?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
* qt_helpers.cpp:
[lyx.git] / src / Paragraph.cpp
blob6f1c408aba621a66bf3bbece0cd359db3b04b7bb
1 /**
2 * \file Paragraph.cpp
3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
5 *
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Jean-Marc Lasgouttes
9 * \author Angus Leeming
10 * \author John Levon
11 * \author André Pönitz
12 * \author Dekel Tsur
13 * \author Jürgen Vigna
14 *
15 * Full author contact details are available in file CREDITS.
16 */
17
18 #include <config.h>
19
20 #include "Paragraph.h"
21
22 #include "LayoutFile.h"
23 #include "Buffer.h"
24 #include "BufferParams.h"
25 #include "Changes.h"
26 #include "Counters.h"
27 #include "Encoding.h"
28 #include "InsetList.h"
29 #include "Language.h"
30 #include "LaTeXFeatures.h"
31 #include "Layout.h"
32 #include "Length.h"
33 #include "Font.h"
34 #include "FontList.h"
35 #include "LyXRC.h"
36 #include "OutputParams.h"
37 #include "output_latex.h"
38 #include "output_xhtml.h"
39 #include "paragraph_funcs.h"
40 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
42 #include "sgml.h"
43 #include "TextClass.h"
44 #include "TexRow.h"
45 #include "Text.h"
46 #include "VSpace.h"
47 #include "WordLangTuple.h"
48 #include "WordList.h"
49
50 #include "frontends/alert.h"
51
52 #include "insets/InsetBibitem.h"
53 #include "insets/InsetLabel.h"
54
55 #include "support/debug.h"
56 #include "support/docstring_list.h"
57 #include "support/ExceptionMessage.h"
58 #include "support/gettext.h"
59 #include "support/lassert.h"
60 #include "support/lstrings.h"
61 #include "support/textutils.h"
62
63 #include <sstream>
64 #include <vector>
65
66 using namespace std;
67 using namespace lyx::support;
68
69 namespace lyx {
70
71 namespace {
72 /// Inset identifier (above 0x10ffff, for ucs-4)
73 char_type const META_INSET = 0x200001;
74 };
75
76 /////////////////////////////////////////////////////////////////////
77 //
78 // Paragraph::Private
79 //
80 /////////////////////////////////////////////////////////////////////
81
82 class Paragraph::Private
83 {
84 public:
85 ///
86 Private(Paragraph * owner, Layout const & layout);
87 /// "Copy constructor"
88 Private(Private const &, Paragraph * owner);
89 /// Copy constructor from \p beg to \p end
90 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
91
92 ///
93 void insertChar(pos_type pos, char_type c, Change const & change);
94
95 /// Output the surrogate pair formed by \p c and \p next to \p os.
96 /// \return the number of characters written.
97 int latexSurrogatePair(odocstream & os, char_type c, char_type next,
98 OutputParams const &);
99
100 /// Output a space in appropriate formatting (or a surrogate pair
101 /// if the next character is a combining character).
102 /// \return whether a surrogate pair was output.
103 bool simpleTeXBlanks(OutputParams const &,
104 odocstream &, TexRow & texrow,
105 pos_type i,
106 unsigned int & column,
107 Font const & font,
108 Layout const & style);
109
110 /// Output consecutive unicode chars, belonging to the same script as
111 /// specified by the latex macro \p ltx, to \p os starting from \p i.
112 /// \return the number of characters written.
113 int writeScriptChars(odocstream & os, docstring const & ltx,
114 Change const &, Encoding const &, pos_type & i);
115
116 /// This could go to ParagraphParameters if we want to.
117 int startTeXParParams(BufferParams const &, odocstream &, TexRow &,
118 OutputParams const &) const;
119
120 /// This could go to ParagraphParameters if we want to.
121 int endTeXParParams(BufferParams const &, odocstream &, TexRow &,
122 OutputParams const &) const;
123
124 ///
125 void latexInset(BufferParams const &,
126 odocstream &,
127 TexRow & texrow, OutputParams &,
128 Font & running_font,
129 Font & basefont,
130 Font const & outerfont,
131 bool & open_font,
132 Change & running_change,
133 Layout const & style,
134 pos_type & i,
135 unsigned int & column);
136
137 ///
138 void latexSpecialChar(
139 odocstream & os,
140 OutputParams const & runparams,
141 Font const & running_font,
142 Change const & running_change,
143 Layout const & style,
144 pos_type & i,
145 unsigned int & column);
146
147 ///
148 bool latexSpecialT1(
149 char_type const c,
150 odocstream & os,
151 pos_type i,
152 unsigned int & column);
153 ///
154 bool latexSpecialTypewriter(
155 char_type const c,
156 odocstream & os,
157 pos_type i,
158 unsigned int & column);
159 ///
160 bool latexSpecialPhrase(
161 odocstream & os,
162 pos_type & i,
163 unsigned int & column,
164 OutputParams const & runparams);
165
166 ///
167 void validate(LaTeXFeatures & features,
168 Layout const & layout) const;
169
170 /// Checks if the paragraph contains only text and no inset or font change.
171 bool onlyText(Buffer const & buf, Font const & outerfont,
172 pos_type initial) const;
173
174 /// match a string against a particular point in the paragraph
175 bool isTextAt(string const & str, pos_type pos) const;
176
177 /// Which Paragraph owns us?
178 Paragraph * owner_;
179
180 /// In which Inset?
181 Inset const * inset_owner_;
182
183 ///
184 FontList fontlist_;
185
186 ///
187 unsigned int id_;
188 ///
189 static unsigned int paragraph_id;
190 ///
191 ParagraphParameters params_;
192
193 /// for recording and looking up changes
194 Changes changes_;
195
196 ///
197 InsetList insetlist_;
198
199 /// end of label
200 pos_type begin_of_body_;
201
202 typedef docstring TextContainer;
203 ///
204 TextContainer text_;
205
206 typedef std::set<docstring> Words;
207 ///
208 Words words_;
209 ///
210 Layout const * layout_;
211 };
212
213
214 // Initialization of the counter for the paragraph id's,
215 unsigned int Paragraph::Private::paragraph_id = 0;
216
217 namespace {
218
219 struct special_phrase {
220 string phrase;
221 docstring macro;
222 bool builtin;
223 };
224
225 special_phrase const special_phrases[] = {
226 { "LyX", from_ascii("\\LyX{}"), false },
227 { "TeX", from_ascii("\\TeX{}"), true },
228 { "LaTeX2e", from_ascii("\\LaTeXe{}"), true },
229 { "LaTeX", from_ascii("\\LaTeX{}"), true },
230 };
231
232 size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase);
233
234 } // namespace anon
235
236
237 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
238 : owner_(owner), inset_owner_(0), begin_of_body_(0), layout_(&layout)
239 {
240 id_ = paragraph_id++;
241 text_.reserve(100);
242 }
243
244
245 Paragraph::Private::Private(Private const & p, Paragraph * owner)
246 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
247 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
248 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
249 layout_(p.layout_)
250 {
251 id_ = paragraph_id++;
252 }
253
254
255 Paragraph::Private::Private(Private const & p, Paragraph * owner,
256 pos_type beg, pos_type end)
257 : owner_(owner), inset_owner_(p.inset_owner_),
258 params_(p.params_), changes_(p.changes_),
259 insetlist_(p.insetlist_, beg, end),
260 begin_of_body_(p.begin_of_body_), words_(p.words_),
261 layout_(p.layout_)
262 {
263 id_ = paragraph_id++;
264 if (beg >= pos_type(p.text_.size()))
265 return;
266 text_ = p.text_.substr(beg, end - beg);
267
268 FontList::const_iterator fcit = fontlist_.begin();
269 FontList::const_iterator fend = fontlist_.end();
270 for (; fcit != fend; ++fcit) {
271 if (fcit->pos() < beg)
272 continue;
273 if (fcit->pos() >= end) {
274 // Add last entry in the fontlist_.
275 fontlist_.set(text_.size() - 1, fcit->font());
276 break;
277 }
278 // Add a new entry in the fontlist_.
279 fontlist_.set(fcit->pos() - beg, fcit->font());
280 }
281 }
282
283
284 void Paragraph::addChangesToToc(DocIterator const & cdit,
285 Buffer const & buf) const
286 {
287 d->changes_.addToToc(cdit, buf);
288 }
289
290
291 bool Paragraph::isDeleted(pos_type start, pos_type end) const
292 {
293 LASSERT(start >= 0 && start <= size(), /**/);
294 LASSERT(end > start && end <= size() + 1, /**/);
295
296 return d->changes_.isDeleted(start, end);
297 }
298
299
300 bool Paragraph::isChanged(pos_type start, pos_type end) const
301 {
302 LASSERT(start >= 0 && start <= size(), /**/);
303 LASSERT(end > start && end <= size() + 1, /**/);
304
305 return d->changes_.isChanged(start, end);
306 }
307
308
309 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
310 {
311 // keep the logic here in sync with the logic of eraseChars()
312 if (!trackChanges)
313 return true;
314
315 Change const change = d->changes_.lookup(size());
316 return change.inserted() && change.currentAuthor();
317 }
318
319
320 void Paragraph::setChange(Change const & change)
321 {
322 // beware of the imaginary end-of-par character!
323 d->changes_.set(change, 0, size() + 1);
324
325 /*
326 * Propagate the change recursively - but not in case of DELETED!
327 *
328 * Imagine that your co-author makes changes in an existing inset. He
329 * sends your document to you and you come to the conclusion that the
330 * inset should go completely. If you erase it, LyX must not delete all
331 * text within the inset. Otherwise, the change tracked insertions of
332 * your co-author get lost and there is no way to restore them later.
333 *
334 * Conclusion: An inset's content should remain untouched if you delete it
335 */
336
337 if (!change.deleted()) {
338 for (pos_type pos = 0; pos < size(); ++pos) {
339 if (Inset * inset = getInset(pos))
340 inset->setChange(change);
341 }
342 }
343 }
344
345
346 void Paragraph::setChange(pos_type pos, Change const & change)
347 {
348 LASSERT(pos >= 0 && pos <= size(), /**/);
349 d->changes_.set(change, pos);
350
351 // see comment in setChange(Change const &) above
352 if (!change.deleted() && pos < size())
353 if (Inset * inset = getInset(pos))
354 inset->setChange(change);
355 }
356
357
358 Change const & Paragraph::lookupChange(pos_type pos) const
359 {
360 LASSERT(pos >= 0 && pos <= size(), /**/);
361 return d->changes_.lookup(pos);
362 }
363
364
365 void Paragraph::acceptChanges(pos_type start, pos_type end)
366 {
367 LASSERT(start >= 0 && start <= size(), /**/);
368 LASSERT(end > start && end <= size() + 1, /**/);
369
370 for (pos_type pos = start; pos < end; ++pos) {
371 switch (lookupChange(pos).type) {
372 case Change::UNCHANGED:
373 // accept changes in nested inset
374 if (Inset * inset = getInset(pos))
375 inset->acceptChanges();
376 break;
377
378 case Change::INSERTED:
379 d->changes_.set(Change(Change::UNCHANGED), pos);
380 // also accept changes in nested inset
381 if (Inset * inset = getInset(pos))
382 inset->acceptChanges();
383 break;
384
385 case Change::DELETED:
386 // Suppress access to non-existent
387 // "end-of-paragraph char"
388 if (pos < size()) {
389 eraseChar(pos, false);
390 --end;
391 --pos;
392 }
393 break;
394 }
395
396 }
397 }
398
399
400 void Paragraph::rejectChanges(pos_type start, pos_type end)
401 {
402 LASSERT(start >= 0 && start <= size(), /**/);
403 LASSERT(end > start && end <= size() + 1, /**/);
404
405 for (pos_type pos = start; pos < end; ++pos) {
406 switch (lookupChange(pos).type) {
407 case Change::UNCHANGED:
408 // reject changes in nested inset
409 if (Inset * inset = getInset(pos))
410 inset->rejectChanges();
411 break;
412
413 case Change::INSERTED:
414 // Suppress access to non-existent
415 // "end-of-paragraph char"
416 if (pos < size()) {
417 eraseChar(pos, false);
418 --end;
419 --pos;
420 }
421 break;
422
423 case Change::DELETED:
424 d->changes_.set(Change(Change::UNCHANGED), pos);
425
426 // Do NOT reject changes within a deleted inset!
