoding is being used, so we can avoid \textgreek.
946 closing_brace = false;
948 os << ltx.substr(pos, length);
949 int size = text_.size();
950 while (i + 1 < size) {
951 char_type const next = text_[i + 1];
952 // Stop here if next character belongs to another script
953 // or there is a change in change tracking status.
954 if (!Encodings::isKnownScriptChar(next, script) ||
955 runningChange != owner_->lookupChange(i + 1))
959 FontList::const_iterator cit = fontlist_.begin();
960 FontList::const_iterator end = fontlist_.end();
961 for (; cit != end; ++cit) {
962 if (cit->pos() >= i && !found) {
963 prev_font = cit->font();
966 if (cit->pos() >= i + 1)
969 // Stop here if there is a font attribute or encoding change.
970 if (found && cit != end && prev_font != cit->font())
972 docstring const latex = encoding.latexChar(next);
973 docstring::size_type const b1 =
974 latex.find_first_of(from_ascii("{"));
975 docstring::size_type const b2 =
976 latex.find_last_of(from_ascii("}"));
977 int const len = b2 - b1 - 1;
978 os << latex.substr(b1 + 1, len);
990 bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
992 pos_type const len = str.length();
994 // is the paragraph large enough?
995 if (pos + len > int(text_.size()))
998 // does the wanted text start at point?
999 for (string::size_type i = 0; i < str.length(); ++i) {
1000 // Caution: direct comparison of characters works only
1001 // because str is pure ASCII.
1002 if (str[i] != text_[pos + i])
1006 return fontlist_.hasChangeInRange(pos, len);
1010 void Paragraph::Private::latexInset(BufferParams const & bparams,
1013 OutputParams & runparams,
1014 Font & running_font,
1016 Font const & outerfont,
1018 Change & running_change,
1019 Layout const & style,
1021 unsigned int & column)
1023 Inset * inset = owner_->getInset(i);
1024 LASSERT(inset, /**/);
1026 if (style.pass_thru) {
1027 inset->plaintext(os, runparams);
1031 // FIXME: move this to InsetNewline::latex
1032 if (inset->lyxCode() == NEWLINE_CODE) {
1033 // newlines are handled differently here than
1034 // the default in simpleTeXSpecialChars().
1035 if (!style.newline_allowed) {
1039 column += running_font.latexWriteEndChanges(
1040 os, bparams, runparams,
1041 basefont, basefont);
1045 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1048 basefont = owner_->getLayoutFont(bparams, outerfont);
1049 running_font = basefont;
1051 if (runparams.moving_arg)
1056 texrow.start(owner_->id(), i + 1);
1060 if (owner_->isDeleted(i)) {
1061 if( ++runparams.inDeletedInset == 1)
1062 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1065 if (inset->canTrackChanges()) {
1066 column += Changes::latexMarkChange(os, bparams, running_change,
1067 Change(Change::UNCHANGED), runparams);
1068 running_change = Change(Change::UNCHANGED);
1072 odocstream::pos_type const len = os.tellp();
1074 if (inset->forceLTR()
1075 && running_font.isRightToLeft()
1076 // ERT is an exception, it should be output with no
1077 // decorations at all
1078 && inset->lyxCode() != ERT_CODE) {
1079 if (running_font.language()->lang() == "farsi")
1086 // FIXME: Bug: we can have an empty font change here!
1087 // if there has just been a font change, we are going to close it
1088 // right now, which means stupid latex code like \textsf{}. AFAIK,
1089 // this does not harm dvi output. A minor bug, thus (JMarc)
1091 // Some insets cannot be inside a font change command.
1092 // However, even such insets *can* be placed in \L or \R
1093 // or their equivalents (for RTL language switches), so we don't
1094 // close the language in those cases.
1095 // ArabTeX, though, cannot handle this special behavior, it seems.
1096 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1097 || running_font.language()->lang() == "arabic_arabtex";
1098 if (open_font && inset->noFontChange()) {
1099 bool closeLanguage = arabtex
1100 || basefont.isRightToLeft() == running_font.isRightToLeft();
1101 unsigned int count = running_font.latexWriteEndChanges(os,
1102 bparams, runparams, basefont, basefont, closeLanguage);
1104 // if any font properties were closed, update the running_font,
1105 // making sure, however, to leave the language as it was
1107 // FIXME: probably a better way to keep track of the old
1108 // language, than copying the entire font?
1109 Font const copy_font(running_font);
1110 basefont = owner_->getLayoutFont(bparams, outerfont);
1111 running_font = basefont;
1113 running_font.setLanguage(copy_font.language());
1114 // leave font open if language is still open
1115 open_font = (running_font.language() == basefont.language());
1117 runparams.local_font = &basefont;
1124 tmp = inset->latex(os, runparams);
1125 } catch (EncodingException & e) {
1126 // add location information and throw again.
1133 if (running_font.language()->lang() == "farsi")
1140 texrow.newlines(tmp);
1141 texrow.start(owner_->id(), i + 1);
1144 column += (unsigned int)(os.tellp() - len);
1147 if (owner_->isDeleted(i))
1148 --runparams.inDeletedInset;
1152 void Paragraph::Private::latexSpecialChar(
1154 OutputParams const & runparams,
1155 Font const & running_font,
1156 Change const & running_change,
1157 Layout const & style,
1159 unsigned int & column)
1161 char_type const c = text_[i];
1163 if (style.pass_thru || runparams.pass_thru) {
1165 // FIXME UNICODE: This can fail if c cannot
1166 // be encoded in the current encoding.
1171 // If T1 font encoding is used, use the special
1172 // characters it provides.
1173 // NOTE: some languages reset the font encoding
1175 if (!running_font.language()->internalFontEncoding()
1176 && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
1179 // \tt font needs special treatment
1180 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
1181 && latexSpecialTypewriter(c, os, i, column))
1184 // Otherwise, we use what LaTeX provides us.
1187 os << "\\textbackslash{}";
1191 os << "\\textless{}";
1195 os << "\\textgreater{}";
1199 os << "\\textbar{}";
1206 os << "\\char`\\\"{}";
1211 case '%': case '#': case '{':
1219 os << "\\textasciitilde{}";
1224 os << "\\textasciicircum{}";
1231 // avoid being mistaken for optional arguments
1239 // Blanks are printed before font switching.
1240 // Sure? I am not! (try nice-latex)
1241 // I am sure it's correct. LyX might be smarter
1242 // in the future, but for now, nothing wrong is
1248 if (latexSpecialPhrase(os, i, column, runparams))
1254 Encoding const & encoding = *(runparams.encoding);
1255 if (i + 1 < int(text_.size())) {
1256 char_type next = text_[i + 1];
1257 if (Encodings::isCombiningChar(next)) {
1258 column += latexSurrogatePair(os, c, next, runparams) - 1;
1264 docstring const latex = encoding.latexChar(c);
1265 if (Encodings::isKnownScriptChar(c, script)
1266 && prefixIs(latex, from_ascii("\\" + script)))
1267 column += writeScriptChars(os, latex,
1268 running_change, encoding, i) - 1;
1269 else if (latex.length() > 1 && latex[latex.length() - 1] != '}') {
1270 // Prevent eating of a following
1271 // space or command corruption by
1272 // following characters
1273 column += latex.length() + 1;
1274 os << latex << "{}";
1276 column += latex.length() - 1;
1284 bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
1285 pos_type i, unsigned int & column)
1291 // In T1 encoding, these characters exist
1292 // but we should avoid ligatures
1293 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1295 os << "\\textcompwordmark{}";
1302 // soul.sty breaks with \char`\"
1303 os << "\\textquotedbl{}";
1312 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream & os,
1313 pos_type i, unsigned int & column)
1317 // within \ttfamily, "--" is merged to "-" (no endash)
1318 // so we avoid this rather irritating ligature
1319 if (i + 1 < int(text_.size()) && text_[i + 1] == '-') {
1326 // everything else has to be checked separately
1327 // (depending on the encoding)
1334 bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
1335 unsigned int & column, OutputParams const & runparams)
1337 // FIXME: if we have "LaTeX" with a font
1338 // change in the middle (before the 'T', then
1339 // the "TeX" part is still special cased.
