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 Jean-Marc Lasgouttes
9 * \author Angus Leeming
11 * \author André Pönitz
13 * \author Jürgen Vigna
15 * Full author contact details are available in file CREDITS.
20 #include "Paragraph.h"
22 #include "LayoutFile.h"
24 #include "BufferParams.h"
28 #include "InsetList.h"
30 #include "LaTeXFeatures.h"
36 #include "OutputParams.h"
37 #include "output_latex.h"
38 #include "paragraph_funcs.h"
39 #include "ParagraphParameters.h"
41 #include "TextClass.h"
47 #include "frontends/alert.h"
49 #include "insets/InsetBibitem.h"
50 #include "insets/InsetLabel.h"
52 #include "support/lassert.h"
53 #include "support/convert.h"
54 #include "support/debug.h"
55 #include "support/ExceptionMessage.h"
56 #include "support/gettext.h"
57 #include "support/lstrings.h"
58 #include "support/Messages.h"
59 #include "support/textutils.h"
65 using namespace lyx::support;
70 /// Inset identifier (above 0x10ffff, for ucs-4)
71 char_type const META_INSET = 0x200001;
74 /////////////////////////////////////////////////////////////////////
78 /////////////////////////////////////////////////////////////////////
80 class Paragraph::Private
84 Private(Paragraph * owner, Layout const & layout);
85 /// "Copy constructor"
86 Private(Private const &, Paragraph * owner);
87 /// Copy constructor from \p beg to \p end
88 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
91 void insertChar(pos_type pos, char_type c, Change const & change);
93 /// Output the surrogate pair formed by \p c and \p next to \p os.
94 /// \return the number of characters written.
95 int latexSurrogatePair(odocstream & os, char_type c, char_type next,
98 /// Output a space in appropriate formatting (or a surrogate pair
99 /// if the next character is a combining character).
100 /// \return whether a surrogate pair was output.
101 bool simpleTeXBlanks(OutputParams const &,
102 odocstream &, TexRow & texrow,
104 unsigned int & column,
106 Layout const & style);
108 /// Output consecutive unicode chars, belonging to the same script as
109 /// specified by the latex macro \p ltx, to \p os starting from \p i.
110 /// \return the number of characters written.
111 int writeScriptChars(odocstream & os, docstring const & ltx,
112 Change const &, Encoding const &, pos_type & i);
114 /// This could go to ParagraphParameters if we want to.
115 int startTeXParParams(BufferParams const &, odocstream &, TexRow &,
116 OutputParams const &) const;
118 /// This could go to ParagraphParameters if we want to.
119 int endTeXParParams(BufferParams const &, odocstream &, TexRow &,
120 OutputParams const &) const;
123 void latexInset(BufferParams const &,
125 TexRow & texrow, OutputParams &,
128 Font const & outerfont,
130 Change & running_change,
131 Layout const & style,
133 unsigned int & column);
136 void latexSpecialChar(
138 OutputParams const & runparams,
139 Font const & running_font,
140 Change const & running_change,
141 Layout const & style,
143 unsigned int & column);
150 unsigned int & column);
152 bool latexSpecialTypewriter(
156 unsigned int & column);
158 bool latexSpecialPhrase(
161 unsigned int & column,
162 OutputParams const & runparams);
165 void validate(LaTeXFeatures & features,
166 Layout const & layout) const;
168 /// Checks if the paragraph contains only text and no inset or font change.
169 bool onlyText(Buffer const & buf, Font const & outerfont,
170 pos_type initial) const;
172 /// match a string against a particular point in the paragraph
173 bool isTextAt(string const & str, pos_type pos) const;
175 /// Which Paragraph owns us?
179 Inset const * inset_owner_;
187 static unsigned int paragraph_id;
189 ParagraphParameters params_;
191 /// for recording and looking up changes
195 InsetList insetlist_;
198 pos_type begin_of_body_;
200 typedef docstring TextContainer;
204 typedef std::set<docstring> Words;
208 Layout const * layout_;
212 // Initialization of the counter for the paragraph id's,
213 unsigned int Paragraph::Private::paragraph_id = 0;
217 struct special_phrase {
223 special_phrase const special_phrases[] = {
224 { "LyX", from_ascii("\\LyX{}"), false },
225 { "TeX", from_ascii("\\TeX{}"), true },
226 { "LaTeX2e", from_ascii("\\LaTeXe{}"), true },
227 { "LaTeX", from_ascii("\\LaTeX{}"), true },
230 size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase);
235 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
236 : owner_(owner), inset_owner_(0), begin_of_body_(0), layout_(&layout)
238 id_ = paragraph_id++;
243 Paragraph::Private::Private(Private const & p, Paragraph * owner)
244 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
245 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
246 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
249 id_ = paragraph_id++;
253 Paragraph::Private::Private(Private const & p, Paragraph * owner,
254 pos_type beg, pos_type end)
255 : owner_(owner), inset_owner_(p.inset_owner_),
256 params_(p.params_), changes_(p.changes_),
257 insetlist_(p.insetlist_, beg, end),
258 begin_of_body_(p.begin_of_body_), words_(p.words_),
261 id_ = paragraph_id++;
262 if (beg >= pos_type(p.text_.size()))
264 text_ = p.text_.substr(beg, end - beg);
266 FontList::const_iterator fcit = fontlist_.begin();
267 FontList::const_iterator fend = fontlist_.end();
268 for (; fcit != fend; ++fcit) {
269 if (fcit->pos() < beg)
271 if (fcit->pos() >= end) {
272 // Add last entry in the fontlist_.
273 fontlist_.set(text_.size() - 1, fcit->font());
276 // Add a new entry in the fontlist_.
277 fontlist_.set(fcit->pos() - beg, fcit->font());
282 void Paragraph::addChangesToToc(DocIterator const & cdit,
283 Buffer const & buf) const
285 d->changes_.addToToc(cdit, buf);
289 bool Paragraph::isChanged(pos_type start, pos_type end) const
291 LASSERT(start >= 0 && start <= size(), /**/);
292 LASSERT(end > start && end <= size() + 1, /**/);
294 return d->changes_.isChanged(start, end);
298 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
300 // keep the logic here in sync with the logic of eraseChars()
304 Change const change = d->changes_.lookup(size());
305 return change.type == Change::INSERTED && change.author == 0;
309 void Paragraph::setChange(Change const & change)
311 // beware of the imaginary end-of-par character!
312 d->changes_.set(change, 0, size() + 1);
315 * Propagate the change recursively - but not in case of DELETED!
317 * Imagine that your co-author makes changes in an existing inset. He
318 * sends your document to you and you come to the conclusion that the
319 * inset should go completely. If you erase it, LyX must not delete all
320 * text within the inset. Otherwise, the change tracked insertions of
321 * your co-author get lost and there is no way to restore them later.
323 * Conclusion: An inset's content should remain untouched if you delete it
326 if (change.type != Change::DELETED) {
327 for (pos_type pos = 0; pos < size(); ++pos) {
328 if (Inset * inset = getInset(pos))
329 inset->setChange(change);
335 void Paragraph::setChange(pos_type pos, Change const & change)
337 LASSERT(pos >= 0 && pos <= size(), /**/);
338 d->changes_.set(change, pos);
340 // see comment in setChange(Change const &) above
341 if (change.type != Change::DELETED && pos < size())
342 if (Inset * inset = getInset(pos))
343 inset->setChange(change);
347 Change const & Paragraph::lookupChange(pos_type pos) const
349 LASSERT(pos >= 0 && pos <= size(), /**/);
350 return d->changes_.lookup(pos);
354 void Paragraph::acceptChanges(BufferParams const & bparams, pos_type start,
357 LASSERT(start >= 0 && start <= size(), /**/);
358 LASSERT(end > start && end <= size() + 1, /**/);
360 for (pos_type pos = start; pos < end; ++pos) {
361 switch (lookupChange(pos).type) {
362 case Change::UNCHANGED:
363 // accept changes in nested inset
364 if (Inset * inset = getInset(pos))
365 inset->acceptChanges(bparams);
368 case Change::INSERTED:
369 d->changes_.set(Change(Change::UNCHANGED), pos);
370 // also accept changes in nested inset
371 if (Inset * inset = getInset(pos))
372 inset->acceptChanges(bparams);
375 case Change::DELETED:
376 // Suppress access to non-existent
377 // "end-of-paragraph char"
379 eraseChar(pos, false);
390 void Paragraph::rejectChanges(BufferParams const & bparams,
391 pos_type start, pos_type end)
393 LASSERT(start >= 0 && start <= size(), /**/);
394 LASSERT(end > start && end <= size() + 1, /**/);
396 for (pos_type pos = start; pos < end; ++pos) {
397 switch (lookupChange(pos).type) {
398 case Change::UNCHANGED:
399 // reject changes in nested inset
400 if (Inset * inset = getInset(pos))
401 inset->rejectChanges(bparams);
404 case Change::INSERTED:
405 // Suppress access to non-existent
406 // "end-of-paragraph char"
408 eraseChar(pos, false);
414 case Change::DELETED:
415 d->changes_.set(Change(Change::UNCHANGED), pos);
417 // Do NOT reject changes within a deleted inset!