427 // There may be insertions of a co-author inside of it!
428
429 break;
430 }
431 }
432 }
433
434
435 void Paragraph::Private::insertChar(pos_type pos, char_type c,
436 Change const & change)
437 {
438 LASSERT(pos >= 0 && pos <= int(text_.size()), /**/);
439
440 // track change
441 changes_.insert(change, pos);
442
443 // This is actually very common when parsing buffers (and
444 // maybe inserting ascii text)
445 if (pos == pos_type(text_.size())) {
446 // when appending characters, no need to update tables
447 text_.push_back(c);
448 return;
449 }
450
451 text_.insert(text_.begin() + pos, c);
452
453 // Update the font table.
454 fontlist_.increasePosAfterPos(pos);
455
456 // Update the insets
457 insetlist_.increasePosAfterPos(pos);
458 }
459
460
461 bool Paragraph::insertInset(pos_type pos, Inset * inset,
462 Change const & change)
463 {
464 LASSERT(inset, /**/);
465 LASSERT(pos >= 0 && pos <= size(), /**/);
466
467 // Paragraph::insertInset() can be used in cut/copy/paste operation where
468 // d->inset_owner_ is not set yet.
469 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
470 return false;
471
472 d->insertChar(pos, META_INSET, change);
473 LASSERT(d->text_[pos] == META_INSET, /**/);
474
475 // Add a new entry in the insetlist_.
476 d->insetlist_.insert(inset, pos);
477 return true;
478 }
479
480
481 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
482 {
483 LASSERT(pos >= 0 && pos <= size(), return false);
484
485 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
486
487 if (trackChanges) {
488 Change change = d->changes_.lookup(pos);
489
490 // set the character to DELETED if
491 // a) it was previously unchanged or
492 // b) it was inserted by a co-author
493
494 if (!change.changed() ||
495 (change.inserted() && !change.currentAuthor())) {
496 setChange(pos, Change(Change::DELETED));
497 return false;
498 }
499
500 if (change.deleted())
501 return false;
502 }
503
504 // Don't physically access the imaginary end-of-paragraph character.
505 // eraseChar() can only mark it as DELETED. A physical deletion of
506 // end-of-par must be handled externally.
507 if (pos == size()) {
508 return false;
509 }
510
511 // track change
512 d->changes_.erase(pos);
513
514 // if it is an inset, delete the inset entry
515 if (d->text_[pos] == META_INSET)
516 d->insetlist_.erase(pos);
517
518 d->text_.erase(d->text_.begin() + pos);
519
520 // Update the fontlist_
521 d->fontlist_.erase(pos);
522
523 // Update the insetlist_
524 d->insetlist_.decreasePosAfterPos(pos);
525
526 return true;
527 }
528
529
530 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
531 {
532 LASSERT(start >= 0 && start <= size(), /**/);
533 LASSERT(end >= start && end <= size() + 1, /**/);
534
535 pos_type i = start;
536 for (pos_type count = end - start; count; --count) {
537 if (!eraseChar(i, trackChanges))
538 ++i;
539 }
540 return end - i;
541 }
542
543
544 int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
545 char_type next, OutputParams const & runparams)
546 {
547 // Writing next here may circumvent a possible font change between
548 // c and next. Since next is only output if it forms a surrogate pair
549 // with c we can ignore this:
550 // A font change inside a surrogate pair does not make sense and is
551 // hopefully impossible to input.
552 // FIXME: change tracking
553 // Is this correct WRT change tracking?
554 Encoding const & encoding = *(runparams.encoding);
555 docstring const latex1 = encoding.latexChar(next);
556 docstring const latex2 = encoding.latexChar(c);
557 if (docstring(1, next) == latex1) {
558 // the encoding supports the combination
559 os << latex2 << latex1;
560 return latex1.length() + latex2.length();
561 } else if (runparams.local_font &&
562 runparams.local_font->language()->lang() == "polutonikogreek") {
563 // polutonikogreek only works without the brackets
564 os << latex1 << latex2;
565 return latex1.length() + latex2.length();
566 } else
567 os << latex1 << '{' << latex2 << '}';
568 return latex1.length() + latex2.length() + 2;
569 }
570
571
572 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
573 odocstream & os, TexRow & texrow,
574 pos_type i,
575 unsigned int & column,
576 Font const & font,
577 Layout const & style)
578 {
579 if (style.pass_thru || runparams.verbatim)
580 return false;
581
582 if (i + 1 < int(text_.size())) {
583 char_type next = text_[i + 1];
584 if (Encodings::isCombiningChar(next)) {
585 // This space has an accent, so we must always output it.
586 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
587 return true;
588 }
589 }
590
591 if (lyxrc.plaintext_linelen > 0
592 && column > lyxrc.plaintext_linelen
593 && i
594 && text_[i - 1] != ' '
595 && (i + 1 < int(text_.size()))
596 // same in FreeSpacing mode
597 && !owner_->isFreeSpacing()
598 // In typewriter mode, we want to avoid
599 // ! . ? : at the end of a line
600 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
601 && (text_[i - 1] == '.'
602 || text_[i - 1] == '?'
603 || text_[i - 1] == ':'
604 || text_[i - 1] == '!'))) {
605 os << '\n';
606 texrow.newline();
607 texrow.start(owner_->id(), i + 1);
608 column = 0;
609 } else if (style.free_spacing) {
610 os << '~';
611 } else {
612 os << ' ';
613 }
614 return false;
615 }
616
617
618 int Paragraph::Private::writeScriptChars(odocstream & os,
619 docstring const & ltx,
620 Change const & runningChange,
621 Encoding const & encoding,
622 pos_type & i)
623 {
624 // FIXME: modifying i here is not very nice...
625
626 // We only arrive here when a proper language for character text_[i] has
627 // not been specified (i.e., it could not be translated in the current
628 // latex encoding) or its latex translation has been forced, and it
629 // belongs to a known script.
630 // Parameter ltx contains the latex translation of text_[i] as specified
631 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
632 // The latex macro name "textXXX" specifies the script to which text_[i]
633 // belongs and we use it in order to check whether characters from the
634 // same script immediately follow, such that we can collect them in a
635 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
636 // for the first char but only "<spec>" for all subsequent chars.
637 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
638 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
639 string script = to_ascii(ltx.substr(1, brace1 - 1));
640 int pos = 0;
641 int length = brace2;
642 bool closing_brace = true;
643 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
644 // Correct encoding is being used, so we can avoid \textgreek.
645 pos = brace1 + 1;
646 length -= pos;
647 closing_brace = false;
648 }
649 os << ltx.substr(pos, length);
650 int size = text_.size();
651 while (i + 1 < size) {
652 char_type const next = text_[i + 1];
653 // Stop here if next character belongs to another script
654 // or there is a change in change tracking status.
655 if (!Encodings::isKnownScriptChar(next, script) ||
656 runningChange != owner_->lookupChange(i + 1))
657 break;
658 Font prev_font;
659 bool found = false;
660 FontList::const_iterator cit = fontlist_.begin();
661 FontList::const_iterator end = fontlist_.end();
662 for (; cit != end; ++cit) {
663 if (cit->pos() >= i && !found) {
664 prev_font = cit->font();
665 found = true;
666 }
667 if (cit->pos() >= i + 1)
668 break;
669 }
670 // Stop here if there is a font attribute or encoding change.
671 if (found && cit != end && prev_font != cit->font())
672 break;
673 docstring const latex = encoding.latexChar(next);
674 docstring::size_type const b1 =
675 latex.find_first_of(from_ascii("{"));
676 docstring::size_type const b2 =
677 latex.find_last_of(from_ascii("}"));
678 int const len = b2 - b1 - 1;
679 os << latex.substr(b1 + 1, len);
680 length += len;
681 ++i;
682 }
683 if (closing_brace) {
684 os << '}';
685 ++length;
686 }
687 return length;
688 }
689
690
691 bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
692 {
693 pos_type const len = str.length();
694
695 // is the paragraph large enough?
696 if (pos + len > int(text_.size()))
697 return false;
698
699 // does the wanted text start at point?
700 for (string::size_type i = 0; i < str.length(); ++i) {
701 // Caution: direct comparison of characters works only
702 // because str is pure ASCII.
703 if (str[i] != text_[pos + i])
704 return false;
705 }
706
707 return fontlist_.hasChangeInRange(pos, len);
708 }
709
710
711 void Paragraph::Private::latexInset(
712 BufferParams const & bparams,
713 odocstream & os,
714 TexRow & texrow,
715 OutputParams & runparams,
716 Font & running_font,
717 Font & basefont,
718 Font const & outerfont,
719 bool & open_font,
720 Change & running_change,
721 Layout const & style,
722 pos_type & i,
723 unsigned int & column)
724 {
725 Inset * inset = owner_->getInset(i);
726 LASSERT(inset, /**/);
727
728 if (style.pass_thru) {
729 inset->plaintext(os, runparams);
730 return;
731 }
732
733 // FIXME: move this to InsetNewline::latex
734 if (inset->lyxCode() == NEWLINE_CODE) {
735 // newlines are handled differently here than
736 // the default in simpleTeXSpecialChars().
737 if (!style.newline_allowed) {
738 os << '\n';
739 } else {
740 if (open_font) {
741 column += running_font.latexWriteEndChanges(
742 os, bparams, runparams,
743 basefont, basefont);
744 open_font = false;
745 }
746
747 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
748 os << '~';
749
750 basefont = owner_->getLayoutFont(bparams, outerfont);
751 running_font = basefont;
752
753 if (runparams.moving_arg)
754 os << "\\protect ";
755
756 }
757 texrow.newline();
758 texrow.start(owner_->id(), i + 1);
759 column = 0;
760 }
761
762 if (owner_->isDeleted(i)) {
763 if( ++runparams.inDeletedInset == 1)
764 runparams.changeOfDeletedInset = owner_->lookupChange(i);
765 }
766
767 if (inset->canTrackChanges()) {
768 column += Changes::latexMarkChange(os, bparams, running_change,
769 Change(Change::UNCHANGED));
770 running_change = Change(Change::UNCHANGED);
771 }
772
773 bool close = false;
774 odocstream::pos_type const len = os.tellp();
775
776 if (inset->forceLTR()
777 && running_font.isRightToLeft()
778 // ERT is an exception, it should be output with no
779 // decorations at all
780 && inset->lyxCode() != ERT_CODE) {
781 if (running_font.language()->lang() == "farsi")
782 os << "\\beginL{}";
783 else
784 os << "\\L{";
785 close = true;
786 }
787
788 // FIXME: Bug: we can have an empty font change here!