1340 // Really we should only operate this on
1341 // "words" for some definition of word
1343 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1344 if (!isTextAt(special_phrases[pnr].phrase, i))
1346 if (runparams.moving_arg)
1348 os << special_phrases[pnr].macro;
1349 i += special_phrases[pnr].phrase.length() - 1;
1350 column += special_phrases[pnr].macro.length() - 1;
1357 void Paragraph::Private::validate(LaTeXFeatures & features) const
1359 if (layout_->inpreamble && inset_owner_) {
1360 bool const is_command = layout_->latextype == LATEX_COMMAND;
1361 Buffer const & buf = inset_owner_->buffer();
1362 BufferParams const & bp = buf.params();
1365 // Using a string stream here circumvents the encoding
1366 // switching machinery of odocstream. Therefore the
1367 // output is wrong if this paragraph contains content
1368 // that needs to switch encoding.
1369 odocstringstream ods;
1371 ods << '\\' << from_ascii(layout_->latexname());
1372 // we have to provide all the optional arguments here, even though
1373 // the last one is the only one we care about.
1374 // Separate handling of optional argument inset.
1375 if (layout_->optargs != 0 || layout_->reqargs != 0)
1376 latexArgInsets(*owner_, ods, features.runparams(),
1377 layout_->reqargs, layout_->optargs);
1379 ods << from_ascii(layout_->latexparam());
1381 docstring::size_type const length = ods.str().length();
1382 // this will output "{" at the beginning, but not at the end
1383 owner_->latex(bp, f, ods, tr, features.runparams(), 0, -1, true);
1384 if (ods.str().length() > length) {
1387 string const snippet = to_utf8(ods.str());
1388 features.addPreambleSnippet(snippet);
1392 if (features.runparams().flavor == OutputParams::HTML
1393 && layout_->htmltitle()) {
1394 features.setHTMLTitle(owner_->asString(AS_STR_INSETS));
1397 // check the params.
1398 if (!params_.spacing().isDefault())
1399 features.require("setspace");
1402 features.useLayout(layout_->name());
1405 fontlist_.validate(features);
1407 // then the indentation
1408 if (!params_.leftIndent().zero())
1409 features.require("ParagraphLeftIndent");
1412 InsetList::const_iterator icit = insetlist_.begin();
1413 InsetList::const_iterator iend = insetlist_.end();
1414 for (; icit != iend; ++icit) {
1416 icit->inset->validate(features);
1417 if (layout_->needprotect &&
1418 icit->inset->lyxCode() == FOOT_CODE)
1419 features.require("NeedLyXFootnoteCode");
1423 // then the contents
1424 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1425 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1426 if (!special_phrases[pnr].builtin
1427 && isTextAt(special_phrases[pnr].phrase, i)) {
1428 features.require(special_phrases[pnr].phrase);
1432 Encodings::validate(text_[i], features);
1436 /////////////////////////////////////////////////////////////////////
1440 /////////////////////////////////////////////////////////////////////
1443 Layout const emptyParagraphLayout;
1446 Paragraph::Paragraph()
1447 : d(new Paragraph::Private(this, emptyParagraphLayout))
1454 Paragraph::Paragraph(Paragraph const & par)
1455 : itemdepth(par.itemdepth),
1456 d(new Paragraph::Private(*par.d, this))
1462 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1463 : itemdepth(par.itemdepth),
1464 d(new Paragraph::Private(*par.d, this, beg, end))
1470 Paragraph & Paragraph::operator=(Paragraph const & par)
1472 // needed as we will destroy the private part before copying it
1474 itemdepth = par.itemdepth;
1478 d = new Private(*par.d, this);
1485 Paragraph::~Paragraph()
1494 // this shall be called just before every "os << ..." action.
1495 void flushString(ostream & os, docstring & s)
1504 void Paragraph::write(ostream & os, BufferParams const & bparams,
1505 depth_type & dth) const
1507 // The beginning or end of a deeper (i.e. nested) area?
1508 if (dth != d->params_.depth()) {
1509 if (d->params_.depth() > dth) {
1510 while (d->params_.depth() > dth) {
1511 os << "\n\\begin_deeper";
1515 while (d->params_.depth() < dth) {
1516 os << "\n\\end_deeper";
1522 // First write the layout
1523 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1525 d->params_.write(os);
1527 Font font1(inherit_font, bparams.language);
1529 Change running_change = Change(Change::UNCHANGED);
1531 // this string is used as a buffer to avoid repetitive calls
1532 // to to_utf8(), which turn out to be expensive (JMarc)
1533 docstring write_buffer;
1536 for (pos_type i = 0; i <= size(); ++i) {
1538 Change const change = lookupChange(i);
1539 if (change != running_change)
1540 flushString(os, write_buffer);
1541 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1542 running_change = change;
1547 // Write font changes
1548 Font font2 = getFontSettings(bparams, i);
1549 if (font2 != font1) {
1550 flushString(os, write_buffer);
1551 font2.lyxWriteChanges(font1, os);
1556 char_type const c = d->text_[i];
1559 if (Inset const * inset = getInset(i)) {
1560 flushString(os, write_buffer);
1561 if (inset->directWrite()) {
1562 // international char, let it write
1563 // code directly so it's shorter in
1569 os << "\\begin_inset ";
1571 os << "\n\\end_inset\n\n";
1577 flushString(os, write_buffer);
1578 os << "\n\\backslash\n";
1582 flushString(os, write_buffer);
1583 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1590 if ((column > 70 && c == ' ')
1592 flushString(os, write_buffer);
1596 // this check is to amend a bug. LyX sometimes
1597 // inserts '\0' this could cause problems.
1599 write_buffer.push_back(c);
1601 LYXERR0("NUL char in structure.");
1607 flushString(os, write_buffer);
1608 os << "\n\\end_layout\n";
1612 void Paragraph::validate(LaTeXFeatures & features) const
1614 d->validate(features);
1618 void Paragraph::insert(pos_type start, docstring const & str,
1619 Font const & font, Change const & change)
1621 for (size_t i = 0, n = str.size(); i != n ; ++i)
1622 insertChar(start + i, str[i], font, change);
1626 void Paragraph::appendChar(char_type c, Font const & font,
1627 Change const & change)
1630 d->changes_.insert(change, d->text_.size());
1631 // when appending characters, no need to update tables
1632 d->text_.push_back(c);
1633 setFont(d->text_.size() - 1, font);
1637 void Paragraph::appendString(docstring const & s, Font const & font,
1638 Change const & change)
1640 pos_type end = s.size();
1641 size_t oldsize = d->text_.size();
1642 size_t newsize = oldsize + end;
1643 size_t capacity = d->text_.capacity();
1644 if (newsize >= capacity)
1645 d->text_.reserve(max(capacity + 100, newsize));
1647 // when appending characters, no need to update tables
1650 // FIXME: Optimize this!
1651 for (size_t i = oldsize; i != newsize; ++i) {
1653 d->changes_.insert(change, 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 bool Paragraph::insertInset(pos_type pos, Inset * inset,
1686 Font const & font, Change const & change)
1688 bool const success = insertInset(pos, inset, change);
1689 // Set the font/language of the inset...
1695 void Paragraph::resetFonts(Font const & font)
1697 d->fontlist_.clear();
1698 d->fontlist_.set(0, font);
1699 d->fontlist_.set(d->text_.size() - 1, font);
1702 // Gets uninstantiated font setting at position.
1703 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1707 LYXERR0("pos: " << pos << " size: " << size());
1708 LASSERT(pos <= size(), /**/);
1711 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1712 if (cit != d->fontlist_.end())
1715 if (pos == size() && !empty())
1716 return getFontSettings(bparams, pos - 1);
1718 // Optimisation: avoid a full font instantiation if there is no
1719 // language change from previous call.
