]> git.lyx.org Git - lyx.git/blob - src/Paragraph.cpp
* src/Buffer.cpp:
[lyx.git] / src / Paragraph.cpp
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 "Buffer.h"
23 #include "BufferParams.h"
24 #include "Changes.h"
25 #include "Counters.h"
26 #include "Encoding.h"
27 #include "InsetList.h"
28 #include "Language.h"
29 #include "LaTeXFeatures.h"
30 #include "Layout.h"
31 #include "Length.h"
32 #include "Font.h"
33 #include "FontList.h"
34 #include "LyXRC.h"
35 #include "OutputParams.h"
36 #include "output_latex.h"
37 #include "paragraph_funcs.h"
38 #include "ParagraphParameters.h"
39 #include "sgml.h"
40 #include "TextClass.h"
41 #include "TexRow.h"
42 #include "VSpace.h"
43
44 #include "frontends/alert.h"
45 #include "frontends/FontMetrics.h"
46
47 #include "insets/InsetBibitem.h"
48 #include "insets/InsetLabel.h"
49
50 #include "support/convert.h"
51 #include "support/debug.h"
52 #include "support/gettext.h"
53 #include "support/lstrings.h"
54 #include "support/Messages.h"
55 #include "support/textutils.h"
56 #include "support/unicode.h"
57
58 #include <sstream>
59 #include <vector>
60
61 using namespace std;
62 using namespace lyx::support;
63
64 namespace lyx {
65
66 namespace {
67 /// Inset identifier (above 0x10ffff, for ucs-4)
68 char_type const META_INSET = 0x200001;
69 };
70
71 /////////////////////////////////////////////////////////////////////
72 //
73 // Paragraph::Private
74 //
75 /////////////////////////////////////////////////////////////////////
76
77 class Paragraph::Private
78 {
79 public:
80         ///
81         Private(Paragraph * owner);
82         /// "Copy constructor"
83         Private(Private const &, Paragraph * owner);
84
85         ///
86         void insertChar(pos_type pos, char_type c, Change const & change);
87
88         /// Output the surrogate pair formed by \p c and \p next to \p os.
89         /// \return the number of characters written.
90         int latexSurrogatePair(odocstream & os, char_type c, char_type next,
91                                Encoding const &);
92
93         /// Output a space in appropriate formatting (or a surrogate pair
94         /// if the next character is a combining character).
95         /// \return whether a surrogate pair was output.
96         bool simpleTeXBlanks(OutputParams const &,
97                              odocstream &, TexRow & texrow,
98                              pos_type i,
99                              unsigned int & column,
100                              Font const & font,
101                              Layout const & style);
102
103         /// Output consecutive unicode chars, belonging to the same script as
104         /// specified by the latex macro \p ltx, to \p os starting from \p i.
105         /// \return the number of characters written.
106         int writeScriptChars(odocstream & os, docstring const & ltx,
107                            Change &, Encoding const &, pos_type & i);
108
109         /// This could go to ParagraphParameters if we want to.
110         int startTeXParParams(BufferParams const &, odocstream &, TexRow &,
111                               bool) const;
112
113         /// This could go to ParagraphParameters if we want to.
114         int endTeXParParams(BufferParams const &, odocstream &, TexRow &,
115                             bool) const;
116
117         ///
118         void latexInset(Buffer const &, BufferParams const &,
119                                    odocstream &,
120                                    TexRow & texrow, OutputParams &,
121                                    Font & running_font,
122                                    Font & basefont,
123                                    Font const & outerfont,
124                                    bool & open_font,
125                                    Change & running_change,
126                                    Layout const & style,
127                                    pos_type & i,
128                                    unsigned int & column);
129
130         ///
131         void latexSpecialChar(
132                                    odocstream & os,
133                                    OutputParams & runparams,
134                                    Font & running_font,
135                                    Change & running_change,
136                                    Layout const & style,
137                                    pos_type & i,
138                                    unsigned int & column);
139
140         ///
141         bool latexSpecialT1(
142                 char_type const c,
143                 odocstream & os,
144                 pos_type & i,
145                 unsigned int & column);
146         ///
147         bool latexSpecialTypewriter(
148                 char_type const c,
149                 odocstream & os,
150                 pos_type & i,
151                 unsigned int & column);
152         ///
153         bool latexSpecialPhrase(
154                 odocstream & os,
155                 pos_type & i,
156                 unsigned int & column,
157                 OutputParams & runparams);
158
159         ///
160         void validate(LaTeXFeatures & features,
161                       Layout const & layout) const;
162
163         /// Checks if the paragraph contains only text and no inset or font change.
164         bool onlyText(Buffer const & buf, Font const & outerfont,
165                       pos_type initial) const;
166
167         /// match a string against a particular point in the paragraph
168         bool isTextAt(string const & str, pos_type pos) const;
169         
170         /// Which Paragraph owns us?
171         Paragraph * owner_;
172
173         /// In which Inset?