789 // if there has just been a font change, we are going to close it
790 // right now, which means stupid latex code like \textsf{}. AFAIK,
791 // this does not harm dvi output. A minor bug, thus (JMarc)
792
793 // Some insets cannot be inside a font change command.
794 // However, even such insets *can* be placed in \L or \R
795 // or their equivalents (for RTL language switches), so we don't
796 // close the language in those cases.
797 // ArabTeX, though, cannot handle this special behavior, it seems.
798 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
799 || running_font.language()->lang() == "arabic_arabtex";
800 if (open_font && inset->noFontChange()) {
801 bool closeLanguage = arabtex
802 || basefont.isRightToLeft() == running_font.isRightToLeft();
803 unsigned int count = running_font.latexWriteEndChanges(os,
804 bparams, runparams, basefont, basefont, closeLanguage);
805 column += count;
806 // if any font properties were closed, update the running_font,
807 // making sure, however, to leave the language as it was
808 if (count > 0) {
809 // FIXME: probably a better way to keep track of the old
810 // language, than copying the entire font?
811 Font const copy_font(running_font);
812 basefont = owner_->getLayoutFont(bparams, outerfont);
813 running_font = basefont;
814 if (!closeLanguage)
815 running_font.setLanguage(copy_font.language());
816 // leave font open if language is still open
817 open_font = (running_font.language() == basefont.language());
818 if (closeLanguage)
819 runparams.local_font = &basefont;
820 }
821 }
822
823 int tmp;
824
825 try {
826 tmp = inset->latex(os, runparams);
827 } catch (EncodingException & e) {
828 // add location information and throw again.
829 e.par_id = id_;
830 e.pos = i;
831 throw(e);
832 }
833
834 if (close) {
835 if (running_font.language()->lang() == "farsi")
836 os << "\\endL{}";
837 else
838 os << '}';
839 }
840
841 if (tmp) {
842 for (int j = 0; j < tmp; ++j)
843 texrow.newline();
844
845 texrow.start(owner_->id(), i + 1);
846 column = 0;
847 } else {
848 column += os.tellp() - len;
849 }
850
851 if (owner_->isDeleted(i))
852 --runparams.inDeletedInset;
853 }
854
855
856 void Paragraph::Private::latexSpecialChar(
857 odocstream & os,
858 OutputParams const & runparams,
859 Font const & running_font,
860 Change const & running_change,
861 Layout const & style,
862 pos_type & i,
863 unsigned int & column)
864 {
865 char_type const c = text_[i];
866
867 if (style.pass_thru) {
868 if (c != '\0')
869 // FIXME UNICODE: This can fail if c cannot
870 // be encoded in the current encoding.
871 os.put(c);
872 return;
873 }
874
875 if (runparams.verbatim) {
876 // FIXME UNICODE: This can fail if c cannot
877 // be encoded in the current encoding.
878 os.put(c);
879 return;
880 }
881
882 // If T1 font encoding is used, use the special
883 // characters it provides.
884 // NOTE: some languages reset the font encoding
885 // internally
886 if (!running_font.language()->internalFontEncoding()
887 && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
888 return;
889
890 // \tt font needs special treatment
891 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
892 && latexSpecialTypewriter(c, os, i, column))
893 return;
894
895 // Otherwise, we use what LaTeX provides us.
896 switch (c) {
897 case '\\':
898 os << "\\textbackslash{}";
899 column += 15;
900 break;
901 case '<':
902 os << "\\textless{}";
903 column += 10;
904 break;
905 case '>':
906 os << "\\textgreater{}";
907 column += 13;
908 break;
909 case '|':
910 os << "\\textbar{}";
911 column += 9;
912 break;
913 case '-':
914 os << '-';
915 break;
916 case '\"':
917 os << "\\char`\\\"{}";
918 column += 9;
919 break;
920
921 case '$': case '&':
922 case '%': case '#': case '{':
923 case '}': case '_':
924 os << '\\';
925 os.put(c);
926 column += 1;
927 break;
928
929 case '~':
930 os << "\\textasciitilde{}";
931 column += 16;
932 break;
933
934 case '^':
935 os << "\\textasciicircum{}";
936 column += 17;
937 break;
938
939 case '*':
940 case '[':
941 case ']':
942 // avoid being mistaken for optional arguments
943 os << '{';
944 os.put(c);
945 os << '}';
946 column += 2;
947 break;
948
949 case ' ':
950 // Blanks are printed before font switching.
951 // Sure? I am not! (try nice-latex)
952 // I am sure it's correct. LyX might be smarter
953 // in the future, but for now, nothing wrong is
954 // written. (Asger)
955 break;
956
957 default:
958 // LyX, LaTeX etc.
959 if (latexSpecialPhrase(os, i, column, runparams))
960 return;
961
962 if (c == '\0')
963 return;
964
965 Encoding const & encoding = *(runparams.encoding);
966 if (i + 1 < int(text_.size())) {
967 char_type next = text_[i + 1];
968 if (Encodings::isCombiningChar(next)) {
969 column += latexSurrogatePair(os, c, next, runparams) - 1;
970 ++i;
971 break;
972 }
973 }
974 string script;
975 docstring const latex = encoding.latexChar(c);
976 if (Encodings::isKnownScriptChar(c, script)
977 && prefixIs(latex, from_ascii("\\" + script)))
978 column += writeScriptChars(os, latex,
979 running_change, encoding, i) - 1;
980 else if (latex.length() > 1 && latex[latex.length() - 1] != '}') {
981 // Prevent eating of a following
982 // space or command corruption by
983 // following characters
984 column += latex.length() + 1;
985 os << latex << "{}";
986 } else {
987 column += latex.length() - 1;
988 os << latex;
989 }
990 break;
991 }
992 }
993
994
995 bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
996 pos_type i, unsigned int & column)
997 {
998 switch (c) {
999 case '>':
1000 case '<':
1001 os.put(c);
1002 // In T1 encoding, these characters exist
1003 // but we should avoid ligatures
1004 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1005 return true;
1006 os << "\\textcompwordmark{}";
1007 column += 19;
1008 return true;
1009 case '|':
1010 os.put(c);
1011 return true;
1012 case '\"':
1013 // soul.sty breaks with \char`\"
1014 os << "\\textquotedbl{}";
1015 column += 14;
1016 return true;
1017 default:
1018 return false;
1019 }
1020 }
1021
1022
1023 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream & os,
1024 pos_type i, unsigned int & column)
1025 {
1026 switch (c) {
1027 case '-':
1028 // within \ttfamily, "--" is merged to "-" (no endash)
1029 // so we avoid this rather irritating ligature
1030 if (i + 1 < int(text_.size()) && text_[i + 1] == '-') {
1031 os << "-{}";
1032 column += 2;
1033 } else
1034 os << '-';
1035 return true;
1036
1037 // everything else has to be checked separately
1038 // (depending on the encoding)
1039 default:
1040 return false;
1041 }
1042 }
1043
1044
1045 bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
1046 unsigned int & column, OutputParams const & runparams)
1047 {
1048 // FIXME: if we have "LaTeX" with a font
1049 // change in the middle (before the 'T', then
1050 // the "TeX" part is still special cased.
1051 // Really we should only operate this on
1052 // "words" for some definition of word
1053
1054 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1055 if (!isTextAt(special_phrases[pnr].phrase, i))
1056 continue;
1057 if (runparams.moving_arg)
1058 os << "\\protect";
1059 os << special_phrases[pnr].macro;
1060 i += special_phrases[pnr].phrase.length() - 1;
1061 column += special_phrases[pnr].macro.length() - 1;
1062 return true;
1063 }
1064 return false;
1065 }
1066
1067
1068 void Paragraph::Private::validate(LaTeXFeatures & features,
1069 Layout const & layout) const
1070 {
1071 // check the params.
1072 if (!params_.spacing().isDefault())
1073 features.require("setspace");
1074
1075 // then the layouts
1076 features.useLayout(layout.name());
1077
1078 // then the fonts
1079 fontlist_.validate(features);
1080
1081 // then the indentation
1082 if (!params_.leftIndent().zero())
1083 features.require("ParagraphLeftIndent");
1084
1085 // then the insets
1086 InsetList::const_iterator icit = insetlist_.begin();
1087 InsetList::const_iterator iend = insetlist_.end();
1088 for (; icit != iend; ++icit) {
1089 if (icit->inset) {
1090 icit->inset->validate(features);
1091 if (layout.needprotect &&
1092 icit->inset->lyxCode() == FOOT_CODE)
1093 features.require("NeedLyXFootnoteCode");
1094 }
1095 }
1096
1097 // then the contents
1098 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1099 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1100 if (!special_phrases[pnr].builtin
1101 && isTextAt(special_phrases[pnr].phrase, i)) {
1102 features.require(special_phrases[pnr].phrase);
1103 break;
1104 }
1105 }
1106 Encodings::validate(text_[i], features);
1107 }
1108 }
1109
1110 /////////////////////////////////////////////////////////////////////
1111 //
1112 // Paragraph
1113 //
1114 /////////////////////////////////////////////////////////////////////
1115
1116 namespace {
1117 Layout const emptyParagraphLayout;
1118 }
1119
1120 Paragraph::Paragraph()
1121 : d(new Paragraph::Private(this, emptyParagraphLayout))
1122 {
1123 itemdepth = 0;
1124 d->params_.clear();
1125 }
1126
1127
1128 Paragraph::Paragraph(Paragraph const & par)
1129 : itemdepth(par.itemdepth),
1130 d(new Paragraph::Private(*par.d, this))
1131 {
1132 registerWords();
1133 }
1134
1135
1136 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1137 : itemdepth(par.itemdepth),
1138 d(new Paragraph::Private(*par.d, this, beg, end))
1139 {
1140 registerWords();
1141 }
1142
1143
1144 Paragraph & Paragraph::operator=(Paragraph const & par)
1145 {
1146 // needed as we will destroy the private part before copying it
1147 if (&par != this) {
1148 itemdepth = par.itemdepth;
1149
1150 deregisterWords();
1151 delete d;
1152 d = new Private(*par.d, this);
1153 registerWords();
1154 }
1155 return *this;
1156 }
1157
1158
1159 Paragraph::~Paragraph()
1160 {
1161 deregisterWords();
1162 delete d;
1163 }
1164
1165
1166 void Paragraph::write(ostream & os, BufferParams const & bparams,
1167 depth_type & dth) const
1168 {
1169 // The beginning or end of a deeper (i.e. nested) area?
1170 if (dth != d->params_.depth()) {
1171 if (d->params_.depth() > dth) {
1172 while (d->params_.depth() > dth) {
1173 os << "\n\\begin_deeper";
1174 ++dth;
1175 }
1176 } else {
1177 while (d->params_.depth() < dth) {
1178 os << "\n\\end_deeper";
1179 --dth;
1180 }
1181 }
1182 }
1183
1184 // First write the layout
1185 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1186
1187 d->params_.write(os);
1188
1189 Font font1(inherit_font, bparams.language);
1190
1191 Change running_change = Change(Change::UNCHANGED);
1192
1193 int column = 0;
1194 for (pos_type i = 0; i <= size(); ++i) {
1195
1196 Change const change = lookupChange(i);
1197 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1198 running_change = change;
1199
1200 if (i == size())
1201 break;
1202
1203 // Write font changes (ignore spelling markers)
1204 Font font2 = getFontSettings(bparams, i);
1205 font2.setMisspelled(false);
1206 if (font2 != font1) {
1207 font2.lyxWriteChanges(font1, os);
1208 column = 0;
1209 font1 = font2;
1210 }
1211
1212 char_type const c = d->text_[i];
1213 switch (c) {
1214 case META_INSET:
1215 if (Inset const * inset = getInset(i)) {
1216 if (inset->directWrite()) {
1217 // international char, let it write
1218 // code directly so it's shorter in
1219 // the file
1220 inset->write(os);
1221 } else {
1222 if (i)
1223 os << '\n';
1224 os << "\\begin_inset ";
1225 inset->write(os);
1226 os << "\n\\end_inset\n\n";
1227 column = 0;
1228 }
1229 }
1230 break;
1231 case '\\':
1232 os << "\n\\backslash\n";
1233 column = 0;
1234 break;
1235 case '.':
1236 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1237 os << ".\n";
1238 column = 0;
1239 } else
1240 os << '.';
1241 break;
1242 default:
1243 if ((column > 70 && c == ' ')
1244 || column > 79) {
1245 os << '\n';
1246 column = 0;
1247 }
1248 // this check is to amend a bug. LyX sometimes
1249 // inserts '\0' this could cause problems.