1720 static Font previous_font;
1721 static Language const * previous_lang = 0;
1722 Language const * lang = getParLanguage(bparams);
1723 if (lang != previous_lang) {
1724 previous_lang = lang;
1725 previous_font = Font(inherit_font, lang);
1727 return previous_font;
1731 FontSpan Paragraph::fontSpan(pos_type pos) const
1733 LASSERT(pos <= size(), /**/);
1736 FontList::const_iterator cit = d->fontlist_.begin();
1737 FontList::const_iterator end = d->fontlist_.end();
1738 for (; cit != end; ++cit) {
1739 if (cit->pos() >= pos) {
1740 if (pos >= beginOfBody())
1741 return FontSpan(max(start, beginOfBody()),
1744 return FontSpan(start,
1745 min(beginOfBody() - 1,
1748 start = cit->pos() + 1;
1751 // This should not happen, but if so, we take no chances.
1752 // LYXERR0("Paragraph::getEndPosOfFontSpan: This should not happen!");
1753 return FontSpan(pos, pos);
1757 // Gets uninstantiated font setting at position 0
1758 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1760 if (!empty() && !d->fontlist_.empty())
1761 return d->fontlist_.begin()->font();
1763 // Optimisation: avoid a full font instantiation if there is no
1764 // language change from previous call.
1765 static Font previous_font;
1766 static Language const * previous_lang = 0;
1767 if (bparams.language != previous_lang) {
1768 previous_lang = bparams.language;
1769 previous_font = Font(inherit_font, bparams.language);
1772 return previous_font;
1776 // Gets the fully instantiated font at a given position in a paragraph
1777 // This is basically the same function as Text::GetFont() in text2.cpp.
1778 // The difference is that this one is used for generating the LaTeX file,
1779 // and thus cosmetic "improvements" are disallowed: This has to deliver
1780 // the true picture of the buffer. (Asger)
1781 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1782 Font const & outerfont) const
1784 LASSERT(pos >= 0, /**/);
1786 Font font = getFontSettings(bparams, pos);
1788 pos_type const body_pos = beginOfBody();
1789 FontInfo & fi = font.fontInfo();
1791 fi.realize(d->layout_->labelfont);
1793 fi.realize(d->layout_->font);
1795 fi.realize(outerfont.fontInfo());
1796 fi.realize(bparams.getFont().fontInfo());
1802 Font const Paragraph::getLabelFont
1803 (BufferParams const & bparams, Font const & outerfont) const
1805 FontInfo tmpfont = d->layout_->labelfont;
1806 tmpfont.realize(outerfont.fontInfo());
1807 tmpfont.realize(bparams.getFont().fontInfo());
1808 return Font(tmpfont, getParLanguage(bparams));
1812 Font const Paragraph::getLayoutFont
1813 (BufferParams const & bparams, Font const & outerfont) const
1815 FontInfo tmpfont = d->layout_->font;
1816 tmpfont.realize(outerfont.fontInfo());
1817 tmpfont.realize(bparams.getFont().fontInfo());
1818 return Font(tmpfont, getParLanguage(bparams));
1822 /// Returns the height of the highest font in range
1823 FontSize Paragraph::highestFontInRange
1824 (pos_type startpos, pos_type endpos, FontSize def_size) const
1826 return d->fontlist_.highestInRange(startpos, endpos, def_size);
1830 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1832 char_type c = d->text_[pos];
1833 if (!lyxrc.rtl_support)
1863 if (uc != c && getFontSettings(bparams, pos).isRightToLeft())
1869 void Paragraph::setFont(pos_type pos, Font const & font)
1871 LASSERT(pos <= size(), /**/);
1873 // First, reduce font against layout/label font
1874 // Update: The setCharFont() routine in text2.cpp already
1875 // reduces font, so we don't need to do that here. (Asger)
1877 d->fontlist_.set(pos, font);
1881 void Paragraph::makeSameLayout(Paragraph const & par)
1883 d->layout_ = par.d->layout_;
1884 d->params_ = par.d->params_;
1888 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1890 if (isFreeSpacing())
1896 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1897 if (eraseChar(pos, trackChanges))
1903 return count > 0 || pos > 0;
1907 bool Paragraph::hasSameLayout(Paragraph const & par) const
1909 return par.d->layout_ == d->layout_
1910 && d->params_.sameLayout(par.d->params_);
1914 depth_type Paragraph::getDepth() const
1916 return d->params_.depth();
1920 depth_type Paragraph::getMaxDepthAfter() const
1922 if (d->layout_->isEnvironment())
1923 return d->params_.depth() + 1;
1925 return d->params_.depth();
1929 char Paragraph::getAlign() const
1931 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1932 return d->layout_->align;
1934 return d->params_.align();
1938 docstring const & Paragraph::labelString() const
1940 return d->params_.labelString();
1944 // the next two functions are for the manual labels
1945 docstring const Paragraph::getLabelWidthString() const
1947 if (d->layout_->margintype == MARGIN_MANUAL
1948 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
1949 return d->params_.labelWidthString();
1951 return _("Senseless with this layout!");
1955 void Paragraph::setLabelWidthString(docstring const & s)
1957 d->params_.labelWidthString(s);
1961 docstring Paragraph::expandLabel(Layout const & layout,
1962 BufferParams const & bparams) const
1964 return expandParagraphLabel(layout, bparams, true);
1968 docstring Paragraph::expandDocBookLabel(Layout const & layout,
1969 BufferParams const & bparams) const
1971 return expandParagraphLabel(layout, bparams, false);
1975 docstring Paragraph::expandParagraphLabel(Layout const & layout,
1976 BufferParams const & bparams, bool process_appendix) const
1978 DocumentClass const & tclass = bparams.documentClass();
1979 string const & lang = getParLanguage(bparams)->code();
1980 bool const in_appendix = process_appendix && d->params_.appendix();
1981 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
1983 if (fmt.empty() && layout.labeltype == LABEL_COUNTER
1984 && !layout.counter.empty())
1985 return tclass.counters().theCounter(layout.counter, lang);
1987 // handle 'inherited level parts' in 'fmt',
1988 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
1989 size_t const i = fmt.find('@', 0);
1990 if (i != docstring::npos) {
1991 size_t const j = fmt.find('@', i + 1);
1992 if (j != docstring::npos) {
1993 docstring parent(fmt, i + 1, j - i - 1);
1994 docstring label = from_ascii("??");
1995 if (tclass.hasLayout(parent))
1996 docstring label = expandParagraphLabel(tclass[parent], bparams,
1998 fmt = docstring(fmt, 0, i) + label
1999 + docstring(fmt, j + 1, docstring::npos);
2003 return tclass.counters().counterLabel(fmt, lang);
2007 void Paragraph::applyLayout(Layout const & new_layout)
2009 d->layout_ = &new_layout;
2010 LyXAlignment const oldAlign = d->params_.align();
2012 if (!(oldAlign & d->layout_->alignpossible)) {
2013 frontend::Alert::warning(_("Alignment not permitted"),
2014 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2015 d->params_.align(LYX_ALIGN_LAYOUT);
2020 pos_type Paragraph::beginOfBody() const
2022 return d->begin_of_body_;
2026 void Paragraph::setBeginOfBody()
2028 if (d->layout_->labeltype != LABEL_MANUAL) {
2029 d->begin_of_body_ = 0;
2033 // Unroll the first two cycles of the loop
2034 // and remember the previous character to
2035 // remove unnecessary getChar() calls
2037 pos_type end = size();
2038 if (i < end && !isNewline(i)) {
2040 char_type previous_char = 0;
2043 previous_char = d->text_[i];
2044 if (!isNewline(i)) {
2046 while (i < end && previous_char != ' ') {
2051 previous_char = temp;
2057 d->begin_of_body_ = i;
2061 bool Paragraph::allowParagraphCustomization() const
2063 return inInset().allowParagraphCustomization();
2067 bool Paragraph::usePlainLayout() const
2069 return inInset().usePlainLayout();