418 // There may be insertions of a co-author inside of it!
426 void Paragraph::Private::insertChar(pos_type pos, char_type c,
427 Change const & change)
429 LASSERT(pos >= 0 && pos <= int(text_.size()), /**/);
432 changes_.insert(change, pos);
434 // This is actually very common when parsing buffers (and
435 // maybe inserting ascii text)
436 if (pos == pos_type(text_.size())) {
437 // when appending characters, no need to update tables
442 text_.insert(text_.begin() + pos, c);
444 // Update the font table.
445 fontlist_.increasePosAfterPos(pos);
448 insetlist_.increasePosAfterPos(pos);
452 bool Paragraph::insertInset(pos_type pos, Inset * inset,
453 Change const & change)
455 LASSERT(inset, /**/);
456 LASSERT(pos >= 0 && pos <= size(), /**/);
458 // Paragraph::insertInset() can be used in cut/copy/paste operation where
459 // d->inset_owner_ is not set yet.
460 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
463 d->insertChar(pos, META_INSET, change);
464 LASSERT(d->text_[pos] == META_INSET, /**/);
466 // Add a new entry in the insetlist_.
467 d->insetlist_.insert(inset, pos);
472 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
474 LASSERT(pos >= 0 && pos <= size(), return false);
476 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
479 Change change = d->changes_.lookup(pos);
481 // set the character to DELETED if
482 // a) it was previously unchanged or
483 // b) it was inserted by a co-author
485 if (change.type == Change::UNCHANGED ||
486 (change.type == Change::INSERTED && change.author != 0)) {
487 setChange(pos, Change(Change::DELETED));
491 if (change.type == Change::DELETED)
495 // Don't physically access the imaginary end-of-paragraph character.
496 // eraseChar() can only mark it as DELETED. A physical deletion of
497 // end-of-par must be handled externally.
503 d->changes_.erase(pos);
505 // if it is an inset, delete the inset entry
506 if (d->text_[pos] == META_INSET)
507 d->insetlist_.erase(pos);
509 d->text_.erase(d->text_.begin() + pos);
511 // Update the fontlist_
512 d->fontlist_.erase(pos);
514 // Update the insetlist_
515 d->insetlist_.decreasePosAfterPos(pos);
521 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
523 LASSERT(start >= 0 && start <= size(), /**/);
524 LASSERT(end >= start && end <= size() + 1, /**/);
527 for (pos_type count = end - start; count; --count) {
528 if (!eraseChar(i, trackChanges))
535 int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
536 char_type next, Encoding const & encoding)
538 // Writing next here may circumvent a possible font change between
539 // c and next. Since next is only output if it forms a surrogate pair
540 // with c we can ignore this:
541 // A font change inside a surrogate pair does not make sense and is
542 // hopefully impossible to input.
543 // FIXME: change tracking
544 // Is this correct WRT change tracking?
545 docstring const latex1 = encoding.latexChar(next);
546 docstring const latex2 = encoding.latexChar(c);
547 if (docstring(1, next) == latex1) {
548 // the encoding supports the combination
549 os << latex2 << latex1;
550 return latex1.length() + latex2.length();
552 os << latex1 << '{' << latex2 << '}';
553 return latex1.length() + latex2.length() + 2;
557 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
558 odocstream & os, TexRow & texrow,
560 unsigned int & column,
562 Layout const & style)
564 if (style.pass_thru || runparams.verbatim)
567 if (i + 1 < int(text_.size())) {
568 char_type next = text_[i + 1];
569 if (Encodings::isCombiningChar(next)) {
570 Encoding const & encoding = *(runparams.encoding);
571 // This space has an accent, so we must always output it.
572 column += latexSurrogatePair(os, ' ', next, encoding) - 1;
577 if (lyxrc.plaintext_linelen > 0
578 && column > lyxrc.plaintext_linelen
580 && text_[i - 1] != ' '
581 && (i + 1 < int(text_.size()))
582 // same in FreeSpacing mode
583 && !owner_->isFreeSpacing()
584 // In typewriter mode, we want to avoid
585 // ! . ? : at the end of a line
586 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
587 && (text_[i - 1] == '.'
588 || text_[i - 1] == '?'
589 || text_[i - 1] == ':'
590 || text_[i - 1] == '!'))) {
593 texrow.start(owner_->id(), i + 1);
595 } else if (style.free_spacing) {
604 int Paragraph::Private::writeScriptChars(odocstream & os,
605 docstring const & ltx,
606 Change const & runningChange,
607 Encoding const & encoding,
610 // FIXME: modifying i here is not very nice...
612 // We only arrive here when a proper language for character text_[i] has
613 // not been specified (i.e., it could not be translated in the current
614 // latex encoding) or its latex translation has been forced, and it
615 // belongs to a known script.
616 // Parameter ltx contains the latex translation of text_[i] as specified
617 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
618 // The latex macro name "textXXX" specifies the script to which text_[i]
619 // belongs and we use it in order to check whether characters from the
620 // same script immediately follow, such that we can collect them in a
621 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
622 // for the first char but only "<spec>" for all subsequent chars.
623 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
624 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
625 string script = to_ascii(ltx.substr(1, brace1 - 1));
628 bool closing_brace = true;
629 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
630 // Correct encoding is being used, so we can avoid \textgreek.
633 closing_brace = false;
635 os << ltx.substr(pos, length);
636 int size = text_.size();
637 while (i + 1 < size) {
638 char_type const next = text_[i + 1];
639 // Stop here if next character belongs to another script
640 // or there is a change in change tracking status.
641 if (!Encodings::isKnownScriptChar(next, script) ||
642 runningChange != owner_->lookupChange(i + 1))
646 FontList::const_iterator cit = fontlist_.begin();
647 FontList::const_iterator end = fontlist_.end();
648 for (; cit != end; ++cit) {
649 if (cit->pos() >= i && !found) {
650 prev_font = cit->font();
653 if (cit->pos() >= i + 1)
656 // Stop here if there is a font attribute or encoding change.
657 if (found && cit != end && prev_font != cit->font())
659 docstring const latex = encoding.latexChar(next);
660 docstring::size_type const b1 =
661 latex.find_first_of(from_ascii("{"));
662 docstring::size_type const b2 =
663 latex.find_last_of(from_ascii("}"));
664 int const len = b2 - b1 - 1;
665 os << latex.substr(b1 + 1, len);
677 bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
679 pos_type const len = str.length();
681 // is the paragraph large enough?
682 if (pos + len > int(text_.size()))
685 // does the wanted text start at point?
686 for (string::size_type i = 0; i < str.length(); ++i) {
687 // Caution: direct comparison of characters works only
688 // because str is pure ASCII.