1250 if (c != '\0')
1251 os << to_utf8(docstring(1, c));
1252 else
1253 LYXERR0("NUL char in structure.");
1254 ++column;
1255 break;
1256 }
1257 }
1258
1259 os << "\n\\end_layout\n";
1260 }
1261
1262
1263 void Paragraph::validate(LaTeXFeatures & features) const
1264 {
1265 d->validate(features, *d->layout_);
1266 }
1267
1268
1269 void Paragraph::insert(pos_type start, docstring const & str,
1270 Font const & font, Change const & change)
1271 {
1272 for (size_t i = 0, n = str.size(); i != n ; ++i)
1273 insertChar(start + i, str[i], font, change);
1274 }
1275
1276
1277 void Paragraph::appendChar(char_type c, Font const & font,
1278 Change const & change)
1279 {
1280 // track change
1281 d->changes_.insert(change, d->text_.size());
1282 // when appending characters, no need to update tables
1283 d->text_.push_back(c);
1284 setFont(d->text_.size() - 1, font);
1285 }
1286
1287
1288 void Paragraph::appendString(docstring const & s, Font const & font,
1289 Change const & change)
1290 {
1291 pos_type end = s.size();
1292 size_t oldsize = d->text_.size();
1293 size_t newsize = oldsize + end;
1294 size_t capacity = d->text_.capacity();
1295 if (newsize >= capacity)
1296 d->text_.reserve(max(capacity + 100, newsize));
1297
1298 // when appending characters, no need to update tables
1299 d->text_.append(s);
1300
1301 // FIXME: Optimize this!
1302 for (size_t i = oldsize; i != newsize; ++i) {
1303 // track change
1304 d->changes_.insert(change, i);
1305 }
1306 d->fontlist_.set(oldsize, font);
1307 d->fontlist_.set(newsize - 1, font);
1308 }
1309
1310
1311 void Paragraph::insertChar(pos_type pos, char_type c,
1312 bool trackChanges)
1313 {
1314 d->insertChar(pos, c, Change(trackChanges ?
1315 Change::INSERTED : Change::UNCHANGED));
1316 }
1317
1318
1319 void Paragraph::insertChar(pos_type pos, char_type c,
1320 Font const & font, bool trackChanges)
1321 {
1322 d->insertChar(pos, c, Change(trackChanges ?
1323 Change::INSERTED : Change::UNCHANGED));
1324 setFont(pos, font);
1325 }
1326
1327
1328 void Paragraph::insertChar(pos_type pos, char_type c,
1329 Font const & font, Change const & change)
1330 {
1331 d->insertChar(pos, c, change);
1332 setFont(pos, font);
1333 }
1334
1335
1336 bool Paragraph::insertInset(pos_type pos, Inset * inset,
1337 Font const & font, Change const & change)
1338 {
1339 bool const success = insertInset(pos, inset, change);
1340 // Set the font/language of the inset...
1341 setFont(pos, font);
1342 return success;
1343 }
1344
1345
1346 void Paragraph::resetFonts(Font const & font)
1347 {
1348 d->fontlist_.clear();
1349 d->fontlist_.set(0, font);
1350 d->fontlist_.set(d->text_.size() - 1, font);
1351 }
1352
1353 // Gets uninstantiated font setting at position.
1354 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1355 pos_type pos) const
1356 {
1357 if (pos > size()) {
1358 LYXERR0("pos: " << pos << " size: " << size());
1359 LASSERT(pos <= size(), /**/);
1360 }
1361
1362 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1363 if (cit != d->fontlist_.end())
1364 return cit->font();
1365
1366 if (pos == size() && !empty())
1367 return getFontSettings(bparams, pos - 1);
1368
1369 // Optimisation: avoid a full font instantiation if there is no
1370 // language change from previous call.
1371 static Font previous_font;
1372 static Language const * previous_lang = 0;
1373 Language const * lang = getParLanguage(bparams);
1374 if (lang != previous_lang) {
1375 previous_lang = lang;
1376 previous_font = Font(inherit_font, lang);
1377 }
1378 return previous_font;
1379 }
1380
1381
1382 FontSpan Paragraph::fontSpan(pos_type pos) const
1383 {
1384 LASSERT(pos <= size(), /**/);
1385 pos_type start = 0;
1386
1387 FontList::const_iterator cit = d->fontlist_.begin();
1388 FontList::const_iterator end = d->fontlist_.end();
1389 for (; cit != end; ++cit) {
1390 if (cit->pos() >= pos) {
1391 if (pos >= beginOfBody())
1392 return FontSpan(max(start, beginOfBody()),
1393 cit->pos());
1394 else
1395 return FontSpan(start,
1396 min(beginOfBody() - 1,
1397 cit->pos()));
1398 }
1399 start = cit->pos() + 1;
1400 }
1401
1402 // This should not happen, but if so, we take no chances.
1403 // LYXERR0("Paragraph::getEndPosOfFontSpan: This should not happen!");
1404 return FontSpan(pos, pos);
1405 }
1406
1407
1408 // Gets uninstantiated font setting at position 0
1409 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1410 {
1411 if (!empty() && !d->fontlist_.empty())
1412 return d->fontlist_.begin()->font();
1413
1414 // Optimisation: avoid a full font instantiation if there is no
1415 // language change from previous call.
1416 static Font previous_font;
1417 static Language const * previous_lang = 0;
1418 if (bparams.language != previous_lang) {
1419 previous_lang = bparams.language;
1420 previous_font = Font(inherit_font, bparams.language);
1421 }
1422
1423 return previous_font;
1424 }
1425
1426
1427 // Gets the fully instantiated font at a given position in a paragraph
1428 // This is basically the same function as Text::GetFont() in text2.cpp.
1429 // The difference is that this one is used for generating the LaTeX file,
1430 // and thus cosmetic "improvements" are disallowed: This has to deliver
1431 // the true picture of the buffer. (Asger)
1432 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1433 Font const & outerfont) const
1434 {
1435 LASSERT(pos >= 0, /**/);
1436
1437 Font font = getFontSettings(bparams, pos);
1438
1439 pos_type const body_pos = beginOfBody();
1440 FontInfo & fi = font.fontInfo();
1441 if (pos < body_pos)
1442 fi.realize(d->layout_->labelfont);
1443 else
1444 fi.realize(d->layout_->font);
1445
1446 fi.realize(outerfont.fontInfo());
1447 fi.realize(bparams.getFont().fontInfo());
1448
1449 return font;
1450 }
1451
1452
1453 Font const Paragraph::getLabelFont
1454 (BufferParams const & bparams, Font const & outerfont) const
1455 {
1456 FontInfo tmpfont = d->layout_->labelfont;
1457 tmpfont.realize(outerfont.fontInfo());
1458 tmpfont.realize(bparams.getFont().fontInfo());
1459 return Font(tmpfont, getParLanguage(bparams));
1460 }
1461
1462
1463 Font const Paragraph::getLayoutFont
1464 (BufferParams const & bparams, Font const & outerfont) const
1465 {
1466 FontInfo tmpfont = d->layout_->font;
1467 tmpfont.realize(outerfont.fontInfo());
1468 tmpfont.realize(bparams.getFont().fontInfo());
1469 return Font(tmpfont, getParLanguage(bparams));
1470 }
1471
1472
1473 /// Returns the height of the highest font in range
1474 FontSize Paragraph::highestFontInRange
1475 (pos_type startpos, pos_type endpos, FontSize def_size) const
1476 {
1477 return d->fontlist_.highestInRange(startpos, endpos, def_size);
1478 }
1479
1480
1481 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1482 {
1483 char_type c = d->text_[pos];
1484 if (!lyxrc.rtl_support)
1485 return c;
1486
1487 char_type uc = c;
1488 switch (c) {
1489 case '(':
1490 uc = ')';
1491 break;
1492 case ')':
1493 uc = '(';
1494 break;
1495 case '[':
1496 uc = ']';
1497 break;
1498 case ']':
1499 uc = '[';
1500 break;
1501 case '{':
1502 uc = '}';
1503 break;
1504 case '}':
1505 uc = '{';
1506 break;
1507 case '<':
1508 uc = '>';
1509 break;
1510 case '>':
1511 uc = '<';
1512 break;
1513 }
1514 if (uc != c && getFontSettings(bparams, pos).isRightToLeft())
1515 return uc;
1516 return c;
1517 }
1518
1519
1520 void Paragraph::setFont(pos_type pos, Font const & font)
1521 {
1522 LASSERT(pos <= size(), /**/);
1523
1524 // First, reduce font against layout/label font
1525 // Update: The setCharFont() routine in text2.cpp already
1526 // reduces font, so we don't need to do that here. (Asger)
1527
1528 d->fontlist_.set(pos, font);
1529 }
1530
1531
1532 void Paragraph::makeSameLayout(Paragraph const & par)
1533 {
1534 d->layout_ = par.d->layout_;
1535 d->params_ = par.d->params_;
1536 }
1537
1538
1539 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1540 {
1541 if (isFreeSpacing())
1542 return false;
1543
1544 int pos = 0;
1545 int count = 0;
1546
1547 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1548 if (eraseChar(pos, trackChanges))
1549 ++count;
1550 else
1551 ++pos;
1552 }
1553
1554 return count > 0 || pos > 0;
1555 }
1556
1557
1558 bool Paragraph::hasSameLayout(Paragraph const & par) const
1559 {
1560 return par.d->layout_ == d->layout_
1561 && d->params_.sameLayout(par.d->params_);
1562 }
1563
1564
1565 depth_type Paragraph::getDepth() const
1566 {
1567 return d->params_.depth();
1568 }
1569
1570
1571 depth_type Paragraph::getMaxDepthAfter() const
1572 {
1573 if (d->layout_->isEnvironment())
1574 return d->params_.depth() + 1;
1575 else
1576 return d->params_.depth();
1577 }
1578
1579
1580 char Paragraph::getAlign() const
1581 {
1582 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1583 return d->layout_->align;
1584 else
1585 return d->params_.align();
1586 }
1587
1588
1589 docstring const & Paragraph::labelString() const
1590 {
1591 return d->params_.labelString();
1592 }
1593
1594
1595 // the next two functions are for the manual labels
1596 docstring const Paragraph::getLabelWidthString() const
1597 {
1598 if (d->layout_->margintype == MARGIN_MANUAL
1599 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
1600 return d->params_.labelWidthString();
1601 else
1602 return _("Senseless with this layout!");
1603 }
1604
1605
1606 void Paragraph::setLabelWidthString(docstring const & s)
1607 {
1608 d->params_.labelWidthString(s);
1609 }
1610
1611
1612 docstring Paragraph::expandLabel(Layout const & layout,
1613 BufferParams const & bparams, bool process_appendix) const
1614 {
1615 DocumentClass const & tclass = bparams.documentClass();
1616 string const & lang = getParLanguage(bparams)->code();
1617 bool const in_appendix = process_appendix && d->params_.appendix();