2073 bool Paragraph::isPassThru() const
2075 return inInset().getLayout().isPassThru() || d->layout_->pass_thru;
2080 // paragraphs inside floats need different alignment tags to avoid
2083 bool noTrivlistCentering(InsetCode code)
2085 return code == FLOAT_CODE
2086 || code == WRAP_CODE
2087 || code == CELL_CODE;
2091 string correction(string const & orig)
2093 if (orig == "flushleft")
2094 return "raggedright";
2095 if (orig == "flushright")
2096 return "raggedleft";
2097 if (orig == "center")
2103 string const corrected_env(string const & suffix, string const & env,
2104 InsetCode code, bool const lastpar)
2106 string output = suffix + "{";
2107 if (noTrivlistCentering(code)) {
2109 // the last paragraph in non-trivlist-aligned
2110 // context is special (to avoid unwanted whitespace)
2111 if (suffix == "\\begin")
2112 return "\\" + correction(env) + "{}";
2115 output += correction(env);
2119 if (suffix == "\\begin")
2125 void adjust_row_column(string const & str, TexRow & texrow, int & column)
2127 if (!contains(str, "\n"))
2128 column += str.size();
2132 column = rsplit(str, tmp, '\n').size();
2139 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2140 odocstream & os, TexRow & texrow,
2141 OutputParams const & runparams) const
2145 if (params_.noindent() && !layout_->pass_thru) {
2146 os << "\\noindent ";
2150 LyXAlignment const curAlign = params_.align();
2152 if (curAlign == layout_->align)
2156 case LYX_ALIGN_NONE:
2157 case LYX_ALIGN_BLOCK:
2158 case LYX_ALIGN_LAYOUT:
2159 case LYX_ALIGN_SPECIAL:
2160 case LYX_ALIGN_DECIMAL:
2162 case LYX_ALIGN_LEFT:
2163 case LYX_ALIGN_RIGHT:
2164 case LYX_ALIGN_CENTER:
2165 if (runparams.moving_arg) {
2172 string const begin_tag = "\\begin";
2173 InsetCode code = ownerCode();
2174 bool const lastpar = runparams.isLastPar;
2177 case LYX_ALIGN_NONE:
2178 case LYX_ALIGN_BLOCK:
2179 case LYX_ALIGN_LAYOUT:
2180 case LYX_ALIGN_SPECIAL:
2181 case LYX_ALIGN_DECIMAL:
2183 case LYX_ALIGN_LEFT: {
2185 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2186 output = corrected_env(begin_tag, "flushleft", code, lastpar);
2188 output = corrected_env(begin_tag, "flushright", code, lastpar);
2189 os << from_ascii(output);
2190 adjust_row_column(output, texrow, column);
2192 } case LYX_ALIGN_RIGHT: {
2194 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2195 output = corrected_env(begin_tag, "flushright", code, lastpar);
2197 output = corrected_env(begin_tag, "flushleft", code, lastpar);
2198 os << from_ascii(output);
2199 adjust_row_column(output, texrow, column);
2201 } case LYX_ALIGN_CENTER: {
2203 output = corrected_env(begin_tag, "center", code, lastpar);
2204 os << from_ascii(output);
2205 adjust_row_column(output, texrow, column);
2214 int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2215 odocstream & os, TexRow & texrow,
2216 OutputParams const & runparams) const
2220 LyXAlignment const curAlign = params_.align();
2222 if (curAlign == layout_->align)
2226 case LYX_ALIGN_NONE:
2227 case LYX_ALIGN_BLOCK:
2228 case LYX_ALIGN_LAYOUT:
2229 case LYX_ALIGN_SPECIAL:
2230 case LYX_ALIGN_DECIMAL:
2232 case LYX_ALIGN_LEFT:
2233 case LYX_ALIGN_RIGHT:
2234 case LYX_ALIGN_CENTER:
2235 if (runparams.moving_arg) {
2242 string const end_tag = "\n\\par\\end";
2243 InsetCode code = ownerCode();
2244 bool const lastpar = runparams.isLastPar;
2247 case LYX_ALIGN_NONE:
2248 case LYX_ALIGN_BLOCK:
2249 case LYX_ALIGN_LAYOUT:
2250 case LYX_ALIGN_SPECIAL:
2251 case LYX_ALIGN_DECIMAL:
2253 case LYX_ALIGN_LEFT: {
2255 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2256 output = corrected_env(end_tag, "flushleft", code, lastpar);
2258 output = corrected_env(end_tag, "flushright", code, lastpar);
2259 os << from_ascii(output);
2260 adjust_row_column(output, texrow, column);
2262 } case LYX_ALIGN_RIGHT: {
2264 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2265 output = corrected_env(end_tag, "flushright", code, lastpar);
2267 output = corrected_env(end_tag, "flushleft", code, lastpar);
2268 os << from_ascii(output);
2269 adjust_row_column(output, texrow, column);
2271 } case LYX_ALIGN_CENTER: {
2273 output = corrected_env(end_tag, "center", code, lastpar);
2274 os << from_ascii(output);
2275 adjust_row_column(output, texrow, column);
2284 // This one spits out the text of the paragraph
2285 void Paragraph::latex(BufferParams const & bparams,
2286 Font const & outerfont,
2287 odocstream & os, TexRow & texrow,
2288 OutputParams const & runparams,
2289 int start_pos, int end_pos, bool force) const
2291 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2293 // FIXME This check should not be needed. Perhaps issue an
2294 // error if it triggers.
2295 Layout const & style = inInset().forcePlainLayout() ?
2296 bparams.documentClass().plainLayout() : *d->layout_;
2298 if (!force && style.inpreamble)
2301 bool const allowcust = allowParagraphCustomization();
2303 // Current base font for all inherited font changes, without any
2304 // change caused by an individual character, except for the language:
2305 // It is set to the language of the first character.
2306 // As long as we are in the label, this font is the base font of the
2307 // label. Before the first body character it is set to the base font
2311 // Maybe we have to create a optional argument.
2312 pos_type body_pos = beginOfBody();
2313 unsigned int column = 0;
2316 // the optional argument is kept in curly brackets in
2317 // case it contains a ']'
2320 basefont = getLabelFont(bparams, outerfont);
2322 basefont = getLayoutFont(bparams, outerfont);
2325 // Which font is currently active?
2326 Font running_font(basefont);
2327 // Do we have an open font change?
2328 bool open_font = false;
2330 Change runningChange = Change(Change::UNCHANGED);
2332 Encoding const * const prev_encoding = runparams.encoding;
2334 texrow.start(id(), 0);
2336 // if the paragraph is empty, the loop will not be entered at all
2338 if (style.isCommand()) {
2343 column += d->startTeXParParams(bparams, os, texrow,
2347 for (pos_type i = 0; i < size(); ++i) {
2348 // First char in paragraph or after label?
2349 if (i == body_pos) {
2352 column += running_font.latexWriteEndChanges(
2353 os, bparams, runparams,
2354 basefont, basefont);
2357 basefont = getLayoutFont(bparams, outerfont);
2358 running_font = basefont;
2360 column += Changes::latexMarkChange(os, bparams,
2361 runningChange, Change(Change::UNCHANGED),
2363 runningChange = Change(Change::UNCHANGED);
2368 if (style.isCommand()) {
2374 column += d->startTeXParParams(bparams, os,
2379 Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
2382 if (bparams.outputChanges && runningChange != change) {
2384 column += running_font.latexWriteEndChanges(
2385 os, bparams, runparams, basefont, basefont);
2388 basefont = getLayoutFont(bparams, outerfont);
2389 running_font = basefont;
2391 column += Changes::latexMarkChange(os, bparams, runningChange,
2393 runningChange = change;
2396 // do not output text which is marked deleted
2397 // if change tracking output is disabled
2398 if (!bparams.outputChanges && change.deleted()) {
2404 // Fully instantiated font
2405 Font const font = getFont(bparams, i, outerfont);
2407 Font const last_font = running_font;
2409 // Do we need to close the previous font?
2411 (font != running_font ||
2412 font.language() != running_font.language()))
2414 column += running_font.latexWriteEndChanges(
2415 os, bparams, runparams, basefont,
2416 (i == body_pos-1) ? basefont : font);
2417 running_font = basefont;
2421 string const running_lang = runparams.use_polyglossia ?