689 if (str[i] != text_[pos + i])
693 return fontlist_.hasChangeInRange(pos, len);
697 void Paragraph::Private::latexInset(
698 BufferParams const & bparams,
701 OutputParams & runparams,
704 Font const & outerfont,
706 Change & running_change,
707 Layout const & style,
709 unsigned int & column)
711 Inset * inset = owner_->getInset(i);
712 LASSERT(inset, /**/);
714 if (style.pass_thru) {
715 inset->plaintext(os, runparams);
719 // FIXME: move this to InsetNewline::latex
720 if (inset->lyxCode() == NEWLINE_CODE) {
721 // newlines are handled differently here than
722 // the default in simpleTeXSpecialChars().
723 if (!style.newline_allowed) {
727 column += running_font.latexWriteEndChanges(
728 os, bparams, runparams,
733 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
736 basefont = owner_->getLayoutFont(bparams, outerfont);
737 running_font = basefont;
739 if (runparams.moving_arg)
744 texrow.start(owner_->id(), i + 1);
748 if (owner_->lookupChange(i).type == Change::DELETED) {
749 if( ++runparams.inDeletedInset == 1)
750 runparams.changeOfDeletedInset = owner_->lookupChange(i);
753 if (inset->canTrackChanges()) {
754 column += Changes::latexMarkChange(os, bparams, running_change,
755 Change(Change::UNCHANGED));
756 running_change = Change(Change::UNCHANGED);
760 odocstringstream ods;
762 if (inset->forceLTR()
763 && running_font.isRightToLeft()
764 // ERT is an exception, it should be output with no decorations at all
765 && inset->lyxCode() != ERT_CODE) {
766 if (running_font.language()->lang() == "farsi")
773 // FIXME: Bug: we can have an empty font change here!
774 // if there has just been a font change, we are going to close it
775 // right now, which means stupid latex code like \textsf{}. AFAIK,
776 // this does not harm dvi output. A minor bug, thus (JMarc)
778 // Some insets cannot be inside a font change command.
779 // However, even such insets *can* be placed in \L or \R
780 // or their equivalents (for RTL language switches), so we don't
781 // close the language in those cases.
782 // ArabTeX, though, cannot handle this special behavior, it seems.
783 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
784 || running_font.language()->lang() == "arabic_arabtex";
785 if (open_font && inset->noFontChange()) {
786 bool closeLanguage = arabtex
787 || basefont.isRightToLeft() == running_font.isRightToLeft();
788 unsigned int count = running_font.latexWriteEndChanges(ods,
789 bparams, runparams, basefont, basefont, closeLanguage);
791 // if any font properties were closed, update the running_font,
792 // making sure, however, to leave the language as it was
794 // FIXME: probably a better way to keep track of the old
795 // language, than copying the entire font?
796 Font const copy_font(running_font);
797 basefont = owner_->getLayoutFont(bparams, outerfont);
798 running_font = basefont;
800 running_font.setLanguage(copy_font.language());
801 // leave font open if language is still open
802 open_font = (running_font.language() == basefont.language());
804 runparams.local_font = &basefont;
811 tmp = inset->latex(ods, runparams);
812 } catch (EncodingException & e) {
813 // add location information and throw again.
820 if (running_font.language()->lang() == "farsi")
829 for (int j = 0; j < tmp; ++j)
832 texrow.start(owner_->id(), i + 1);
835 column += ods.str().size();
838 if (owner_->lookupChange(i).type == Change::DELETED)
839 --runparams.inDeletedInset;
843 void Paragraph::Private::latexSpecialChar(
845 OutputParams const & runparams,
846 Font const & running_font,
847 Change const & running_change,
848 Layout const & style,
850 unsigned int & column)
852 char_type const c = text_[i];
854 if (style.pass_thru) {
856 // FIXME UNICODE: This can fail if c cannot
857 // be encoded in the current encoding.
862 if (runparams.verbatim) {
863 // FIXME UNICODE: This can fail if c cannot
864 // be encoded in the current encoding.
869 if (lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
872 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
873 && latexSpecialTypewriter(c, os, i, column))
876 // Otherwise, we use what LaTeX provides us.
879 os << "\\textbackslash{}";
883 os << "\\textless{}";
887 os << "\\textgreater{}";
898 os << "\\char`\\\"{}";
903 case '%': case '#': case '{':
911 os << "\\textasciitilde{}";
916 os << "\\textasciicircum{}";
921 // avoid being mistaken for optional arguments
929 // Blanks are printed before font switching.
930 // Sure? I am not! (try nice-latex)
931 // I am sure it's correct. LyX might be smarter
932 // in the future, but for now, nothing wrong is
938 if (latexSpecialPhrase(os, i, column, runparams))
944 Encoding const & encoding = *(runparams.encoding);
945 if (i + 1 < int(text_.size())) {
946 char_type next = text_[i + 1];
947 if (Encodings::isCombiningChar(next)) {
948 column += latexSurrogatePair(os, c, next, encoding) - 1;
954 docstring const latex = encoding.latexChar(c);
955 if (Encodings::isKnownScriptChar(c, script)
956 && prefixIs(latex, from_ascii("\\" + script)))
957 column += writeScriptChars(os, latex,
958 running_change, encoding, i) - 1;
959 else if (latex.length() > 1 && latex[latex.length() - 1] != '}') {
960 // Prevent eating of a following
961 // space or command corruption by
962 // following characters
963 column += latex.length() + 1;
966 column += latex.length() - 1;
974 bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
975 pos_type i, unsigned int & column)
981 // In T1 encoding, these characters exist
982 // but we should avoid ligatures
983 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
985 os << "\\textcompwordmark{}";
992 // soul.sty breaks with \char`\"
993 os << "\\textquotedbl{}";
1002 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream & os,
1003 pos_type i, unsigned int & column)
1007 // within \ttfamily, "--" is merged to "-" (no endash)
1008 // so we avoid this rather irritating ligature
1009 if (i + 1 < int(text_.size()) && text_[i + 1] == '-') {
1016 // everything else has to be checked separately
1017 // (depending on the encoding)
1024 bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
1025 unsigned int & column, OutputParams const & runparams)
1027 // FIXME: if we have "LaTeX" with a font
1028 // change in the middle (before the 'T', then
1029 // the "TeX" part is still special cased.
1030 // Really we should only operate this on
1031 // "words" for some definition of word
1033 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1034 if (!isTextAt(special_phrases[pnr].phrase, i))
1036 if (runparams.moving_arg)
1038 os << special_phrases[pnr].macro;
1039 i += special_phrases[pnr].phrase.length() - 1;
1040 column += special_phrases[pnr].macro.length() - 1;
1047 void Paragraph::Private::validate(LaTeXFeatures & features,
1048 Layout const & layout) const
1050 // check the params.
1051 if (!params_.spacing().isDefault())
1052 features.require("setspace");
1055 features.useLayout(layout.name());
1058 fontlist_.validate(features);
1060 // then the indentation
1061 if (!params_.leftIndent().zero())
1062 features.require("ParagraphLeftIndent");
1065 InsetList::const_iterator icit = insetlist_.begin();
1066 InsetList::const_iterator iend = insetlist_.end();
1067 for (; icit != iend; ++icit) {
1069 icit->inset->validate(features);
1070 if (layout.needprotect &&
1071 icit->inset->lyxCode() == FOOT_CODE)
1072 features.require("NeedLyXFootnoteCode");
1076 // then the contents
1077 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1078 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1079 if (!special_phrases[pnr].builtin
1080 && isTextAt(special_phrases[pnr].phrase, i)) {
1081 features.require(special_phrases[pnr].phrase);
1085 Encodings::validate(text_[i], features);
1089 /////////////////////////////////////////////////////////////////////
1093 /////////////////////////////////////////////////////////////////////
1096 Layout const emptyParagraphLayout;
1099 Paragraph::Paragraph()
1100 : d(new Paragraph::Private(this, emptyParagraphLayout))
1107 Paragraph::Paragraph(Paragraph const & par)
1108 : itemdepth(par.itemdepth),
1109 d(new Paragraph::Private(*par.d, this))
1115 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1116 : itemdepth(par.itemdepth),
1117 d(new Paragraph::Private(*par.d, this, beg, end))
1123 Paragraph & Paragraph::operator=(Paragraph const & par)
1125 // needed as we will destroy the private part before copying it
1127 itemdepth = par.itemdepth;
1131 d = new Private(*par.d, this);
1138 Paragraph::~Paragraph()
1145 void Paragraph::write(ostream & os, BufferParams const & bparams,
1146 depth_type & dth) const
1148 // The beginning or end of a deeper (i.e. nested) area?