1618 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
1619
1620 if (fmt.empty() && layout.labeltype == LABEL_COUNTER
1621 && !layout.counter.empty())
1622 return tclass.counters().theCounter(layout.counter, lang);
1623
1624 // handle 'inherited level parts' in 'fmt',
1625 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
1626 size_t const i = fmt.find('@', 0);
1627 if (i != docstring::npos) {
1628 size_t const j = fmt.find('@', i + 1);
1629 if (j != docstring::npos) {
1630 docstring parent(fmt, i + 1, j - i - 1);
1631 docstring label = from_ascii("??");
1632 if (tclass.hasLayout(parent))
1633 docstring label = expandLabel(tclass[parent], bparams,
1634 process_appendix);
1635 fmt = docstring(fmt, 0, i) + label
1636 + docstring(fmt, j + 1, docstring::npos);
1637 }
1638 }
1639
1640 return tclass.counters().counterLabel(fmt, lang);
1641 }
1642
1643
1644 void Paragraph::applyLayout(Layout const & new_layout)
1645 {
1646 d->layout_ = &new_layout;
1647 LyXAlignment const oldAlign = d->params_.align();
1648
1649 if (!(oldAlign & d->layout_->alignpossible)) {
1650 frontend::Alert::warning(_("Alignment not permitted"),
1651 _("The new layout does not permit the alignment previously used.\nSetting to default."));
1652 d->params_.align(LYX_ALIGN_LAYOUT);
1653 }
1654 }
1655
1656
1657 pos_type Paragraph::beginOfBody() const
1658 {
1659 return d->begin_of_body_;
1660 }
1661
1662
1663 void Paragraph::setBeginOfBody()
1664 {
1665 if (d->layout_->labeltype != LABEL_MANUAL) {
1666 d->begin_of_body_ = 0;
1667 return;
1668 }
1669
1670 // Unroll the first two cycles of the loop
1671 // and remember the previous character to
1672 // remove unnecessary getChar() calls
1673 pos_type i = 0;
1674 pos_type end = size();
1675 if (i < end && !isNewline(i)) {
1676 ++i;
1677 char_type previous_char = 0;
1678 char_type temp = 0;
1679 if (i < end) {
1680 previous_char = d->text_[i];
1681 if (!isNewline(i)) {
1682 ++i;
1683 while (i < end && previous_char != ' ') {
1684 temp = d->text_[i];
1685 if (isNewline(i))
1686 break;
1687 ++i;
1688 previous_char = temp;
1689 }
1690 }
1691 }
1692 }
1693
1694 d->begin_of_body_ = i;
1695 }
1696
1697
1698 bool Paragraph::forcePlainLayout() const
1699 {
1700 return inInset().forcePlainLayout();
1701 }
1702
1703
1704 bool Paragraph::allowParagraphCustomization() const
1705 {
1706 return inInset().allowParagraphCustomization();
1707 }
1708
1709
1710 bool Paragraph::usePlainLayout() const
1711 {
1712 return inInset().usePlainLayout();
1713 }
1714
1715
1716 namespace {
1717
1718 // paragraphs inside floats need different alignment tags to avoid
1719 // unwanted space
1720
1721 bool noTrivlistCentering(InsetCode code)
1722 {
1723 return code == FLOAT_CODE
1724 || code == WRAP_CODE
1725 || code == CELL_CODE;
1726 }
1727
1728
1729 string correction(string const & orig)
1730 {
1731 if (orig == "flushleft")
1732 return "raggedright";
1733 if (orig == "flushright")
1734 return "raggedleft";
1735 if (orig == "center")
1736 return "centering";
1737 return orig;
1738 }
1739
1740
1741 string const corrected_env(string const & suffix, string const & env,
1742 InsetCode code, bool const lastpar)
1743 {
1744 string output = suffix + "{";
1745 if (noTrivlistCentering(code)) {
1746 if (lastpar) {
1747 // the last paragraph in non-trivlist-aligned
1748 // context is special (to avoid unwanted whitespace)
1749 if (suffix == "\\begin")
1750 return "\\" + correction(env) + "{}";
1751 return string();
1752 }
1753 output += correction(env);
1754 } else
1755 output += env;
1756 output += "}";
1757 if (suffix == "\\begin")
1758 output += "\n";
1759 return output;
1760 }
1761
1762
1763 void adjust_row_column(string const & str, TexRow & texrow, int & column)
1764 {
1765 if (!contains(str, "\n"))
1766 column += str.size();
1767 else {
1768 string tmp;
1769 texrow.newline();
1770 column = rsplit(str, tmp, '\n').size();
1771 }
1772 }
1773
1774 } // namespace anon
1775
1776
1777 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
1778 odocstream & os, TexRow & texrow,
1779 OutputParams const & runparams) const
1780 {
1781 int column = 0;
1782
1783 if (params_.noindent()) {
1784 os << "\\noindent ";
1785 column += 10;
1786 }
1787
1788 LyXAlignment const curAlign = params_.align();
1789
1790 if (curAlign == layout_->align)
1791 return column;
1792
1793 switch (curAlign) {
1794 case LYX_ALIGN_NONE:
1795 case LYX_ALIGN_BLOCK:
1796 case LYX_ALIGN_LAYOUT:
1797 case LYX_ALIGN_SPECIAL:
1798 break;
1799 case LYX_ALIGN_LEFT:
1800 case LYX_ALIGN_RIGHT:
1801 case LYX_ALIGN_CENTER:
1802 if (runparams.moving_arg) {
1803 os << "\\protect";
1804 column += 8;
1805 }
1806 break;
1807 }
1808
1809 string const begin_tag = "\\begin";
1810 InsetCode code = owner_->ownerCode();
1811 bool const lastpar = runparams.isLastPar;
1812
1813 switch (curAlign) {
1814 case LYX_ALIGN_NONE:
1815 case LYX_ALIGN_BLOCK:
1816 case LYX_ALIGN_LAYOUT:
1817 case LYX_ALIGN_SPECIAL:
1818 break;
1819 case LYX_ALIGN_LEFT: {
1820 string output;
1821 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1822 output = corrected_env(begin_tag, "flushleft", code, lastpar);
1823 else
1824 output = corrected_env(begin_tag, "flushright", code, lastpar);
1825 os << from_ascii(output);
1826 adjust_row_column(output, texrow, column);
1827 break;
1828 } case LYX_ALIGN_RIGHT: {
1829 string output;
1830 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1831 output = corrected_env(begin_tag, "flushright", code, lastpar);
1832 else
1833 output = corrected_env(begin_tag, "flushleft", code, lastpar);
1834 os << from_ascii(output);
1835 adjust_row_column(output, texrow, column);
1836 break;
1837 } case LYX_ALIGN_CENTER: {
1838 string output;
1839 output = corrected_env(begin_tag, "center", code, lastpar);
1840 os << from_ascii(output);
1841 adjust_row_column(output, texrow, column);
1842 break;
1843 }
1844 }
1845
1846 return column;
1847 }
1848
1849
1850 int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
1851 odocstream & os, TexRow & texrow,
1852 OutputParams const & runparams) const
1853 {
1854 int column = 0;
1855
1856 LyXAlignment const curAlign = params_.align();
1857
1858 if (curAlign == layout_->align)
1859 return column;
1860
1861 switch (curAlign) {
1862 case LYX_ALIGN_NONE:
1863 case LYX_ALIGN_BLOCK:
1864 case LYX_ALIGN_LAYOUT:
1865 case LYX_ALIGN_SPECIAL:
1866 break;
1867 case LYX_ALIGN_LEFT:
1868 case LYX_ALIGN_RIGHT:
1869 case LYX_ALIGN_CENTER:
1870 if (runparams.moving_arg) {
1871 os << "\\protect";
1872 column = 8;
1873 }
1874 break;
1875 }
1876
1877 string const end_tag = "\n\\par\\end";
1878 InsetCode code = owner_->ownerCode();
1879 bool const lastpar = runparams.isLastPar;
1880
1881 switch (curAlign) {
1882 case LYX_ALIGN_NONE:
1883 case LYX_ALIGN_BLOCK:
1884 case LYX_ALIGN_LAYOUT:
1885 case LYX_ALIGN_SPECIAL:
1886 break;
1887 case LYX_ALIGN_LEFT: {
1888 string output;
1889 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1890 output = corrected_env(end_tag, "flushleft", code, lastpar);
1891 else
1892 output = corrected_env(end_tag, "flushright", code, lastpar);
1893 os << from_ascii(output);
1894 adjust_row_column(output, texrow, column);
1895 break;
1896 } case LYX_ALIGN_RIGHT: {
1897 string output;
1898 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1899 output = corrected_env(end_tag, "flushright", code, lastpar);
1900 else
1901 output = corrected_env(end_tag, "flushleft", code, lastpar);
1902 os << from_ascii(output);
1903 adjust_row_column(output, texrow, column);
1904 break;
1905 } case LYX_ALIGN_CENTER: {
1906 string output;
1907 output = corrected_env(end_tag, "center", code, lastpar);
1908 os << from_ascii(output);
1909 adjust_row_column(output, texrow, column);
1910 break;
1911 }
1912 }
1913
1914 return column;
1915 }
1916
1917
1918 // This one spits out the text of the paragraph
1919 bool Paragraph::latex(BufferParams const & bparams,
1920 Font const & outerfont,
1921 odocstream & os, TexRow & texrow,
1922 OutputParams const & runparams,
1923 int start_pos, int end_pos) const
1924 {
1925 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
1926
1927 bool return_value = false;
1928
1929 bool const allowcust = allowParagraphCustomization();
1930
1931 // FIXME This check should not be needed. Perhaps issue an
1932 // error if it triggers.
1933 Layout const & style = forcePlainLayout() ?
1934 bparams.documentClass().plainLayout() : *d->layout_;
1935
1936 // Current base font for all inherited font changes, without any
1937 // change caused by an individual character, except for the language:
1938 // It is set to the language of the first character.
1939 // As long as we are in the label, this font is the base font of the
1940 // label. Before the first body character it is set to the base font
1941 // of the body.
1942 Font basefont;
1943
1944 // Maybe we have to create a optional argument.
1945 pos_type body_pos = beginOfBody();
1946 unsigned int column = 0;
1947
1948 if (body_pos > 0) {
1949 os << '[';
1950 column += 1;
1951 basefont = getLabelFont(bparams, outerfont);
1952 } else {
1953 basefont = getLayoutFont(bparams, outerfont);
1954 }
1955
1956 // Which font is currently active?
1957 Font running_font(basefont);
1958 // Do we have an open font change?
1959 bool open_font = false;
1960
1961 Change runningChange = Change(Change::UNCHANGED);
1962
1963 Encoding const * const prev_encoding = runparams.encoding;
1964
1965 texrow.start(id(), 0);
1966
1967 // if the paragraph is empty, the loop will not be entered at all
1968 if (empty()) {
1969 if (style.isCommand()) {
1970 os << '{';
1971 ++column;
1972 }
1973 if (allowcust)
1974 column += d->startTeXParParams(bparams, os, texrow,
1975 runparams);
1976 }
1977
1978 for (pos_type i = 0; i < size(); ++i) {
1979 // First char in paragraph or after label?