2422 running_font.language()->polyglossia() : running_font.language()->babel();
2423 // close babel's font environment before opening CJK.
2424 string const lang_end_command = runparams.use_polyglossia ?
2425 "\\end{$$lang}" : lyxrc.language_command_end;
2426 if (!running_lang.empty() &&
2427 font.language()->encoding()->package() == Encoding::CJK) {
2428 string end_tag = subst(lang_end_command,
2431 os << from_ascii(end_tag);
2432 column += end_tag.length();
2435 // Switch file encoding if necessary (and allowed)
2436 if (!runparams.pass_thru && !style.pass_thru &&
2437 runparams.encoding->package() != Encoding::none &&
2438 font.language()->encoding()->package() != Encoding::none) {
2439 pair<bool, int> const enc_switch = switchEncoding(os, bparams,
2440 runparams, *(font.language()->encoding()));
2441 if (enc_switch.first) {
2442 column += enc_switch.second;
2443 runparams.encoding = font.language()->encoding();
2447 char_type const c = d->text_[i];
2449 // Do we need to change font?
2450 if ((font != running_font ||
2451 font.language() != running_font.language()) &&
2454 odocstringstream ods;
2455 column += font.latexWriteStartChanges(ods, bparams,
2456 runparams, basefont,
2458 running_font = font;
2460 docstring fontchange = ods.str();
2461 // check whether the fontchange ends with a \\textcolor
2462 // modifier and the text starts with a space (bug 4473)
2463 docstring const last_modifier = rsplit(fontchange, '\\');
2464 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2465 os << fontchange << from_ascii("{}");
2466 // check if the fontchange ends with a trailing blank
2467 // (like "\small " (see bug 3382)
2468 else if (suffixIs(fontchange, ' ') && c == ' ')
2469 os << fontchange.substr(0, fontchange.size() - 1)
2470 << from_ascii("{}");
2475 // FIXME: think about end_pos implementation...
2476 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2477 // FIXME: integrate this case in latexSpecialChar
2478 // Do not print the separation of the optional argument
2479 // if style.pass_thru is false. This works because
2480 // latexSpecialChar ignores spaces if
2481 // style.pass_thru is false.
2482 if (i != body_pos - 1) {
2483 if (d->simpleTeXBlanks(
2484 runparams, os, texrow,
2485 i, column, font, style)) {
2486 // A surrogate pair was output. We
2487 // must not call latexSpecialChar
2488 // in this iteration, since it would output
2489 // the combining character again.
2496 OutputParams rp = runparams;
2497 rp.free_spacing = style.free_spacing;
2498 rp.local_font = &font;
2499 rp.intitle = style.intitle;
2501 // Two major modes: LaTeX or plain
2502 // Handle here those cases common to both modes
2503 // and then split to handle the two modes separately.
2504 if (c == META_INSET) {
2505 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2506 d->latexInset(bparams, os,
2507 texrow, rp, running_font,
2508 basefont, outerfont, open_font,
2509 runningChange, style, i, column);
2512 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2514 d->latexSpecialChar(os, rp, running_font, runningChange,
2516 } catch (EncodingException & e) {
2517 if (runparams.dryrun) {
2518 os << "<" << _("LyX Warning: ")
2519 << _("uncodable character") << " '";
2523 // add location information and throw again.
2532 // Set the encoding to that returned from latexSpecialChar (see
2533 // comment for encoding member in OutputParams.h)
2534 runparams.encoding = rp.encoding;
2537 // If we have an open font definition, we have to close it
2539 #ifdef FIXED_LANGUAGE_END_DETECTION
2541 running_font.latexWriteEndChanges(os, bparams,
2542 runparams, basefont,
2543 next_->getFont(bparams, 0, outerfont));
2545 running_font.latexWriteEndChanges(os, bparams,
2546 runparams, basefont, basefont);
2549 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2550 //FIXME: there as we start another \selectlanguage with the next paragraph if
2551 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2552 running_font.latexWriteEndChanges(os, bparams, runparams,
2553 basefont, basefont);
2557 column += Changes::latexMarkChange(os, bparams, runningChange,
2558 Change(Change::UNCHANGED), runparams);
2560 // Needed if there is an optional argument but no contents.
2561 if (body_pos > 0 && body_pos == size()) {
2565 if (allowcust && d->endTeXParParams(bparams, os, texrow, runparams)
2566 && runparams.encoding != prev_encoding) {
2567 runparams.encoding = prev_encoding;
2568 if (!runparams.isFullUnicode())
2569 os << setEncoding(prev_encoding->iconvName());
2572 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2576 bool Paragraph::emptyTag() const
2578 for (pos_type i = 0; i < size(); ++i) {
2579 if (Inset const * inset = getInset(i)) {
2580 InsetCode lyx_code = inset->lyxCode();
2581 // FIXME testing like that is wrong. What is
2583 if (lyx_code != TOC_CODE &&
2584 lyx_code != INCLUDE_CODE &&
2585 lyx_code != GRAPHICS_CODE &&
2586 lyx_code != ERT_CODE &&
2587 lyx_code != LISTINGS_CODE &&
2588 lyx_code != FLOAT_CODE &&
2589 lyx_code != TABULAR_CODE) {
2593 char_type c = d->text_[i];
2594 if (c != ' ' && c != '\t')
2602 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2605 for (pos_type i = 0; i < size(); ++i) {
2606 if (Inset const * inset = getInset(i)) {
2607 InsetCode lyx_code = inset->lyxCode();
2608 if (lyx_code == LABEL_CODE) {
2609 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2610 docstring const & id = il->getParam("name");
2611 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2619 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2623 for (i = 0; i < size(); ++i) {
2624 if (Inset const * inset = getInset(i)) {
2625 inset->docbook(os, runparams);
2627 char_type c = d->text_[i];
2630 os << sgml::escapeChar(c);
2637 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2641 for (i = 0; i < size(); ++i) {
2642 if (Inset const * inset = getInset(i)) {
2643 inset->xhtml(xs, runparams);
2645 char_type c = d->text_[i];
2655 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2658 pos_type size = text_.size();
2659 for (pos_type i = initial; i < size; ++i) {
2660 Font font = owner_->getFont(buf.params(), i, outerfont);
2661 if (text_[i] == META_INSET)
2663 if (i != initial && font != font_old)
2672 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2674 OutputParams const & runparams,
2675 Font const & outerfont,
2676 pos_type initial) const
2678 bool emph_flag = false;
2680 Layout const & style = *d->layout_;
2682 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2684 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2687 // parsing main loop
2688 for (pos_type i = initial; i < size(); ++i) {
2689 Font font = getFont(buf.params(), i, outerfont);
2691 // handle <emphasis> tag
2692 if (font_old.emph() != font.fontInfo().emph()) {
2693 if (font.fontInfo().emph() == FONT_ON) {
2696 } else if (i != initial) {
2697 os << "</emphasis>";
2702 if (Inset const * inset = getInset(i)) {
2703 inset->docbook(os, runparams);
2705 char_type c = d->text_[i];
2707 if (style.pass_thru)
2710 os << sgml::escapeChar(c);
2712 font_old = font.fontInfo();
2716 os << "</emphasis>";
2719 if (style.free_spacing)
2721 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2726 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2728 OutputParams const & runparams,
2729 Font const & outerfont,
2730 pos_type initial) const
2734 bool emph_flag = false;
2735 bool bold_flag = false;
2738 Layout const & style = *d->layout_;
2740 if (!runparams.for_toc && runparams.html_make_pars) {
2741 // generate a magic label for this paragraph
2742 string const attr = "id='" + magicLabel() + "'";
2743 xs << html::CompTag("a", attr);
2747 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2749 // parsing main loop
2750 for (pos_type i = initial; i < size(); ++i) {
2751 // let's not show deleted material in the output
2755 Font font = getFont(buf.params(), i, outerfont);
2758 if (font_old.emph() != font.fontInfo().emph()) {
2759 if (font.fontInfo().emph() == FONT_ON) {
2760 xs << html::StartTag("em");
2762 } else if (emph_flag && i != initial) {
2763 xs << html::EndTag("em");
2768 if (font_old.series() != font.fontInfo().series()) {
2769 if (font.fontInfo().series() == BOLD_SERIES) {
2770 xs << html::StartTag("strong");
2772 } else if (bold_flag && i != initial) {
2773 xs << html::EndTag("strong");
2778 // Other such tags? What about the other text ranges?