1149 if (dth != d->params_.depth()) {
1150 if (d->params_.depth() > dth) {
1151 while (d->params_.depth() > dth) {
1152 os << "\n\\begin_deeper";
1156 while (d->params_.depth() < dth) {
1157 os << "\n\\end_deeper";
1163 // First write the layout
1164 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1166 d->params_.write(os);
1168 Font font1(inherit_font, bparams.language);
1170 Change running_change = Change(Change::UNCHANGED);
1173 for (pos_type i = 0; i <= size(); ++i) {
1175 Change change = lookupChange(i);
1176 Changes::lyxMarkChange(os, column, running_change, change);
1177 running_change = change;
1182 // Write font changes
1183 Font const & font2 = getFontSettings(bparams, i);
1184 if (font2 != font1) {
1185 font2.lyxWriteChanges(font1, os);
1190 char_type const c = d->text_[i];
1193 if (Inset const * inset = getInset(i)) {
1194 if (inset->directWrite()) {
1195 // international char, let it write
1196 // code directly so it's shorter in
1202 os << "\\begin_inset ";
1204 os << "\n\\end_inset\n\n";
1210 os << "\n\\backslash\n";
1214 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1221 if ((column > 70 && c == ' ')
1226 // this check is to amend a bug. LyX sometimes
1227 // inserts '\0' this could cause problems.
1229 os << to_utf8(docstring(1, c));
1231 LYXERR0("NUL char in structure.");
1237 os << "\n\\end_layout\n";
1241 void Paragraph::validate(LaTeXFeatures & features) const
1243 d->validate(features, *d->layout_);
1247 void Paragraph::insert(pos_type start, docstring const & str,
1248 Font const & font, Change const & change)
1250 for (size_t i = 0, n = str.size(); i != n ; ++i)
1251 insertChar(start + i, str[i], font, change);
1255 void Paragraph::appendChar(char_type c, Font const & font,
1256 Change const & change)
1259 d->changes_.insert(change, d->text_.size());
1260 // when appending characters, no need to update tables
1261 d->text_.push_back(c);
1262 setFont(d->text_.size() - 1, font);
1266 void Paragraph::appendString(docstring const & s, Font const & font,
1267 Change const & change)
1269 pos_type end = s.size();
1270 size_t oldsize = d->text_.size();
1271 size_t newsize = oldsize + end;
1272 size_t capacity = d->text_.capacity();
1273 if (newsize >= capacity)
1274 d->text_.reserve(max(capacity + 100, newsize));
1276 // when appending characters, no need to update tables
1279 // FIXME: Optimize this!
1280 for (pos_type i = oldsize; i != newsize; ++i) {
1282 d->changes_.insert(change, i);
1284 d->fontlist_.set(oldsize, font);
1285 d->fontlist_.set(newsize - 1, font);
1289 void Paragraph::insertChar(pos_type pos, char_type c,
1292 d->insertChar(pos, c, Change(trackChanges ?
1293 Change::INSERTED : Change::UNCHANGED));
1297 void Paragraph::insertChar(pos_type pos, char_type c,
1298 Font const & font, bool trackChanges)
1300 d->insertChar(pos, c, Change(trackChanges ?
1301 Change::INSERTED : Change::UNCHANGED));
1306 void Paragraph::insertChar(pos_type pos, char_type c,
1307 Font const & font, Change const & change)
1309 d->insertChar(pos, c, change);
1314 bool Paragraph::insertInset(pos_type pos, Inset * inset,
1315 Font const & font, Change const & change)
1317 bool const success = insertInset(pos, inset, change);
1318 // Set the font/language of the inset...
1324 void Paragraph::resetFonts(Font const & font)
1326 d->fontlist_.clear();
1327 d->fontlist_.set(0, font);
1328 d->fontlist_.set(d->text_.size() - 1, font);
1331 // Gets uninstantiated font setting at position.
1332 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1336 LYXERR0("pos: " << pos << " size: " << size());
1337 LASSERT(pos <= size(), /**/);
1340 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1341 if (cit != d->fontlist_.end())
1344 if (pos == size() && !empty())
1345 return getFontSettings(bparams, pos - 1);
1347 // Optimisation: avoid a full font instantiation if there is no
1348 // language change from previous call.
1349 static Font previous_font;
1350 static Language const * previous_lang = 0;
1351 Language const * lang = getParLanguage(bparams);
1352 if (lang != previous_lang) {
1353 previous_lang = lang;
1354 previous_font = Font(inherit_font, lang);
1356 return previous_font;
1360 FontSpan Paragraph::fontSpan(pos_type pos) const
1362 LASSERT(pos <= size(), /**/);
1365 FontList::const_iterator cit = d->fontlist_.begin();
1366 FontList::const_iterator end = d->fontlist_.end();
1367 for (; cit != end; ++cit) {
1368 if (cit->pos() >= pos) {
1369 if (pos >= beginOfBody())
1370 return FontSpan(max(start, beginOfBody()),
1373 return FontSpan(start,
1374 min(beginOfBody() - 1,
1377 start = cit->pos() + 1;
1380 // This should not happen, but if so, we take no chances.
1381 // LYXERR0("Paragraph::getEndPosOfFontSpan: This should not happen!");
1382 return FontSpan(pos, pos);
1386 // Gets uninstantiated font setting at position 0
1387 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1389 if (!empty() && !d->fontlist_.empty())
1390 return d->fontlist_.begin()->font();
1392 // Optimisation: avoid a full font instantiation if there is no
1393 // language change from previous call.
1394 static Font previous_font;
1395 static Language const * previous_lang = 0;
1396 if (bparams.language != previous_lang) {
1397 previous_lang = bparams.language;
1398 previous_font = Font(inherit_font, bparams.language);
1401 return previous_font;
1405 // Gets the fully instantiated font at a given position in a paragraph
1406 // This is basically the same function as Text::GetFont() in text2.cpp.