1980 if (i == body_pos) {
1981 if (body_pos > 0) {
1982 if (open_font) {
1983 column += running_font.latexWriteEndChanges(
1984 os, bparams, runparams,
1985 basefont, basefont);
1986 open_font = false;
1987 }
1988 basefont = getLayoutFont(bparams, outerfont);
1989 running_font = basefont;
1990
1991 column += Changes::latexMarkChange(os, bparams,
1992 runningChange, Change(Change::UNCHANGED));
1993 runningChange = Change(Change::UNCHANGED);
1994
1995 os << "] ";
1996 column +=2;
1997 }
1998 if (style.isCommand()) {
1999 os << '{';
2000 ++column;
2001 }
2002
2003 if (allowcust)
2004 column += d->startTeXParParams(bparams, os,
2005 texrow,
2006 runparams);
2007 }
2008
2009 Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
2010 : lookupChange(i);
2011
2012 if (bparams.outputChanges && runningChange != change) {
2013 if (open_font) {
2014 column += running_font.latexWriteEndChanges(
2015 os, bparams, runparams, basefont, basefont);
2016 open_font = false;
2017 }
2018 basefont = getLayoutFont(bparams, outerfont);
2019 running_font = basefont;
2020
2021 column += Changes::latexMarkChange(os, bparams, runningChange, change);
2022 runningChange = change;
2023 }
2024
2025 // do not output text which is marked deleted
2026 // if change tracking output is disabled
2027 if (!bparams.outputChanges && change.deleted()) {
2028 continue;
2029 }
2030
2031 ++column;
2032
2033 // Fully instantiated font
2034 Font const font = getFont(bparams, i, outerfont);
2035
2036 Font const last_font = running_font;
2037
2038 // Do we need to close the previous font?
2039 if (open_font &&
2040 (font != running_font ||
2041 font.language() != running_font.language()))
2042 {
2043 column += running_font.latexWriteEndChanges(
2044 os, bparams, runparams, basefont,
2045 (i == body_pos-1) ? basefont : font);
2046 running_font = basefont;
2047 open_font = false;
2048 }
2049
2050 // close babel's font environment before opening CJK.
2051 if (!running_font.language()->babel().empty() &&
2052 font.language()->encoding()->package() == Encoding::CJK) {
2053 string end_tag = subst(lyxrc.language_command_end,
2054 "$$lang",
2055 running_font.language()->babel());
2056 os << from_ascii(end_tag);
2057 column += end_tag.length();
2058 }
2059
2060 // Switch file encoding if necessary (and allowed)
2061 if (!runparams.verbatim &&
2062 runparams.encoding->package() != Encoding::none &&
2063 font.language()->encoding()->package() != Encoding::none) {
2064 pair<bool, int> const enc_switch = switchEncoding(os, bparams,
2065 runparams, *(font.language()->encoding()));
2066 if (enc_switch.first) {
2067 column += enc_switch.second;
2068 runparams.encoding = font.language()->encoding();
2069 }
2070 }
2071
2072 char_type const c = d->text_[i];
2073
2074 // Do we need to change font?
2075 if ((font != running_font ||
2076 font.language() != running_font.language()) &&
2077 i != body_pos - 1)
2078 {
2079 odocstringstream ods;
2080 column += font.latexWriteStartChanges(ods, bparams,
2081 runparams, basefont,
2082 last_font);
2083 running_font = font;
2084 open_font = true;
2085 docstring fontchange = ods.str();
2086 // check whether the fontchange ends with a \\textcolor
2087 // modifier and the text starts with a space (bug 4473)
2088 docstring const last_modifier = rsplit(fontchange, '\\');
2089 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2090 os << fontchange << from_ascii("{}");
2091 // check if the fontchange ends with a trailing blank
2092 // (like "\small " (see bug 3382)
2093 else if (suffixIs(fontchange, ' ') && c == ' ')
2094 os << fontchange.substr(0, fontchange.size() - 1)
2095 << from_ascii("{}");
2096 else
2097 os << fontchange;
2098 }
2099
2100 // FIXME: think about end_pos implementation...
2101 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2102 // FIXME: integrate this case in latexSpecialChar
2103 // Do not print the separation of the optional argument
2104 // if style.pass_thru is false. This works because
2105 // latexSpecialChar ignores spaces if
2106 // style.pass_thru is false.
2107 if (i != body_pos - 1) {
2108 if (d->simpleTeXBlanks(
2109 runparams, os, texrow,
2110 i, column, font, style)) {
2111 // A surrogate pair was output. We
2112 // must not call latexSpecialChar
2113 // in this iteration, since it would output
2114 // the combining character again.
2115 ++i;
2116 continue;
2117 }
2118 }
2119 }
2120
2121 OutputParams rp = runparams;
2122 rp.free_spacing = style.free_spacing;
2123 rp.local_font = &font;
2124 rp.intitle = style.intitle;
2125
2126 // Two major modes: LaTeX or plain
2127 // Handle here those cases common to both modes
2128 // and then split to handle the two modes separately.
2129 if (c == META_INSET) {
2130 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2131 d->latexInset(bparams, os,
2132 texrow, rp, running_font,
2133 basefont, outerfont, open_font,
2134 runningChange, style, i, column);
2135 }
2136 } else {
2137 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2138 try {
2139 d->latexSpecialChar(os, rp, running_font, runningChange,
2140 style, i, column);
2141 } catch (EncodingException & e) {
2142 if (runparams.dryrun) {
2143 os << "<" << _("LyX Warning: ")
2144 << _("uncodable character") << " '";
2145 os.put(c);
2146 os << "'>";
2147 } else {
2148 // add location information and throw again.
2149 e.par_id = id();
2150 e.pos = i;
2151 throw(e);
2152 }
2153 }
2154 }
2155 }
2156
2157 // Set the encoding to that returned from latexSpecialChar (see
2158 // comment for encoding member in OutputParams.h)
2159 runparams.encoding = rp.encoding;
2160 }
2161
2162 // If we have an open font definition, we have to close it
2163 if (open_font) {
2164 #ifdef FIXED_LANGUAGE_END_DETECTION
2165 if (next_) {
2166 running_font
2167 .latexWriteEndChanges(os, bparams, runparams,
2168 basefont,
2169 next_->getFont(bparams, 0, outerfont));
2170 } else {
2171 running_font.latexWriteEndChanges(os, bparams,
2172 runparams, basefont, basefont);
2173 }
2174 #else
2175 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2176 //FIXME: there as we start another \selectlanguage with the next paragraph if
2177 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2178 running_font.latexWriteEndChanges(os, bparams, runparams,
2179 basefont, basefont);
2180 #endif
2181 }
2182
2183 column += Changes::latexMarkChange(os, bparams, runningChange, Change(Change::UNCHANGED));
2184
2185 // Needed if there is an optional argument but no contents.
2186 if (body_pos > 0 && body_pos == size()) {
2187 os << "]~";
2188 return_value = false;
2189 }
2190
2191 if (allowcust && d->endTeXParParams(bparams, os, texrow, runparams))
2192 runparams.encoding = prev_encoding;
2193
2194 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2195 return return_value;
2196 }
2197
2198
2199 bool Paragraph::emptyTag() const
2200 {
2201 for (pos_type i = 0; i < size(); ++i) {
2202 if (Inset const * inset = getInset(i)) {
2203 InsetCode lyx_code = inset->lyxCode();
2204 // FIXME testing like that is wrong. What is
2205 // the intent?
2206 if (lyx_code != TOC_CODE &&
2207 lyx_code != INCLUDE_CODE &&
2208 lyx_code != GRAPHICS_CODE &&
2209 lyx_code != ERT_CODE &&
2210 lyx_code != LISTINGS_CODE &&
2211 lyx_code != FLOAT_CODE &&
2212 lyx_code != TABULAR_CODE) {
2213 return false;
2214 }
2215 } else {
2216 char_type c = d->text_[i];
2217 if (c != ' ' && c != '\t')
2218 return false;
2219 }
2220 }
2221 return true;
2222 }
2223
2224
2225 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2226 const
2227 {
2228 for (pos_type i = 0; i < size(); ++i) {
2229 if (Inset const * inset = getInset(i)) {
2230 InsetCode lyx_code = inset->lyxCode();
2231 if (lyx_code == LABEL_CODE) {
2232 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2233 docstring const & id = il->getParam("name");
2234 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2235 }
2236 }
2237 }
2238 return string();
2239 }
2240
2241
2242 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2243 const
2244 {
2245 pos_type i;
2246 for (i = 0; i < size(); ++i) {
2247 if (Inset const * inset = getInset(i)) {
2248 inset->docbook(os, runparams);
2249 } else {
2250 char_type c = d->text_[i];
2251 if (c == ' ')
2252 break;
2253 os << sgml::escapeChar(c);
2254 }
2255 }
2256 return i;
2257 }
2258
2259
2260 pos_type Paragraph::firstWordLyXHTML(odocstream & os, OutputParams const & runparams)
2261 const
2262 {
2263 pos_type i;
2264 for (i = 0; i < size(); ++i) {
2265 if (Inset const * inset = getInset(i)) {
2266 inset->xhtml(os, runparams);
2267 } else {
2268 char_type c = d->text_[i];
2269 if (c == ' ')
2270 break;
2271 os << html::escapeChar(c);
2272 }
2273 }
2274 return i;
2275 }
2276
2277
2278 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2279 {
2280 Font font_old;
2281 pos_type size = text_.size();
2282 for (pos_type i = initial; i < size; ++i) {
2283 Font font = owner_->getFont(buf.params(), i, outerfont);
2284 if (text_[i] == META_INSET)
2285 return false;
2286 if (i != initial && font != font_old)
2287 return false;
2288 font_old = font;
2289 }
2290
2291 return true;
2292 }
2293
2294
2295 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2296 odocstream & os,
2297 OutputParams const & runparams,
2298 Font const & outerfont,
2299 pos_type initial) const
2300 {
2301 bool emph_flag = false;
2302
2303 Layout const & style = *d->layout_;
2304 FontInfo font_old =
2305 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2306
2307 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2308 os << "]]>";
2309
2310 // parsing main loop
2311 for (pos_type i = initial; i < size(); ++i) {
2312 Font font = getFont(buf.params(), i, outerfont);
2313
2314 // handle <emphasis> tag
2315 if (font_old.emph() != font.fontInfo().emph()) {
2316 if (font.fontInfo().emph() == FONT_ON) {
2317 os << "<emphasis>";
2318 emph_flag = true;
2319 } else if (i != initial) {
2320 os << "</emphasis>";
2321 emph_flag = false;
2322 }
2323 }
2324
2325 if (Inset const * inset = getInset(i)) {
2326 inset->docbook(os, runparams);
2327 } else {
2328 char_type c = d->text_[i];
2329
2330 if (style.pass_thru)
2331 os.put(c);
2332 else
2333 os << sgml::escapeChar(c);
2334 }
2335 font_old = font.fontInfo();
2336 }
2337
2338 if (emph_flag) {
2339 os << "</emphasis>";
2340 }
2341
2342 if (style.free_spacing)
2343 os << '\n';
2344 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2345 os << "<![CDATA[";
2346 }
2347
2348
2349 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2350 odocstream & os,
2351 OutputParams const & runparams,
2352 Font const & outerfont,
2353 pos_type initial) const
2354 {
2355 docstring retval;
2356
2357 // FIXME We really need to manage the tag nesting here.
2358 // Probably in the same sort of way as in output_xhtml.