2780 Inset const * inset = getInset(i);
2782 if (!runparams.for_toc || inset->isInToc()) {
2783 OutputParams np = runparams;
2784 if (!inset->getLayout().htmlisblock())
2785 np.html_in_par = true;
2786 retval += inset->xhtml(xs, np);
2789 char_type c = d->text_[i];
2791 if (style.pass_thru)
2793 else if (c == '-') {
2796 if (j < size() && d->text_[j] == '-') {
2798 if (j < size() && d->text_[j] == '-') {
2799 str += from_ascii("—");
2802 str += from_ascii("–");
2808 // We don't want to escape the entities. Note that
2809 // it is safe to do this, since str can otherwise
2810 // only be "-". E.g., it can't be "<".
2811 xs << XHTMLStream::ESCAPE_NONE << str;
2815 font_old = font.fontInfo();
2823 bool Paragraph::isHfill(pos_type pos) const
2825 Inset const * inset = getInset(pos);
2826 return inset && (inset->lyxCode() == SPACE_CODE &&
2827 inset->isStretchableSpace());
2831 bool Paragraph::isNewline(pos_type pos) const
2833 Inset const * inset = getInset(pos);
2834 return inset && inset->lyxCode() == NEWLINE_CODE;
2838 bool Paragraph::isLineSeparator(pos_type pos) const
2840 char_type const c = d->text_[pos];
2841 if (isLineSeparatorChar(c))
2843 Inset const * inset = getInset(pos);
2844 return inset && inset->isLineSeparator();
2848 bool Paragraph::isWordSeparator(pos_type pos) const
2850 if (Inset const * inset = getInset(pos))
2851 return !inset->isLetter();
2852 char_type const c = d->text_[pos];
2853 // We want to pass the ' and escape chars to the spellchecker
2854 static docstring const quote = from_utf8(lyxrc.spellchecker_esc_chars + '\'');
2855 return (!isLetterChar(c) && !isDigitASCII(c) && !contains(quote, c))
2860 bool Paragraph::isSameSpellRange(pos_type pos1, pos_type pos2) const
2863 || d->speller_state_.getRange(pos1) == d->speller_state_.getRange(pos2);
2867 bool Paragraph::isChar(pos_type pos) const
2869 if (Inset const * inset = getInset(pos))
2870 return inset->isChar();
2871 char_type const c = d->text_[pos];
2872 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
2876 bool Paragraph::isSpace(pos_type pos) const
2878 if (Inset const * inset = getInset(pos))
2879 return inset->isSpace();
2880 char_type const c = d->text_[pos];
2881 return lyx::isSpace(c);
2886 Paragraph::getParLanguage(BufferParams const & bparams) const
2889 return getFirstFontSettings(bparams).language();
2890 // FIXME: we should check the prev par as well (Lgb)
2891 return bparams.language;
2895 bool Paragraph::isRTL(BufferParams const & bparams) const
2897 return lyxrc.rtl_support
2898 && getParLanguage(bparams)->rightToLeft()
2899 && !inInset().getLayout().forceLTR();
2903 void Paragraph::changeLanguage(BufferParams const & bparams,
2904 Language const * from, Language const * to)
2906 // change language including dummy font change at the end
2907 for (pos_type i = 0; i <= size(); ++i) {
2908 Font font = getFontSettings(bparams, i);
2909 if (font.language() == from) {
2910 font.setLanguage(to);
2914 d->requestSpellCheck(size());
2918 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
2920 Language const * doc_language = bparams.language;
2921 FontList::const_iterator cit = d->fontlist_.begin();
2922 FontList::const_iterator end = d->fontlist_.end();
2924 for (; cit != end; ++cit)
2925 if (cit->font().language() != ignore_language &&
2926 cit->font().language() != latex_language &&
2927 cit->font().language() != doc_language)
2933 void Paragraph::getLanguages(std::set<Language const *> & languages) const
2935 FontList::const_iterator cit = d->fontlist_.begin();
2936 FontList::const_iterator end = d->fontlist_.end();
2938 for (; cit != end; ++cit) {
2939 Language const * lang = cit->font().language();
2940 if (lang != ignore_language &&
2941 lang != latex_language)
2942 languages.insert(lang);
2947 docstring Paragraph::asString(int options) const
2949 return asString(0, size(), options);
2953 docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
2955 odocstringstream os;
2958 && options & AS_STR_LABEL
2959 && !d->params_.labelString().empty())
2960 os << d->params_.labelString() << ' ';
2962 for (pos_type i = beg; i < end; ++i) {
2963 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
2965 char_type const c = d->text_[i];
2966 if (isPrintable(c) || c == '\t'
2967 || (c == '\n' && (options & AS_STR_NEWLINES)))
2969 else if (c == META_INSET && (options & AS_STR_INSETS)) {
2970 getInset(i)->toString(os);
2971 if (getInset(i)->asInsetMath())
2980 void Paragraph::forToc(docstring & os, size_t maxlen) const
2982 if (!d->params_.labelString().empty())
2983 os += d->params_.labelString() + ' ';
2984 for (pos_type i = 0; i < size() && os.length() < maxlen; ++i) {
2987 char_type const c = d->text_[i];
2990 else if (c == '\t' || c == '\n')
2992 else if (c == META_INSET)
2993 getInset(i)->forToc(os, maxlen);
2998 docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputParams & runparams) const
3000 odocstringstream os;
3003 && options & AS_STR_LABEL
3004 && !d->params_.labelString().empty())
3005 os << d->params_.labelString() << ' ';
3007 for (pos_type i = beg; i < end; ++i) {
3008 char_type const c = d->text_[i];
3009 if (isPrintable(c) || c == '\t'
3010 || (c == '\n' && (options & AS_STR_NEWLINES)))
3012 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3013 getInset(i)->plaintext(os, runparams);
3021 void Paragraph::setInsetOwner(Inset const * inset)
3023 d->inset_owner_ = inset;
3027 int Paragraph::id() const
3033 void Paragraph::setId(int id)
3039 Layout const & Paragraph::layout() const
3045 void Paragraph::setLayout(Layout const & layout)
3047 d->layout_ = &layout;
3051 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3053 setLayout(tc.defaultLayout());
3057 void Paragraph::setPlainLayout(DocumentClass const & tc)
3059 setLayout(tc.plainLayout());
3063 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3065 if (usePlainLayout())
3066 setPlainLayout(tclass);
3068 setDefaultLayout(tclass);
3072 Inset const & Paragraph::inInset() const
3074 LASSERT(d->inset_owner_, throw ExceptionMessage(BufferException,
3075 _("Memory problem"), _("Paragraph not properly initialized")));
3076 return *d->inset_owner_;
3080 ParagraphParameters & Paragraph::params()
3086 ParagraphParameters const & Paragraph::params() const
3092 bool Paragraph::isFreeSpacing() const
3094 if (d->layout_->free_spacing)
3096 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3100 bool Paragraph::allowEmpty() const
3102 if (d->layout_->keepempty)
3104 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3108 char_type Paragraph::transformChar(char_type c, pos_type pos) const
3110 if (!Encodings::isArabicChar(c))
3113 char_type prev_char = ' ';
3114 char_type next_char = ' ';
3116 for (pos_type i = pos - 1; i >= 0; --i) {
3117 char_type const par_char = d->text_[i];
3118 if (!Encodings::isArabicComposeChar(par_char)) {
3119 prev_char = par_char;
3124 for (pos_type i = pos + 1, end = size(); i < end; ++i) {
3125 char_type const par_char = d->text_[i];
3126 if (!Encodings::isArabicComposeChar(par_char)) {
3127 next_char = par_char;
3132 if (Encodings::isArabicChar(next_char)) {
3133 if (Encodings::isArabicChar(prev_char) &&
3134 !Encodings::isArabicSpecialChar(prev_char))
3135 return Encodings::transformChar(c, Encodings::FORM_MEDIAL);
3137 return Encodings::transformChar(c, Encodings::FORM_INITIAL);
3139 if (Encodings::isArabicChar(prev_char) &&
3140 !Encodings::isArabicSpecialChar(prev_char))
3141 return Encodings::transformChar(c, Encodings::FORM_FINAL);
3143 return Encodings::transformChar(c, Encodings::FORM_ISOLATED);
3148 int Paragraph::checkBiblio(Buffer const & buffer)
3151 // This is getting more and more a mess. ...We really should clean
3152 // up this bibitem issue for 1.6.