1407 // The difference is that this one is used for generating the LaTeX file,
1408 // and thus cosmetic "improvements" are disallowed: This has to deliver
1409 // the true picture of the buffer. (Asger)
1410 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1411 Font const & outerfont) const
1413 LASSERT(pos >= 0, /**/);
1415 Font font = getFontSettings(bparams, pos);
1417 pos_type const body_pos = beginOfBody();
1418 FontInfo & fi = font.fontInfo();
1420 fi.realize(d->layout_->labelfont);
1422 fi.realize(d->layout_->font);
1424 fi.realize(outerfont.fontInfo());
1425 fi.realize(bparams.getFont().fontInfo());
1431 Font const Paragraph::getLabelFont
1432 (BufferParams const & bparams, Font const & outerfont) const
1434 FontInfo tmpfont = d->layout_->labelfont;
1435 tmpfont.realize(outerfont.fontInfo());
1436 tmpfont.realize(bparams.getFont().fontInfo());
1437 return Font(tmpfont, getParLanguage(bparams));
1441 Font const Paragraph::getLayoutFont
1442 (BufferParams const & bparams, Font const & outerfont) const
1444 FontInfo tmpfont = d->layout_->font;
1445 tmpfont.realize(outerfont.fontInfo());
1446 tmpfont.realize(bparams.getFont().fontInfo());
1447 return Font(tmpfont, getParLanguage(bparams));
1451 /// Returns the height of the highest font in range
1452 FontSize Paragraph::highestFontInRange
1453 (pos_type startpos, pos_type endpos, FontSize def_size) const
1455 return d->fontlist_.highestInRange(startpos, endpos, def_size);
1459 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1461 char_type c = d->text_[pos];
1462 if (!lyxrc.rtl_support)
1492 if (uc != c && getFontSettings(bparams, pos).isRightToLeft())
1498 void Paragraph::setFont(pos_type pos, Font const & font)
1500 LASSERT(pos <= size(), /**/);
1502 // First, reduce font against layout/label font
1503 // Update: The setCharFont() routine in text2.cpp already
1504 // reduces font, so we don't need to do that here. (Asger)
1506 d->fontlist_.set(pos, font);
1510 void Paragraph::makeSameLayout(Paragraph const & par)
1512 d->layout_ = par.d->layout_;
1513 d->params_ = par.d->params_;
1517 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1519 if (isFreeSpacing())
1525 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1526 if (eraseChar(pos, trackChanges))
1532 return count > 0 || pos > 0;
1536 bool Paragraph::hasSameLayout(Paragraph const & par) const
1538 return par.d->layout_ == d->layout_
1539 && d->params_.sameLayout(par.d->params_);
1543 depth_type Paragraph::getDepth() const
1545 return d->params_.depth();
1549 depth_type Paragraph::getMaxDepthAfter() const
1551 if (d->layout_->isEnvironment())
1552 return d->params_.depth() + 1;
1554 return d->params_.depth();
1558 char Paragraph::getAlign() const
1560 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1561 return d->layout_->align;
1563 return d->params_.align();
1567 docstring const & Paragraph::labelString() const
1569 return d->params_.labelString();
1573 // the next two functions are for the manual labels
1574 docstring const Paragraph::getLabelWidthString() const
1576 if (d->layout_->margintype == MARGIN_MANUAL)
1577 return d->params_.labelWidthString();
1579 return _("Senseless with this layout!");
1583 void Paragraph::setLabelWidthString(docstring const & s)
1585 d->params_.labelWidthString(s);
1589 docstring const Paragraph::translateIfPossible(docstring const & s,
1590 BufferParams const & bparams) const
1592 if (!isAscii(s) || s.empty()) {
1593 // This must be a user defined layout. We cannot translate
1594 // this, since gettext accepts only ascii keys.
1597 // Probably standard layout, try to translate
1598 Messages & m = getMessages(getParLanguage(bparams)->code());
1599 return m.get(to_ascii(s));
1603 docstring Paragraph::expandLabel(Layout const & layout,
1604 BufferParams const & bparams, bool process_appendix) const
1606 DocumentClass const & tclass = bparams.documentClass();
1609 if (process_appendix && d->params_.appendix())
1610 fmt = translateIfPossible(layout.labelstring_appendix(),
1613 fmt = translateIfPossible(layout.labelstring(), bparams);
1615 if (fmt.empty() && layout.labeltype == LABEL_COUNTER
1616 && !layout.counter.empty())
1617 fmt = "\\the" + layout.counter;
1619 // handle 'inherited level parts' in 'fmt',
1620 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
1621 size_t const i = fmt.find('@', 0);
1622 if (i != docstring::npos) {
1623 size_t const j = fmt.find('@', i + 1);
1624 if (j != docstring::npos) {
1625 docstring parent(fmt, i + 1, j - i - 1);
1626 docstring label = from_ascii("??");
1627 if (tclass.hasLayout(parent))
1628 docstring label = expandLabel(tclass[parent], bparams,
1630 fmt = docstring(fmt, 0, i) + label
1631 + docstring(fmt, j + 1, docstring::npos);
1635 return tclass.counters().counterLabel(fmt);
1639 void Paragraph::applyLayout(Layout const & new_layout)
1641 d->layout_ = &new_layout;
1642 LyXAlignment const oldAlign = d->params_.align();
1644 if (!(oldAlign & d->layout_->alignpossible)) {
1645 frontend::Alert::warning(_("Alignment not permitted"),
1646 _("The new layout does not permit the alignment previously used.\nSetting to default."));
1647 d->params_.align(LYX_ALIGN_LAYOUT);
1652 pos_type Paragraph::beginOfBody() const
1654 return d->begin_of_body_;
1658 void Paragraph::setBeginOfBody()
1660 if (d->layout_->labeltype != LABEL_MANUAL) {
1661 d->begin_of_body_ = 0;
1665 // Unroll the first two cycles of the loop
1666 // and remember the previous character to
1667 // remove unnecessary getChar() calls
1669 pos_type end = size();
1670 if (i < end && !isNewline(i)) {
1672 char_type previous_char = 0;
1675 previous_char = d->text_[i];
1676 if (!isNewline(i)) {
1678 while (i < end && previous_char != ' ') {
1683 previous_char = temp;
1689 d->begin_of_body_ = i;
1693 bool Paragraph::forcePlainLayout() const
1695 return inInset().forcePlainLayout();
1699 bool Paragraph::allowParagraphCustomization() const
1701 return inInset().allowParagraphCustomization();
1705 bool Paragraph::usePlainLayout() const
1707 return inInset().usePlainLayout();
1713 // paragraphs inside floats need different alignment tags to avoid
1716 bool noTrivlistCentering(InsetCode code)
1718 return code == FLOAT_CODE
1719 || code == WRAP_CODE
1720 || code == CELL_CODE;
1724 string correction(string const & orig)
1726 if (orig == "flushleft")
1727 return "raggedright";
1728 if (orig == "flushright")
1729 return "raggedleft";
1730 if (orig == "center")
1736 string const corrected_env(string const & suffix, string const & env,
1737 InsetCode code, bool const lastpar)
1739 string output = suffix + "{";
1740 if (noTrivlistCentering(code)) {
1742 // the last paragraph in non-trivlist-aligned
1743 // context is special (to avoid unwanted whitespace)
1744 if (suffix == "\\begin")
1745 return "\\" + correction(env) + "{}";
1748 output += correction(env);
1752 if (suffix == "\\begin")
1758 void adjust_row_column(string const & str, TexRow & texrow, int & column)
1760 if (!contains(str, "\n"))
1761 column += str.size();
1765 column = rsplit(str, tmp, '\n').size();
1772 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
1773 odocstream & os, TexRow & texrow,
1774 OutputParams const & runparams) const
1778 if (params_.noindent()) {
1779 os << "\\noindent ";
1783 LyXAlignment const curAlign = params_.align();
1785 if (curAlign == layout_->align)
1789 case LYX_ALIGN_NONE:
1790 case LYX_ALIGN_BLOCK:
1791 case LYX_ALIGN_LAYOUT:
1792 case LYX_ALIGN_SPECIAL:
1794 case LYX_ALIGN_LEFT:
1795 case LYX_ALIGN_RIGHT:
1796 case LYX_ALIGN_CENTER:
1797 if (runparams.moving_arg) {
1804 string const begin_tag = "\\begin";
1805 InsetCode code = owner_->ownerCode();
1806 bool const lastpar = runparams.isLastPar;
1809 case LYX_ALIGN_NONE:
1810 case LYX_ALIGN_BLOCK:
1811 case LYX_ALIGN_LAYOUT:
1812 case LYX_ALIGN_SPECIAL:
1814 case LYX_ALIGN_LEFT: {
1816 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1817 output = corrected_env(begin_tag, "flushleft", code, lastpar);
1819 output = corrected_env(begin_tag, "flushright", code, lastpar);
1820 os << from_ascii(output);
1821 adjust_row_column(output, texrow, column);
1823 } case LYX_ALIGN_RIGHT: {
1825 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1826 output = corrected_env(begin_tag, "flushright", code, lastpar);
1828 output = corrected_env(begin_tag, "flushleft", code, lastpar);
1829 os << from_ascii(output);
1830 adjust_row_column(output, texrow, column);
1832 } case LYX_ALIGN_CENTER: {
1834 output = corrected_env(begin_tag, "center", code, lastpar);
1835 os << from_ascii(output);
1836 adjust_row_column(output, texrow, column);
1845 int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
1846 odocstream & os, TexRow & texrow,
1847 OutputParams const & runparams) const
1851 switch (params_.align()) {
1852 case LYX_ALIGN_NONE:
1853 case LYX_ALIGN_BLOCK:
1854 case LYX_ALIGN_LAYOUT:
1855 case LYX_ALIGN_SPECIAL:
1857 case LYX_ALIGN_LEFT:
1858 case LYX_ALIGN_RIGHT:
1859 case LYX_ALIGN_CENTER:
1860 if (runparams.moving_arg) {
1867 string const end_tag = "\n\\par\\end";
1868 InsetCode code = owner_->ownerCode();
1869 bool const lastpar = runparams.isLastPar;
1871 switch (params_.align()) {
1872 case LYX_ALIGN_NONE:
1873 case LYX_ALIGN_BLOCK:
1874 case LYX_ALIGN_LAYOUT:
1875 case LYX_ALIGN_SPECIAL:
1877 case LYX_ALIGN_LEFT: {
1879 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1880 output = corrected_env(end_tag, "flushleft", code, lastpar);
1882 output = corrected_env(end_tag, "flushright", code, lastpar);
1883 os << from_ascii(output);
1884 adjust_row_column(output, texrow, column);
1886 } case LYX_ALIGN_RIGHT: {
1888 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1889 output = corrected_env(end_tag, "flushright", code, lastpar);
1891 output = corrected_env(end_tag, "flushleft", code, lastpar);
1892 os << from_ascii(output);
1893 adjust_row_column(output, texrow, column);
1895 } case LYX_ALIGN_CENTER: {
1897 output = corrected_env(end_tag, "center", code, lastpar);
1898 os << from_ascii(output);
1899 adjust_row_column(output, texrow, column);
1908 // This one spits out the text of the paragraph
1909 bool Paragraph::latex(BufferParams const & bparams,
1910 Font const & outerfont,
1911 odocstream & os, TexRow & texrow,
1912 OutputParams const & runparams) const
1914 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
1916 bool return_value = false;
1918 bool const allowcust = allowParagraphCustomization();
1920 // FIXME This check should not be needed. Perhaps issue an
1921 // error if it triggers.