2359 bool emph_flag = false;
2360 bool bold_flag = false;
2361 std::string closing_tag;
2362
2363 Layout const & style = *d->layout_;
2364 FontInfo font_old =
2365 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2366
2367 //if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2368 // os << "]]>";
2369
2370 // parsing main loop
2371 for (pos_type i = initial; i < size(); ++i) {
2372 Font font = getFont(buf.params(), i, outerfont);
2373
2374 // emphasis
2375 if (font_old.emph() != font.fontInfo().emph()) {
2376 if (font.fontInfo().emph() == FONT_ON) {
2377 os << "<em>";
2378 emph_flag = true;
2379 } else if (emph_flag && i != initial) {
2380 os << "</em>";
2381 emph_flag = false;
2382 }
2383 }
2384 // bold
2385 if (font_old.series() != font.fontInfo().series()) {
2386 if (font.fontInfo().series() == BOLD_SERIES) {
2387 os << "<strong>";
2388 bold_flag = true;
2389 } else if (bold_flag && i != initial) {
2390 os << "</strong>";
2391 bold_flag = false;
2392 }
2393 }
2394 // FIXME Other such tags?
2395
2396 if (Inset const * inset = getInset(i)) {
2397 retval += inset->xhtml(os, runparams);
2398 } else {
2399 char_type c = d->text_[i];
2400
2401 if (style.pass_thru)
2402 os.put(c);
2403 else if (c == '-') {
2404 docstring str;
2405 int j = i + 1;
2406 if (j < size() && d->text_[j] == '-') {
2407 j += 1;
2408 if (j < size() && d->text_[j] == '-') {
2409 str += from_ascii("&mdash;");
2410 i += 2;
2411 } else {
2412 str += from_ascii("&ndash;");
2413 i += 1;
2414 }
2415 }
2416 else
2417 str += c;
2418 os << str;
2419 } else
2420 os << html::escapeChar(c);
2421 }
2422 font_old = font.fontInfo();
2423 }
2424
2425 // FIXME This could be out of order. See above.
2426 if (emph_flag)
2427 os << "</em>";
2428 if (bold_flag)
2429 os << "</strong>";
2430
2431 return retval;
2432 }
2433
2434
2435 bool Paragraph::isHfill(pos_type pos) const
2436 {
2437 Inset const * inset = getInset(pos);
2438 return inset && (inset->lyxCode() == SPACE_CODE &&
2439 inset->isStretchableSpace());
2440 }
2441
2442
2443 bool Paragraph::isNewline(pos_type pos) const
2444 {
2445 Inset const * inset = getInset(pos);
2446 return inset && inset->lyxCode() == NEWLINE_CODE;
2447 }
2448
2449
2450 bool Paragraph::isLineSeparator(pos_type pos) const
2451 {
2452 char_type const c = d->text_[pos];
2453 if (isLineSeparatorChar(c))
2454 return true;
2455 Inset const * inset = getInset(pos);
2456 return inset && inset->isLineSeparator();
2457 }
2458
2459
2460 bool Paragraph::isWordSeparator(pos_type pos) const
2461 {
2462 if (Inset const * inset = getInset(pos))
2463 return !inset->isLetter();
2464 char_type const c = d->text_[pos];
2465 // We want to pass the ' and escape chars to the spellchecker
2466 static docstring const quote = from_utf8(lyxrc.spellchecker_esc_chars + '\'');
2467 return (!isLetterChar(c) && !isDigit(c) && !contains(quote, c))
2468 || pos == size();
2469 }
2470
2471
2472 bool Paragraph::isChar(pos_type pos) const
2473 {
2474 if (Inset const * inset = getInset(pos))
2475 return inset->isChar();
2476 char_type const c = d->text_[pos];
2477 return !isLetterChar(c) && !isDigit(c) && !lyx::isSpace(c);
2478 }
2479
2480
2481 bool Paragraph::isSpace(pos_type pos) const
2482 {
2483 if (Inset const * inset = getInset(pos))
2484 return inset->isSpace();
2485 char_type const c = d->text_[pos];
2486 return lyx::isSpace(c);
2487 }
2488
2489
2490 Language const *
2491 Paragraph::getParLanguage(BufferParams const & bparams) const
2492 {
2493 if (!empty())
2494 return getFirstFontSettings(bparams).language();
2495 // FIXME: we should check the prev par as well (Lgb)
2496 return bparams.language;
2497 }
2498
2499
2500 bool Paragraph::isRTL(BufferParams const & bparams) const
2501 {
2502 return lyxrc.rtl_support
2503 && getParLanguage(bparams)->rightToLeft()
2504 && !inInset().getLayout().forceLTR();
2505 }
2506
2507
2508 void Paragraph::changeLanguage(BufferParams const & bparams,
2509 Language const * from, Language const * to)
2510 {
2511 // change language including dummy font change at the end
2512 for (pos_type i = 0; i <= size(); ++i) {
2513 Font font = getFontSettings(bparams, i);
2514 if (font.language() == from) {
2515 font.setLanguage(to);
2516 setFont(i, font);
2517 }
2518 }
2519 }
2520
2521
2522 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
2523 {
2524 Language const * doc_language = bparams.language;
2525 FontList::const_iterator cit = d->fontlist_.begin();
2526 FontList::const_iterator end = d->fontlist_.end();
2527
2528 for (; cit != end; ++cit)
2529 if (cit->font().language() != ignore_language &&
2530 cit->font().language() != latex_language &&
2531 cit->font().language() != doc_language)
2532 return true;
2533 return false;
2534 }
2535
2536
2537 docstring Paragraph::asString(int options) const
2538 {
2539 return asString(0, size(), options);
2540 }
2541
2542
2543 docstring Paragraph::asString(pos_type beg, pos_type end, int options) const
2544 {
2545 odocstringstream os;
2546
2547 if (beg == 0
2548 && options & AS_STR_LABEL
2549 && !d->params_.labelString().empty())
2550 os << d->params_.labelString() << ' ';
2551
2552 for (pos_type i = beg; i < end; ++i) {
2553 char_type const c = d->text_[i];
2554 if (isPrintable(c) || c == '\t'
2555 || (c == '\n' && (options & AS_STR_NEWLINES)))
2556 os.put(c);
2557 else if (c == META_INSET && (options & AS_STR_INSETS)) {
2558 getInset(i)->tocString(os);
2559 if (getInset(i)->asInsetMath())
2560 os << " ";
2561 }
2562 }
2563
2564 return os.str();
2565 }
2566
2567
2568 docstring Paragraph::stringify(pos_type beg, pos_type end, int options, OutputParams & runparams) const
2569 {
2570 odocstringstream os;
2571
2572 if (beg == 0
2573 && options & AS_STR_LABEL
2574 && !d->params_.labelString().empty())
2575 os << d->params_.labelString() << ' ';
2576
2577 for (pos_type i = beg; i < end; ++i) {
2578 char_type const c = d->text_[i];
2579 if (isPrintable(c) || c == '\t'
2580 || (c == '\n' && (options & AS_STR_NEWLINES)))
2581 os.put(c);
2582 else if (c == META_INSET && (options & AS_STR_INSETS)) {
2583 getInset(i)->plaintext(os, runparams);
2584 }
2585 }
2586
2587 return os.str();
2588 }
2589
2590
2591 void Paragraph::setInsetOwner(Inset const * inset)
2592 {
2593 d->inset_owner_ = inset;
2594 }
2595
2596
2597 int Paragraph::id() const
2598 {
2599 return d->id_;
2600 }
2601
2602
2603 Layout const & Paragraph::layout() const
2604 {
2605 return *d->layout_;
2606 }
2607
2608
2609 void Paragraph::setLayout(Layout const & layout)
2610 {
2611 d->layout_ = &layout;
2612 }
2613
2614
2615 void Paragraph::setDefaultLayout(DocumentClass const & tc)
2616 {
2617 setLayout(tc.defaultLayout());
2618 }
2619
2620
2621 void Paragraph::setPlainLayout(DocumentClass const & tc)
2622 {
2623 setLayout(tc.plainLayout());
2624 }
2625
2626
2627 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
2628 {
2629 if (usePlainLayout())
2630 setPlainLayout(tclass);
2631 else
2632 setDefaultLayout(tclass);
2633 }
2634
2635
2636 Inset const & Paragraph::inInset() const
2637 {
2638 LASSERT(d->inset_owner_, throw ExceptionMessage(BufferException,
2639 _("Memory problem"), _("Paragraph not properly initialized")));
2640 return *d->inset_owner_;
2641 }
2642
2643
2644 InsetCode Paragraph::ownerCode() const
2645 {
2646 return d->inset_owner_ ? d->inset_owner_->lyxCode() : NO_CODE;
2647 }
2648
2649
2650 ParagraphParameters & Paragraph::params()
2651 {
2652 return d->params_;
2653 }
2654
2655
2656 ParagraphParameters const & Paragraph::params() const
2657 {
2658 return d->params_;
2659 }
2660
2661
2662 bool Paragraph::isFreeSpacing() const
2663 {
2664 if (d->layout_->free_spacing)
2665 return true;
2666 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
2667 }
2668
2669
2670 bool Paragraph::allowEmpty() const
2671 {
2672 if (d->layout_->keepempty)
2673 return true;
2674 return d->inset_owner_ && d->inset_owner_->allowEmpty();
2675 }
2676
2677
2678 char_type Paragraph::transformChar(char_type c, pos_type pos) const
2679 {
2680 if (!Encodings::isArabicChar(c))
2681 return c;
2682
2683 char_type prev_char = ' ';
2684 char_type next_char = ' ';
2685
2686 for (pos_type i = pos - 1; i >= 0; --i) {
2687 char_type const par_char = d->text_[i];
2688 if (!Encodings::isArabicComposeChar(par_char)) {
2689 prev_char = par_char;
2690 break;
2691 }
2692 }
2693
2694 for (pos_type i = pos + 1, end = size(); i < end; ++i) {
2695 char_type const par_char = d->text_[i];
2696 if (!Encodings::isArabicComposeChar(par_char)) {
2697 next_char = par_char;
2698 break;
2699 }
2700 }
2701
2702 if (Encodings::isArabicChar(next_char)) {
2703 if (Encodings::isArabicChar(prev_char) &&
2704 !Encodings::isArabicSpecialChar(prev_char))
2705 return Encodings::transformChar(c, Encodings::FORM_MEDIAL);
2706 else
2707 return Encodings::transformChar(c, Encodings::FORM_INITIAL);
2708 } else {
2709 if (Encodings::isArabicChar(prev_char) &&
2710 !Encodings::isArabicSpecialChar(prev_char))
2711 return Encodings::transformChar(c, Encodings::FORM_FINAL);
2712 else
2713 return Encodings::transformChar(c, Encodings::FORM_ISOLATED);
2714 }
2715 }
2716
2717
2718 int Paragraph::checkBiblio(Buffer const & buffer)
2719 {
2720 // FIXME From JS:
2721 // This is getting more and more a mess. ...We really should clean
2722 // up this bibitem issue for 1.6. See also bug 2743.
2723
2724 // Add bibitem insets if necessary
2725 if (d->layout_->labeltype != LABEL_BIBLIO)
2726 return 0;
2727
2728 bool hasbibitem = !d->insetlist_.empty()
2729 // Insist on it being in pos 0
2730 && d->text_[0] == META_INSET
2731 && d->insetlist_.begin()->inset->lyxCode() == BIBITEM_CODE;
2732
2733 bool track_changes = buffer.params().trackChanges;
2734
2735 docstring oldkey;
2736 docstring oldlabel;
2737
2738 // remove a bibitem in pos != 0
2739 // restore it later in pos 0 if necessary
2740 // (e.g. if a user inserts contents _before_ the item)
2741 // we're assuming there's only one of these, which there
2742 // should be.