3154 // Add bibitem insets if necessary
3155 if (d->layout_->labeltype != LABEL_BIBLIO)
3158 bool hasbibitem = !d->insetlist_.empty()
3159 // Insist on it being in pos 0
3160 && d->text_[0] == META_INSET
3161 && d->insetlist_.begin()->inset->lyxCode() == BIBITEM_CODE;
3163 bool track_changes = buffer.params().trackChanges;
3168 // remove a bibitem in pos != 0
3169 // restore it later in pos 0 if necessary
3170 // (e.g. if a user inserts contents _before_ the item)
3171 // we're assuming there's only one of these, which there
3173 int erasedInsetPosition = -1;
3174 InsetList::iterator it = d->insetlist_.begin();
3175 InsetList::iterator end = d->insetlist_.end();
3176 for (; it != end; ++it)
3177 if (it->inset->lyxCode() == BIBITEM_CODE
3179 InsetCommand * olditem = it->inset->asInsetCommand();
3180 oldkey = olditem->getParam("key");
3181 oldlabel = olditem->getParam("label");
3182 erasedInsetPosition = it->pos;
3183 eraseChar(erasedInsetPosition, track_changes);
3187 // There was an InsetBibitem at the beginning, and we didn't
3188 // have to erase one.
3189 if (hasbibitem && erasedInsetPosition < 0)
3192 // There was an InsetBibitem at the beginning and we did have to
3193 // erase one. So we give its properties to the beginning inset.
3195 InsetCommand * inset = d->insetlist_.begin()->inset->asInsetCommand();
3196 if (!oldkey.empty())
3197 inset->setParam("key", oldkey);
3198 inset->setParam("label", oldlabel);
3199 return -erasedInsetPosition;
3202 // There was no inset at the beginning, so we need to create one with
3203 // the key and label of the one we erased.
3204 InsetBibitem * inset =
3205 new InsetBibitem(const_cast<Buffer *>(&buffer), InsetCommandParams(BIBITEM_CODE));
3206 // restore values of previously deleted item in this par.
3207 if (!oldkey.empty())
3208 inset->setParam("key", oldkey);
3209 inset->setParam("label", oldlabel);
3210 insertInset(0, inset,
3211 Change(track_changes ? Change::INSERTED : Change::UNCHANGED));
3217 void Paragraph::checkAuthors(AuthorList const & authorList)
3219 d->changes_.checkAuthors(authorList);
3223 bool Paragraph::isChanged(pos_type pos) const
3225 return lookupChange(pos).changed();
3229 bool Paragraph::isInserted(pos_type pos) const
3231 return lookupChange(pos).inserted();
3235 bool Paragraph::isDeleted(pos_type pos) const
3237 return lookupChange(pos).deleted();
3241 InsetList const & Paragraph::insetList() const
3243 return d->insetlist_;
3247 void Paragraph::setBuffer(Buffer & b)
3249 d->insetlist_.setBuffer(b);
3253 Inset * Paragraph::releaseInset(pos_type pos)
3255 Inset * inset = d->insetlist_.release(pos);
3256 /// does not honour change tracking!
3257 eraseChar(pos, false);
3262 Inset * Paragraph::getInset(pos_type pos)
3264 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3265 ? d->insetlist_.get(pos) : 0;
3269 Inset const * Paragraph::getInset(pos_type pos) const
3271 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3272 ? d->insetlist_.get(pos) : 0;
3276 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3277 pos_type & right, TextCase action)
3279 // process sequences of modified characters; in change
3280 // tracking mode, this approach results in much better
3281 // usability than changing case on a char-by-char basis
3284 bool const trackChanges = bparams.trackChanges;
3286 bool capitalize = true;
3288 for (; pos < right; ++pos) {
3289 char_type oldChar = d->text_[pos];
3290 char_type newChar = oldChar;
3292 // ignore insets and don't play with deleted text!
3293 if (oldChar != META_INSET && !isDeleted(pos)) {
3295 case text_lowercase:
3296 newChar = lowercase(oldChar);
3298 case text_capitalization:
3300 newChar = uppercase(oldChar);
3304 case text_uppercase:
3305 newChar = uppercase(oldChar);
3310 if (isWordSeparator(pos) || isDeleted(pos)) {
3311 // permit capitalization again
3315 if (oldChar != newChar) {
3317 if (pos != right - 1)
3319 // step behind the changing area
3323 int erasePos = pos - changes.size();
3324 for (size_t i = 0; i < changes.size(); i++) {
3325 insertChar(pos, changes[i],
3326 getFontSettings(bparams,
3329 if (!eraseChar(erasePos, trackChanges)) {
3332 ++right; // expand selection
3340 int Paragraph::find(docstring const & str, bool cs, bool mw,
3341 pos_type start_pos, bool del) const
3343 pos_type pos = start_pos;
3344 int const strsize = str.length();
3346 pos_type const parsize = d->text_.size();
3347 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3348 // Ignore ligature break and hyphenation chars while searching
3349 while (pos < parsize - 1 && isInset(pos)) {
3350 const InsetSpecialChar *isc = dynamic_cast<const InsetSpecialChar*>(getInset(pos));
3352 || (isc->kind() != InsetSpecialChar::HYPHENATION
3353 && isc->kind() != InsetSpecialChar::LIGATURE_BREAK))
3357 if (cs && str[i] != d->text_[pos])
3359 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3361 if (!del && isDeleted(pos))
3368 // if necessary, check whether string matches word
3370 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3373 && !isWordSeparator(pos))
3377 return pos - start_pos;
3381 char_type Paragraph::getChar(pos_type pos) const
3383 return d->text_[pos];
3387 pos_type Paragraph::size() const
3389 return d->text_.size();
3393 bool Paragraph::empty() const
3395 return d->text_.empty();
3399 bool Paragraph::isInset(pos_type pos) const
3401 return d->text_[pos] == META_INSET;
3405 bool Paragraph::isSeparator(pos_type pos) const
3407 //FIXME: Are we sure this can be the only separator?
3408 return d->text_[pos] == ' ';
3412 void Paragraph::deregisterWords()
3414 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3415 Private::LangWordsMap::const_iterator ite = d->words_.end();
3416 for (; itl != ite; ++itl) {
3417 WordList * wl = theWordList(itl->first);
3418 Private::Words::const_iterator it = (itl->second).begin();
3419 Private::Words::const_iterator et = (itl->second).end();
3420 for (; it != et; ++it)
3427 void Paragraph::locateWord(pos_type & from, pos_type & to,
3428 word_location const loc) const
3431 case WHOLE_WORD_STRICT:
3432 if (from == 0 || from == size()
3433 || isWordSeparator(from)
3434 || isWordSeparator(from - 1)) {
3438 // no break here, we go to the next
3441 // If we are already at the beginning of a word, do nothing
3442 if (!from || isWordSeparator(from - 1))
3444 // no break here, we go to the next
3447 // always move the cursor to the beginning of previous word
3448 while (from && !isWordSeparator(from - 1))
3452 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3455 // no need to move the 'from' cursor
3459 while (to < size() && !isWordSeparator(to))
3464 void Paragraph::collectWords()
3466 // This is the value that needs to be exposed in the preferences
3467 // to resolve bug #6760.