1922 Layout const & style = forcePlainLayout() ?
1923 bparams.documentClass().plainLayout() : *d->layout_;
1925 // Current base font for all inherited font changes, without any
1926 // change caused by an individual character, except for the language:
1927 // It is set to the language of the first character.
1928 // As long as we are in the label, this font is the base font of the
1929 // label. Before the first body character it is set to the base font
1933 // Maybe we have to create a optional argument.
1934 pos_type body_pos = beginOfBody();
1935 unsigned int column = 0;
1938 // the optional argument is kept in curly brackets in
1939 // case it contains a ']'
1942 basefont = getLabelFont(bparams, outerfont);
1944 basefont = getLayoutFont(bparams, outerfont);
1947 // Which font is currently active?
1948 Font running_font(basefont);
1949 // Do we have an open font change?
1950 bool open_font = false;
1952 Change runningChange = Change(Change::UNCHANGED);
1954 texrow.start(id(), 0);
1956 // if the paragraph is empty, the loop will not be entered at all
1958 if (style.isCommand()) {
1963 column += d->startTeXParParams(bparams, os, texrow,
1967 for (pos_type i = 0; i < size(); ++i) {
1968 // First char in paragraph or after label?
1969 if (i == body_pos) {
1972 column += running_font.latexWriteEndChanges(
1973 os, bparams, runparams,
1974 basefont, basefont);
1977 basefont = getLayoutFont(bparams, outerfont);
1978 running_font = basefont;
1980 column += Changes::latexMarkChange(os, bparams,
1981 runningChange, Change(Change::UNCHANGED));
1982 runningChange = Change(Change::UNCHANGED);
1987 if (style.isCommand()) {
1993 column += d->startTeXParParams(bparams, os,
1998 Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
2001 if (bparams.outputChanges && runningChange != change) {
2003 column += running_font.latexWriteEndChanges(
2004 os, bparams, runparams, basefont, basefont);
2007 basefont = getLayoutFont(bparams, outerfont);
2008 running_font = basefont;
2010 column += Changes::latexMarkChange(os, bparams, runningChange, change);
2011 runningChange = change;
2014 // do not output text which is marked deleted
2015 // if change tracking output is disabled
2016 if (!bparams.outputChanges && change.type == Change::DELETED) {
2022 // Fully instantiated font
2023 Font const font = getFont(bparams, i, outerfont);
2025 Font const last_font = running_font;
2027 // Do we need to close the previous font?
2029 (font != running_font ||
2030 font.language() != running_font.language()))
2032 column += running_font.latexWriteEndChanges(
2033 os, bparams, runparams, basefont,
2034 (i == body_pos-1) ? basefont : font);
2035 running_font = basefont;
2039 // close babel's font environment before opening CJK.
2040 if (!running_font.language()->babel().empty() &&
2041 font.language()->encoding()->package() == Encoding::CJK) {
2042 string end_tag = subst(lyxrc.language_command_end,
2044 running_font.language()->babel());
2045 os << from_ascii(end_tag);
2046 column += end_tag.length();
2049 // Switch file encoding if necessary (and allowed)
2050 if (!runparams.verbatim &&
2051 runparams.encoding->package() != Encoding::none &&
2052 font.language()->encoding()->package() != Encoding::none) {
2053 odocstringstream ods;
2054 pair<bool, int> const enc_switch = switchEncoding(ods, bparams,
2055 runparams, *(font.language()->encoding()));
2057 if (enc_switch.first) {
2058 column += enc_switch.second;
2059 runparams.encoding = font.language()->encoding();
2063 char_type const c = d->text_[i];
2065 // Do we need to change font?
2066 if ((font != running_font ||
2067 font.language() != running_font.language()) &&
2070 odocstringstream ods;
2071 column += font.latexWriteStartChanges(ods, bparams,
2072 runparams, basefont,
2074 running_font = font;
2076 docstring fontchange = ods.str();
2077 // check if the fontchange ends with a trailing blank
2078 // (like "\small " (see bug 3382)
2079 if (suffixIs(fontchange, ' ') && c == ' ')
2080 os << fontchange.substr(0, fontchange.size() - 1)
2081 << from_ascii("{}");
2087 // FIXME: integrate this case in latexSpecialChar
2088 // Do not print the separation of the optional argument
2089 // if style.pass_thru is false. This works because
2090 // latexSpecialChar ignores spaces if
2091 // style.pass_thru is false.
2092 if (i != body_pos - 1) {
2093 if (d->simpleTeXBlanks(
2094 runparams, os, texrow,
2095 i, column, font, style)) {
2096 // A surrogate pair was output. We
2097 // must not call latexSpecialChar
2098 // in this iteration, since it would output
2099 // the combining character again.
2106 OutputParams rp = runparams;
2107 rp.free_spacing = style.free_spacing;
2108 rp.local_font = &font;
2109 rp.intitle = style.intitle;
2111 // Two major modes: LaTeX or plain
2112 // Handle here those cases common to both modes
2113 // and then split to handle the two modes separately.
2114 if (c == META_INSET)
2115 d->latexInset(bparams, os,
2116 texrow, rp, running_font,
2117 basefont, outerfont, open_font,
2118 runningChange, style, i, column);
2121 d->latexSpecialChar(os, rp, running_font, runningChange,
2123 } catch (EncodingException & e) {
2124 if (runparams.dryrun) {
2125 os << "<" << _("LyX Warning: ")
2126 << _("uncodable character") << " '";
2130 // add location information and throw again.
2138 // Set the encoding to that returned from latexSpecialChar (see
2139 // comment for encoding member in OutputParams.h)
2140 runparams.encoding = rp.encoding;
2143 // If we have an open font definition, we have to close it
2145 #ifdef FIXED_LANGUAGE_END_DETECTION
2148 .latexWriteEndChanges(os, bparams, runparams,
2150 next_->getFont(bparams, 0, outerfont));
2152 running_font.latexWriteEndChanges(os, bparams,
2153 runparams, basefont, basefont);
2156 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2157 //FIXME: there as we start another \selectlanguage with the next paragraph if
2158 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2159 running_font.latexWriteEndChanges(os, bparams, runparams,
2160 basefont, basefont);
2164 column += Changes::latexMarkChange(os, bparams, runningChange, Change(Change::UNCHANGED));
2166 // Needed if there is an optional argument but no contents.