2743 int erasedInsetPosition = -1;
2744 InsetList::iterator it = d->insetlist_.begin();
2745 InsetList::iterator end = d->insetlist_.end();
2746 for (; it != end; ++it)
2747 if (it->inset->lyxCode() == BIBITEM_CODE
2748 && it->pos > 0) {
2749 InsetBibitem * olditem = static_cast<InsetBibitem *>(it->inset);
2750 oldkey = olditem->getParam("key");
2751 oldlabel = olditem->getParam("label");
2752 erasedInsetPosition = it->pos;
2753 eraseChar(erasedInsetPosition, track_changes);
2754 break;
2755 }
2756
2757 // There was an InsetBibitem at the beginning, and we didn't
2758 // have to erase one.
2759 if (hasbibitem && erasedInsetPosition < 0)
2760 return 0;
2761
2762 // There was an InsetBibitem at the beginning and we did have to
2763 // erase one. So we give its properties to the beginning inset.
2764 if (hasbibitem) {
2765 InsetBibitem * inset =
2766 static_cast<InsetBibitem *>(d->insetlist_.begin()->inset);
2767 if (!oldkey.empty())
2768 inset->setParam("key", oldkey);
2769 inset->setParam("label", oldlabel);
2770 return -erasedInsetPosition;
2771 }
2772
2773 // There was no inset at the beginning, so we need to create one with
2774 // the key and label of the one we erased.
2775 InsetBibitem * inset =
2776 new InsetBibitem(buffer, InsetCommandParams(BIBITEM_CODE));
2777 // restore values of previously deleted item in this par.
2778 if (!oldkey.empty())
2779 inset->setParam("key", oldkey);
2780 inset->setParam("label", oldlabel);
2781 insertInset(0, static_cast<Inset *>(inset),
2782 Change(track_changes ? Change::INSERTED : Change::UNCHANGED));
2783
2784 return 1;
2785 }
2786
2787
2788 void Paragraph::checkAuthors(AuthorList const & authorList)
2789 {
2790 d->changes_.checkAuthors(authorList);
2791 }
2792
2793
2794 bool Paragraph::isChanged(pos_type pos) const
2795 {
2796 return lookupChange(pos).changed();
2797 }
2798
2799
2800 bool Paragraph::isInserted(pos_type pos) const
2801 {
2802 return lookupChange(pos).inserted();
2803 }
2804
2805
2806 bool Paragraph::isDeleted(pos_type pos) const
2807 {
2808 return lookupChange(pos).deleted();
2809 }
2810
2811
2812 InsetList const & Paragraph::insetList() const
2813 {
2814 return d->insetlist_;
2815 }
2816
2817
2818 void Paragraph::setBuffer(Buffer & b)
2819 {
2820 d->insetlist_.setBuffer(b);
2821 }
2822
2823
2824 Inset * Paragraph::releaseInset(pos_type pos)
2825 {
2826 Inset * inset = d->insetlist_.release(pos);
2827 /// does not honour change tracking!
2828 eraseChar(pos, false);
2829 return inset;
2830 }
2831
2832
2833 Inset * Paragraph::getInset(pos_type pos)
2834 {
2835 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
2836 ? d->insetlist_.get(pos) : 0;
2837 }
2838
2839
2840 Inset const * Paragraph::getInset(pos_type pos) const
2841 {
2842 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
2843 ? d->insetlist_.get(pos) : 0;
2844 }
2845
2846
2847 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
2848 pos_type & right, TextCase action)
2849 {
2850 // process sequences of modified characters; in change
2851 // tracking mode, this approach results in much better
2852 // usability than changing case on a char-by-char basis
2853 docstring changes;
2854
2855 bool const trackChanges = bparams.trackChanges;
2856
2857 bool capitalize = true;
2858
2859 for (; pos < right; ++pos) {
2860 char_type oldChar = d->text_[pos];
2861 char_type newChar = oldChar;
2862
2863 // ignore insets and don't play with deleted text!
2864 if (oldChar != META_INSET && !isDeleted(pos)) {
2865 switch (action) {
2866 case text_lowercase:
2867 newChar = lowercase(oldChar);
2868 break;
2869 case text_capitalization:
2870 if (capitalize) {
2871 newChar = uppercase(oldChar);
2872 capitalize = false;
2873 }
2874 break;
2875 case text_uppercase:
2876 newChar = uppercase(oldChar);
2877 break;
2878 }
2879 }
2880
2881 if (isWordSeparator(pos) || isDeleted(pos)) {
2882 // permit capitalization again
2883 capitalize = true;
2884 }
2885
2886 if (oldChar != newChar) {
2887 changes += newChar;
2888 if (pos != right - 1)
2889 continue;
2890 // step behind the changing area
2891 pos++;
2892 }
2893
2894 int erasePos = pos - changes.size();
2895 for (size_t i = 0; i < changes.size(); i++) {
2896 insertChar(pos, changes[i],
2897 getFontSettings(bparams,
2898 erasePos),
2899 trackChanges);
2900 if (!eraseChar(erasePos, trackChanges)) {
2901 ++erasePos;
2902 ++pos; // advance
2903 ++right; // expand selection
2904 }
2905 }
2906 changes.clear();
2907 }
2908 }
2909
2910
2911 bool Paragraph::find(docstring const & str, bool cs, bool mw,
2912 pos_type pos, bool del) const
2913 {
2914 int const strsize = str.length();
2915 int i = 0;
2916 pos_type const parsize = d->text_.size();
2917 for (i = 0; pos + i < parsize; ++i) {
2918 if (i >= strsize)
2919 break;
2920 if (cs && str[i] != d->text_[pos + i])
2921 break;
2922 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos + i]))
2923 break;
2924 if (!del && isDeleted(pos + i))
2925 break;
2926 }
2927
2928 if (i != strsize)
2929 return false;
2930
2931 // if necessary, check whether string matches word
2932 if (mw) {
2933 if (pos > 0 && !isWordSeparator(pos - 1))
2934 return false;
2935 if (pos + strsize < parsize
2936 && !isWordSeparator(pos + strsize))
2937 return false;
2938 }
2939
2940 return true;
2941 }
2942
2943
2944 char_type Paragraph::getChar(pos_type pos) const
2945 {
2946 return d->text_[pos];
2947 }
2948
2949
2950 pos_type Paragraph::size() const
2951 {
2952 return d->text_.size();
2953 }
2954
2955
2956 bool Paragraph::empty() const
2957 {
2958 return d->text_.empty();
2959 }
2960
2961
2962 bool Paragraph::isInset(pos_type pos) const
2963 {
2964 return d->text_[pos] == META_INSET;
2965 }
2966
2967
2968 bool Paragraph::isSeparator(pos_type pos) const
2969 {
2970 //FIXME: Are we sure this can be the only separator?
2971 return d->text_[pos] == ' ';
2972 }
2973
2974
2975 void Paragraph::deregisterWords()
2976 {
2977 Private::Words::const_iterator it;
2978 WordList & wl = theWordList();
2979 for (it = d->words_.begin(); it != d->words_.end(); ++it)
2980 wl.remove(*it);
2981 d->words_.clear();
2982 }
2983
2984
2985 void Paragraph::locateWord(pos_type & from, pos_type & to,
2986 word_location const loc) const
2987 {
2988 switch (loc) {
2989 case WHOLE_WORD_STRICT:
2990 if (from == 0 || from == size()
2991 || isWordSeparator(from)
2992 || isWordSeparator(from - 1)) {
2993 to = from;
2994 return;
2995 }
2996 // no break here, we go to the next
2997
2998 case WHOLE_WORD:
2999 // If we are already at the beginning of a word, do nothing
3000 if (!from || isWordSeparator(from - 1))
3001 break;
3002 // no break here, we go to the next
3003
3004 case PREVIOUS_WORD:
3005 // always move the cursor to the beginning of previous word
3006 while (from && !isWordSeparator(from - 1))
3007 --from;
3008 break;
3009 case NEXT_WORD:
3010 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3011 break;
3012 case PARTIAL_WORD:
3013 // no need to move the 'from' cursor
3014 break;
3015 }
3016 to = from;
3017 while (to < size() && !isWordSeparator(to))
3018 ++to;
3019 }
3020
3021
3022 void Paragraph::collectWords()
3023 {
3024 pos_type n = size();
3025 WordLangTuple wl;
3026 docstring_list suggestions;
3027 for (pos_type pos = 0; pos < n; ++pos) {
3028 if (isWordSeparator(pos))
3029 continue;
3030 pos_type from = pos;
3031 locateWord(from, pos, WHOLE_WORD);
3032 if (pos - from >= 6) {
3033 docstring word = asString(from, pos, AS_STR_NONE);
3034 d->words_.insert(word);
3035 }
3036 if (lyxrc.spellcheck_continuously
3037 && spellCheck(from, pos, wl, suggestions)) {
3038 for (size_t i = 0; i != suggestions.size(); ++i)
3039 d->words_.insert(suggestions[i]);
3040 }
3041 }
3042 }
3043
3044
3045 void Paragraph::registerWords()
3046 {
3047 Private::Words::const_iterator it;
3048 WordList & wl = theWordList();
3049 for (it = d->words_.begin(); it != d->words_.end(); ++it)
3050 wl.insert(*it);
3051 }
3052
3053
3054 void Paragraph::updateWords()
3055 {
3056 deregisterWords();
3057 collectWords();
3058 registerWords();
3059 }
3060
3061
3062 bool Paragraph::spellCheck(pos_type & from, pos_type & to, WordLangTuple & wl,
3063 docstring_list & suggestions, bool do_suggestion) const
3064 {
3065 SpellChecker * speller = theSpellChecker();
3066 if (!speller)
3067 return false;
3068
3069 locateWord(from, to, WHOLE_WORD);
3070 if (from == to || from >= pos_type(d->text_.size()))
3071 return false;
3072
3073 docstring word = asString(from, to, AS_STR_INSETS);
3074 string const lang_code = lyxrc.spellchecker_alt_lang.empty()
3075 ? getFontSettings(d->inset_owner_->buffer().params(), from).language()->code()
3076 : lyxrc.spellchecker_alt_lang;
3077 wl = WordLangTuple(word, lang_code);
3078 SpellChecker::Result res = speller->check(wl);
3079 // Just ignore any error that the spellchecker reports.
3080 // FIXME: we should through out an exception and catch it in the GUI to
3081 // display the error.
3082 if (!speller->error().empty())
3083 return false;
3084
3085 bool const misspelled = res != SpellChecker::OK
3086 && res != SpellChecker::IGNORED_WORD;
3087
3088 if (lyxrc.spellcheck_continuously)
3089 d->fontlist_.setMisspelled(from, to, misspelled);
3090
3091 if (misspelled && do_suggestion)
3092 speller->suggest(wl, suggestions);
3093 else
3094 suggestions.clear();
3095
3096 return misspelled;
3097 }
3098
3099
3100 bool Paragraph::isMisspelled(pos_type pos) const
3101 {
3102 pos_type from = pos;
3103 pos_type to = pos;
3104 WordLangTuple wl;
3105 docstring_list suggestions;
3106 return spellCheck(from, to, wl, suggestions, false);
3107 }
3108
3109
3110 } // namespace lyx
LyX document processor (git-svn mirror) - obsolete