3468 static int minlength = 6;
3469 pos_type n = size();
3470 for (pos_type pos = 0; pos < n; ++pos) {
3471 if (isWordSeparator(pos))
3473 pos_type from = pos;
3474 locateWord(from, pos, WHOLE_WORD);
3475 if (pos - from >= minlength) {
3476 docstring word = asString(from, pos, AS_STR_NONE);
3477 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
3478 if (cit == d->fontlist_.end())
3480 Language const * lang = cit->font().language();
3481 d->words_[*lang].insert(word);
3487 void Paragraph::registerWords()
3489 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3490 Private::LangWordsMap::const_iterator ite = d->words_.end();
3491 for (; itl != ite; ++itl) {
3492 WordList * wl = theWordList(itl->first);
3493 Private::Words::const_iterator it = (itl->second).begin();
3494 Private::Words::const_iterator et = (itl->second).end();
3495 for (; it != et; ++it)
3501 void Paragraph::updateWords()
3509 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3511 SkipPositionsIterator begin = skips.begin();
3512 SkipPositions::iterator end = skips.end();
3513 if (pos > 0 && begin < end) {
3515 if (end->last == pos - 1) {
3520 skips.insert(end, FontSpan(pos, pos));
3524 Language * Paragraph::Private::locateSpellRange(
3525 pos_type & from, pos_type & to,
3526 SkipPositions & skips) const
3528 // skip leading white space
3529 while (from < to && owner_->isWordSeparator(from))
3531 // don't check empty range
3534 // get current language
3535 Language * lang = getSpellLanguage(from);
3536 pos_type last = from;
3537 bool samelang = true;
3538 bool sameinset = true;
3539 while (last < to && samelang && sameinset) {
3540 // hop to end of word
3541 while (last < to && !owner_->isWordSeparator(last)) {
3542 if (owner_->getInset(last)) {
3543 appendSkipPosition(skips, last);
3544 } else if (owner_->isDeleted(last)) {
3545 appendSkipPosition(skips, last);
3549 // hop to next word while checking for insets
3550 while (sameinset && last < to && owner_->isWordSeparator(last)) {
3551 if (Inset const * inset = owner_->getInset(last))
3552 sameinset = inset->isChar() && inset->isLetter();
3553 if (sameinset && owner_->isDeleted(last)) {
3554 appendSkipPosition(skips, last);
3559 if (sameinset && last < to) {
3560 // now check for language change
3561 samelang = lang == getSpellLanguage(last);
3564 // if language change detected backstep is needed
3572 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
3575 const_cast<Language *>(owner_->getFontSettings(
3576 inset_owner_->buffer().params(), from).language());
3577 if (lang == inset_owner_->buffer().params().language
3578 && !lyxrc.spellchecker_alt_lang.empty()) {
3580 string const lang_variety =
3581 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3582 lang->setCode(lang_code);
3583 lang->setVariety(lang_variety);
3589 void Paragraph::requestSpellCheck(pos_type pos)
3591 d->requestSpellCheck(pos == -1 ? size() : pos);
3595 bool Paragraph::needsSpellCheck() const
3597 SpellChecker::ChangeNumber speller_change_number = 0;
3598 if (theSpellChecker())
3599 speller_change_number = theSpellChecker()->changeNumber();
3600 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
3601 d->speller_state_.needsCompleteRefresh(speller_change_number);
3603 return d->needsSpellCheck();
3607 bool Paragraph::Private::ignoreWord(docstring const & word) const
3609 // Ignore words with digits
3610 // FIXME: make this customizable
3611 // (note that some checkers ignore words with digits by default)
3612 docstring::const_iterator cit = word.begin();
3613 docstring::const_iterator const end = word.end();
3614 for (; cit != end; ++cit) {
3615 if (isNumber((*cit)))
3622 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
3623 WordLangTuple & wl, docstring_list & suggestions,
3624 bool do_suggestion, bool check_learned) const
3626 SpellChecker::Result result = SpellChecker::WORD_OK;
3627 SpellChecker * speller = theSpellChecker();
3631 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
3634 locateWord(from, to, WHOLE_WORD);
3635 if (from == to || from >= size())
3638 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3639 Language * lang = d->getSpellLanguage(from);
3641 wl = WordLangTuple(word, lang);
3646 if (needsSpellCheck() || check_learned) {
3647 if (!d->ignoreWord(word)) {
3648 bool const trailing_dot = to < size() && d->text_[to] == '.';
3649 result = speller->check(wl);
3650 if (SpellChecker::misspelled(result) && trailing_dot) {
3651 wl = WordLangTuple(word.append(from_ascii(".")), lang);
3652 result = speller->check(wl);
3653 if (!SpellChecker::misspelled(result)) {
3654 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
3656 from << ".." << to << "]");
3658 // spell check with dot appended failed
3659 // restore original word/lang value
3660 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3661 wl = WordLangTuple(word, lang);
3665 d->setMisspelled(from, to, result);
3667 result = d->speller_state_.getState(from);
3670 bool const misspelled_ = SpellChecker::misspelled(result) ;
3671 if (misspelled_ && do_suggestion)
3672 speller->suggest(wl, suggestions);
3673 else if (misspelled_)
3674 LYXERR(Debug::GUI, "misspelled word: \"" <<
3676 from << ".." << to << "]");
3678 suggestions.clear();
3684 void Paragraph::Private::markMisspelledWords(
3685 pos_type const & first, pos_type const & last,
3686 SpellChecker::Result result,
3687 docstring const & word,
3688 SkipPositions const & skips)
3690 if (!SpellChecker::misspelled(result)) {
3691 setMisspelled(first, last, SpellChecker::WORD_OK);
3695 SpellChecker * speller = theSpellChecker();
3696 // locate and enumerate the error positions
3697 int nerrors = speller->numMisspelledWords();
3699 SkipPositionsIterator it = skips.begin();
3700 SkipPositionsIterator et = skips.end();
3701 for (int index = 0; index < nerrors; ++index) {
3704 speller->misspelledWord(index, wstart, wlen);
3705 /// should not happen if speller supports range checks
3706 if (!wlen) continue;
3707 docstring const misspelled = word.substr(wstart, wlen);
3708 wstart += first + numskipped;
3709 if (snext < wstart) {
3710 /// mark the range of correct spelling
3711 numskipped += countSkips(it, et, wstart);
3712 setMisspelled(snext,
3713 wstart - 1, SpellChecker::WORD_OK);
3715 snext = wstart + wlen;
3716 numskipped += countSkips(it, et, snext);
3717 /// mark the range of misspelling
3718 setMisspelled(wstart, snext, result);
3719 LYXERR(Debug::GUI, "misspelled word: \"" <<
3720 misspelled << "\" [" <<
3721 wstart << ".." << (snext-1) << "]");
3724 if (snext <= last) {
3725 /// mark the range of correct spelling at end
3726 setMisspelled(snext, last, SpellChecker::WORD_OK);
3731 void Paragraph::spellCheck() const
3733 SpellChecker * speller = theSpellChecker();
3734 if (!speller || !size() ||!needsSpellCheck())
3738 d->rangeOfSpellCheck(start, endpos);
3739 if (speller->canCheckParagraph()) {
3740 // loop until we leave the range
3741 for (pos_type first = start; first < endpos; ) {
3742 pos_type last = endpos;
3743 Private::SkipPositions skips;
3744 Language * lang = d->locateSpellRange(first, last, skips);
3745 if (first >= endpos)
3747 // start the spell checker on the unit of meaning
3748 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
3749 WordLangTuple wl = WordLangTuple(word, lang);
3750 SpellChecker::Result result = word.size() ?
3751 speller->check(wl) : SpellChecker::WORD_OK;
3752 d->markMisspelledWords(first, last, result, word, skips);
3756 static docstring_list suggestions;
3757 pos_type to = endpos;
3758 while (start < endpos) {
3760 spellCheck(start, to, wl, suggestions, false);
3764 d->readySpellCheck();
3768 bool Paragraph::isMisspelled(pos_type pos) const
3770 return SpellChecker::misspelled(d->speller_state_.getState(pos));
3774 string Paragraph::magicLabel() const
3777 ss << "magicparlabel-" << id();