2167 if (body_pos > 0 && body_pos == size()) {
2169 return_value = false;
2173 column += d->endTeXParParams(bparams, os, texrow,
2177 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2178 return return_value;
2182 bool Paragraph::emptyTag() const
2184 for (pos_type i = 0; i < size(); ++i) {
2185 if (Inset const * inset = getInset(i)) {
2186 InsetCode lyx_code = inset->lyxCode();
2187 if (lyx_code != TOC_CODE &&
2188 lyx_code != INCLUDE_CODE &&
2189 lyx_code != GRAPHICS_CODE &&
2190 lyx_code != ERT_CODE &&
2191 lyx_code != LISTINGS_CODE &&
2192 lyx_code != FLOAT_CODE &&
2193 lyx_code != TABULAR_CODE) {
2197 char_type c = d->text_[i];
2198 if (c != ' ' && c != '\t')
2206 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2209 for (pos_type i = 0; i < size(); ++i) {
2210 if (Inset const * inset = getInset(i)) {
2211 InsetCode lyx_code = inset->lyxCode();
2212 if (lyx_code == LABEL_CODE) {
2213 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2214 docstring const & id = il->getParam("name");
2215 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2223 pos_type Paragraph::firstWord(odocstream & os, OutputParams const & runparams)
2227 for (i = 0; i < size(); ++i) {
2228 if (Inset const * inset = getInset(i)) {
2229 inset->docbook(os, runparams);
2231 char_type c = d->text_[i];
2234 os << sgml::escapeChar(c);
2241 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2244 pos_type size = text_.size();
2245 for (pos_type i = initial; i < size; ++i) {
2246 Font font = owner_->getFont(buf.params(), i, outerfont);
2247 if (text_[i] == META_INSET)
2249 if (i != initial && font != font_old)
2258 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2260 OutputParams const & runparams,
2261 Font const & outerfont,
2262 pos_type initial) const
2264 bool emph_flag = false;
2266 Layout const & style = *d->layout_;
2268 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2270 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2273 // parsing main loop
2274 for (pos_type i = initial; i < size(); ++i) {
2275 Font font = getFont(buf.params(), i, outerfont);
2277 // handle <emphasis> tag
2278 if (font_old.emph() != font.fontInfo().emph()) {
2279 if (font.fontInfo().emph() == FONT_ON) {
2282 } else if (i != initial) {
2283 os << "</emphasis>";
2288 if (Inset const * inset = getInset(i)) {
2289 inset->docbook(os, runparams);
2291 char_type c = d->text_[i];
2293 if (style.pass_thru)
2296 os << sgml::escapeChar(c);
2298 font_old = font.fontInfo();
2302 os << "</emphasis>";
2305 if (style.free_spacing)
2307 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2312 bool Paragraph::isHfill(pos_type pos) const
2314 Inset const * inset = getInset(pos);
2315 return inset && (inset->lyxCode() == SPACE_CODE &&
2316 inset->isStretchableSpace());
2320 bool Paragraph::isNewline(pos_type pos) const
2322 Inset const * inset = getInset(pos);
2323 return inset && inset->lyxCode() == NEWLINE_CODE;
2327 bool Paragraph::isLineSeparator(pos_type pos) const
2329 char_type const c = d->text_[pos];
2330 if (isLineSeparatorChar(c))
2332 Inset const * inset = getInset(pos);
2333 return inset && inset->isLineSeparator();
2337 /// Used by the spellchecker
2338 bool Paragraph::isLetter(pos_type pos) const
2340 if (Inset const * inset = getInset(pos))
2341 return inset->isLetter();
2342 char_type const c = d->text_[pos];
2343 return isLetterChar(c) || isDigit(c);
2347 bool Paragraph::isChar(pos_type pos) const
2349 if (Inset const * inset = getInset(pos))
2350 return inset->isChar();
2351 char_type const c = d->text_[pos];
2352 return !isLetterChar(c) && !isDigit(c) && !lyx::isSpace(c);
2356 bool Paragraph::isSpace(pos_type pos) const
2358 if (Inset const * inset = getInset(pos))
2359 return inset->isSpace();
2360 char_type const c = d->text_[pos];
2361 return lyx::isSpace(c);
2366 Paragraph::getParLanguage(BufferParams const & bparams) const
2369 return getFirstFontSettings(bparams).language();
2370 // FIXME: we should check the prev par as well (Lgb)
2371 return bparams.language;
2375 bool Paragraph::isRTL(BufferParams const & bparams) const
2377 return lyxrc.rtl_support
2378 && getParLanguage(bparams)->rightToLeft()
2379 && ownerCode() != ERT_CODE
2380 && ownerCode() != LISTINGS_CODE;
2384 void Paragraph::changeLanguage(BufferParams const & bparams,
2385 Language const * from, Language const * to)
2387 // change language including dummy font change at the end
2388 for (pos_type i = 0; i <= size(); ++i) {
2389 Font font = getFontSettings(bparams, i);
2390 if (font.language() == from) {
2391 font.setLanguage(to);
2398 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
2400 Language const * doc_language = bparams.language;
2401 FontList::const_iterator cit = d->fontlist_.begin();
2402 FontList::const_iterator end = d->fontlist_.end();
2404 for (; cit != end; ++cit)
2405 if (cit->font().language() != ignore_language &&
2406 cit->font().language() != latex_language &&
2407 cit->font().language() != doc_language)
2413 docstring Paragraph::asString(int options) const
2415 return asString(0, size(), options);
2419 docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
2421 odocstringstream os;
2424 && options & AS_STR_LABEL
2425 && !d->params_.labelString().empty())
2426 os << d->params_.labelString() << ' ';
2428 for (pos_type i = beg; i < end; ++i) {
2429 char_type const c = d->text_[i];
2430 if (isPrintable(c) || c == '\t'
2431 || (c == '\n' && options & AS_STR_NEWLINES))
2433 else if (c == META_INSET && options & AS_STR_INSETS)
2434 getInset(i)->tocString(os);
2441 void Paragraph::setInsetOwner(Inset const * inset)
2443 d->inset_owner_ = inset;
2447 int Paragraph::id() const
2453 Layout const & Paragraph::layout() const
2459 void Paragraph::setLayout(Layout const & layout)
2461 d->layout_ = &layout;
2465 void Paragraph::setDefaultLayout(DocumentClass const & tc)
2467 setLayout(tc.defaultLayout());
2471 void Paragraph::setPlainLayout(DocumentClass const & tc)
2473 setLayout(tc.plainLayout());
2477 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
2479 if (usePlainLayout())
2480 setPlainLayout(tclass);
2482 setDefaultLayout(tclass);
2486 Inset const & Paragraph::inInset() const
2488 LASSERT(d->inset_owner_, throw ExceptionMessage(BufferException,
2489 _("Memory problem"), _("Paragraph not properly initiliazed")));
2490 return *d->inset_owner_;
2494 InsetCode Paragraph::ownerCode() const
2496 return d->inset_owner_ ? d->inset_owner_->lyxCode() : NO_CODE;
2500 ParagraphParameters & Paragraph::params()
2506 ParagraphParameters const & Paragraph::params() const
2512 bool Paragraph::isFreeSpacing() const
2514 if (d->layout_->free_spacing)
2516 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
2520 bool Paragraph::allowEmpty() const
2522 if (d->layout_->keepempty)
2524 return d->inset_owner_ && d->inset_owner_->allowEmpty();
2528 char_type Paragraph::transformChar(char_type c, pos_type pos) const
2530 if (!Encodings::isArabicChar(c))
2533 char_type prev_char = ' ';
2534 char_type next_char = ' ';
2536 for (pos_type i = pos - 1; i >= 0; --i) {
2537 char_type const par_char = d->text_[i];
2538 if (!Encodings::isArabicComposeChar(par_char)) {
2539 prev_char = par_char;
2544 for (pos_type i = pos + 1, end = size(); i < end; ++i) {
2545 char_type const par_char = d->text_[i];
2546 if (!Encodings::isArabicComposeChar(par_char)) {
2547 next_char = par_char;
2552 if (Encodings::isArabicChar(next_char)) {
2553 if (Encodings::isArabicChar(prev_char) &&
2554 !Encodings::isArabicSpecialChar(prev_char))
2555 return Encodings::transformChar(c, Encodings::FORM_MEDIAL);
2557 return Encodings::transformChar(c, Encodings::FORM_INITIAL);
2559 if (Encodings::isArabicChar(prev_char) &&
2560 !Encodings::isArabicSpecialChar(prev_char))
2561 return Encodings::transformChar(c, Encodings::FORM_FINAL);
2563 return Encodings::transformChar(c, Encodings::FORM_ISOLATED);
2568 int Paragraph::checkBiblio(Buffer const & buffer)
2571 // This is getting more and more a mess. ...We really should clean
2572 // up this bibitem issue for 1.6. See also bug 2743.
2574 // Add bibitem insets if necessary
2575 if (d->layout_->labeltype != LABEL_BIBLIO)
2578 bool hasbibitem = !d->insetlist_.empty()
2579 // Insist on it being in pos 0
2580 && d->text_[0] == META_INSET
2581 && d->insetlist_.begin()->inset->lyxCode() == BIBITEM_CODE;
2583 bool track_changes = buffer.params().trackChanges;
2588 // remove a bibitem in pos != 0
2589 // restore it later in pos 0 if necessary
2590 // (e.g. if a user inserts contents _before_ the item)
2591 // we're assuming there's only one of these, which there
2593 int erasedInsetPosition = -1;
2594 InsetList::iterator it = d->insetlist_.begin();
2595 InsetList::iterator end = d->insetlist_.end();
2596 for (; it != end; ++it)
2597 if (it->inset->lyxCode() == BIBITEM_CODE
2599 InsetBibitem * olditem = static_cast<InsetBibitem *>(it->inset);
2600 oldkey = olditem->getParam("key");
2601 oldlabel = olditem->getParam("label");
2602 erasedInsetPosition = it->pos;
2603 eraseChar(erasedInsetPosition, track_changes);
2607 // There was an InsetBibitem at the beginning, and we didn't
2608 // have to erase one.
2609 if (hasbibitem && erasedInsetPosition < 0)
2612 // There was an InsetBibitem at the beginning and we did have to
2613 // erase one. So we give its properties to the beginning inset.
2615 InsetBibitem * inset =
2616 static_cast<InsetBibitem *>(d->insetlist_.begin()->inset);
2617 if (!oldkey.empty())
2618 inset->setParam("key", oldkey);
2619 inset->setParam("label", oldlabel);
2620 return -erasedInsetPosition;
2623 // There was no inset at the beginning, so we need to create one with
2624 // the key and label of the one we erased.
2625 InsetBibitem * inset =
2626 new InsetBibitem(buffer, InsetCommandParams(BIBITEM_CODE));
2627 // restore values of previously deleted item in this par.
2628 if (!oldkey.empty())
2629 inset->setParam("key", oldkey);
2630 inset->setParam("label", oldlabel);
2631 insertInset(0, static_cast<Inset *>(inset),
2632 Change(track_changes ? Change::INSERTED : Change::UNCHANGED));
2638 void Paragraph::checkAuthors(AuthorList const & authorList)
2640 d->changes_.checkAuthors(authorList);
2644 bool Paragraph::isUnchanged(pos_type pos) const
2646 return lookupChange(pos).type == Change::UNCHANGED;
2650 bool Paragraph::isInserted(pos_type pos) const
2652 return lookupChange(pos).type == Change::INSERTED;
2656 bool Paragraph::isDeleted(pos_type pos) const
2658 return lookupChange(pos).type == Change::DELETED;
2662 InsetList const & Paragraph::insetList() const
2664 return d->insetlist_;
2668 void Paragraph::setBuffer(Buffer & b)
2670 d->insetlist_.setBuffer(b);
2674 Inset * Paragraph::releaseInset(pos_type pos)
2676 Inset * inset = d->insetlist_.release(pos);
2677 /// does not honour change tracking!
2678 eraseChar(pos, false);
2683 Inset * Paragraph::getInset(pos_type pos)
2685 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
2686 ? d->insetlist_.get(pos) : 0;
2690 Inset const * Paragraph::getInset(pos_type pos) const
2692 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
2693 ? d->insetlist_.get(pos) : 0;
2697 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
2698 pos_type & right, TextCase action)
2700 // process sequences of modified characters; in change
2701 // tracking mode, this approach results in much better
2702 // usability than changing case on a char-by-char basis
2705 bool const trackChanges = bparams.trackChanges;
2707 bool capitalize = true;
2709 for (; pos < right; ++pos) {
2710 char_type oldChar = d->text_[pos];
2711 char_type newChar = oldChar;
2713 // ignore insets and don't play with deleted text!
2714 if (oldChar != META_INSET && !isDeleted(pos)) {
2716 case text_lowercase:
2717 newChar = lowercase(oldChar);
2719 case text_capitalization:
2721 newChar = uppercase(oldChar);
2725 case text_uppercase:
2726 newChar = uppercase(oldChar);
2731 if (!isLetter(pos) || isDeleted(pos)) {
2732 // permit capitalization again
2736 if (oldChar != newChar)
2739 if (oldChar == newChar || pos == right - 1) {
2740 if (oldChar != newChar) {
2741 // step behind the changing area
2744 int erasePos = pos - changes.size();
2745 for (size_t i = 0; i < changes.size(); i++) {
2746 insertChar(pos, changes[i],
2747 getFontSettings(bparams,
2750 if (!eraseChar(erasePos, trackChanges)) {
2753 ++right; // expand selection
2762 bool Paragraph::find(docstring const & str, bool cs, bool mw,
2763 pos_type pos, bool del) const
2765 int const strsize = str.length();
2767 pos_type const parsize = d->text_.size();
2768 for (i = 0; pos + i < parsize; ++i) {
2771 if (cs && str[i] != d->text_[pos + i])
2773 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos + i]))
2775 if (!del && isDeleted(pos + i))
2782 // if necessary, check whether string matches word
2784 if (pos > 0 && isLetter(pos - 1))
2786 if (pos + strsize < parsize
2787 && isLetter(pos + strsize))
2795 char_type Paragraph::getChar(pos_type pos) const
2797 return d->text_[pos];
2801 pos_type Paragraph::size() const
2803 return d->text_.size();
2807 bool Paragraph::empty() const
2809 return d->text_.empty();
2813 bool Paragraph::isInset(pos_type pos) const
2815 return d->text_[pos] == META_INSET;
2819 bool Paragraph::isSeparator(pos_type pos) const
2821 //FIXME: Are we sure this can be the only separator?
2822 return d->text_[pos] == ' ';
2826 void Paragraph::deregisterWords()
2828 Private::Words::const_iterator it;
2829 WordList & wl = theWordList();
2830 for (it = d->words_.begin(); it != d->words_.end(); ++it)
2836 void Paragraph::collectWords(CursorSlice const & sl)
2839 bool inword = false;
2841 //lyxerr << "Words: ";
2842 pos_type n = size();
2843 for (pos_type pos = 0; pos != n; ++pos) {
2847 if (!isLetter(pos)) {
2856 CursorSlice from = sl;
2857 CursorSlice to = sl;
2860 from.text()->getWord(from, to, WHOLE_WORD);
2861 if (to.pos() - from.pos() < 6)
2863 docstring word = asString(from.pos(), to.pos(), false);
2864 d->words_.insert(word);
2865 //lyxerr << word << " ";
2867 //lyxerr << std::endl;
2871 void Paragraph::registerWords()
2873 Private::Words::const_iterator it;
2874 WordList & wl = theWordList();
2875 for (it = d->words_.begin(); it != d->words_.end(); ++it)
2880 void Paragraph::updateWords(CursorSlice const & sl)
2882 LASSERT(&sl.paragraph() == this, /**/);