174         Inset * inset_owner_;
175
176         ///
177         FontList fontlist_;
178
179         ///
180         unsigned int id_;
181         ///
182         static unsigned int paragraph_id;
183         ///
184         ParagraphParameters params_;
185
186         /// for recording and looking up changes
187         Changes changes_;
188
189         ///
190         InsetList insetlist_;
191
192         ///
193         LayoutPtr layout_;
194
195         /// end of label
196         pos_type begin_of_body_;
197
198         typedef docstring TextContainer;
199         ///
200         TextContainer text_;
201 };
202
203
204 // Initialization of the counter for the paragraph id's,
205 unsigned int Paragraph::Private::paragraph_id = 0;
206
207 namespace {
208
209 struct special_phrase {
210         string phrase;
211         docstring macro;
212         bool builtin;
213 };
214
215 special_phrase const special_phrases[] = {
216         { "LyX", from_ascii("\\LyX{}"), false },
217         { "TeX", from_ascii("\\TeX{}"), true },
218         { "LaTeX2e", from_ascii("\\LaTeXe{}"), true },
219         { "LaTeX", from_ascii("\\LaTeX{}"), true },
220 };
221
222 size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase);
223
224 } // namespace anon
225
226
227 Paragraph::Private::Private(Paragraph * owner)
228         : owner_(owner), inset_owner_(0), begin_of_body_(0)
229 {
230         id_ = paragraph_id++;
231         text_.reserve(100);
232 }
233
234
235 Paragraph::Private::Private(Private const & p, Paragraph * owner)
236         : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_), 
237           params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
238           layout_(p.layout_), begin_of_body_(p.begin_of_body_), text_(p.text_)
239 {
240         id_ = paragraph_id++;
241 }
242
243
244 bool Paragraph::isChanged(pos_type start, pos_type end) const
245 {
246         BOOST_ASSERT(start >= 0 && start <= size());
247         BOOST_ASSERT(end > start && end <= size() + 1);
248
249         return d->changes_.isChanged(start, end);
250 }
251
252
253 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const {
254         // keep the logic here in sync with the logic of eraseChars()
255
256         if (!trackChanges) {
257                 return true;
258         }
259
260         Change change = d->changes_.lookup(size());
261
262         return change.type == Change::INSERTED && change.author == 0;
263 }
264
265
266 void Paragraph::setChange(Change const & change)
267 {
268         // beware of the imaginary end-of-par character!
269         d->changes_.set(change, 0, size() + 1);
270
271         /*
272          * Propagate the change recursively - but not in case of DELETED!
273          *
274          * Imagine that your co-author makes changes in an existing inset. He
275          * sends your document to you and you come to the conclusion that the
276          * inset should go completely. If you erase it, LyX must not delete all
277          * text within the inset. Otherwise, the change tracked insertions of
278          * your co-author get lost and there is no way to restore them later.
279          *
280          * Conclusion: An inset's content should remain untouched if you delete it
281          */
282
283         if (change.type != Change::DELETED) {
284                 for (pos_type pos = 0; pos < size(); ++pos) {
285                         if (isInset(pos))
286                                 getInset(pos)->setChange(change);
287                 }
288         }
289 }
290
291
292 void Paragraph::setChange(pos_type pos, Change const & change)
293 {
294         BOOST_ASSERT(pos >= 0 && pos <= size());
295
296         d->changes_.set(change, pos);
297
298         // see comment in setChange(Change const &) above
299
300         if (change.type != Change::DELETED &&
301             pos < size() && isInset(pos)) {
302                 getInset(pos)->setChange(change);
303         }
304 }
305
306
307 Change const & Paragraph::lookupChange(pos_type pos) const
308 {
309         BOOST_ASSERT(pos >= 0 && pos <= size());
310
311         return d->changes_.lookup(pos);
312 }
313
314
315 void Paragraph::acceptChanges(BufferParams const & bparams, pos_type start,
316                 pos_type end)
317 {
318         BOOST_ASSERT(start >= 0 && start <= size());
319         BOOST_ASSERT(end > start && end <= size() + 1);
320
321         for (pos_type pos = start; pos < end; ++pos) {
322                 switch (lookupChange(pos).type) {
323                         case Change::UNCHANGED:
324                                 // accept changes in nested inset
325                                 if (pos < size() && isInset(pos))
326                                         getInset(pos)->acceptChanges(bparams);
327
328                                 break;
329
330                         case Change::INSERTED:
331                                 d->changes_.set(Change(Change::UNCHANGED), pos);
332                                 // also accept changes in nested inset
333                                 if (pos < size() && isInset(pos)) {
334                                         getInset(pos)->acceptChanges(bparams);
335                                 }
336                                 break;
337
338                         case Change::DELETED:
339                                 // Suppress access to non-existent
340                                 // "end-of-paragraph char"
341                                 if (pos < size()) {
342                                         eraseChar(pos, false);
343                                         --end;
344                                         --pos;
345                                 }
346                                 break;
347                 }
348
349         }
350 }
351
352
353 void Paragraph::rejectChanges(BufferParams const & bparams,
354                 pos_type start, pos_type end)
355 {
356         BOOST_ASSERT(start >= 0 && start <= size());
357         BOOST_ASSERT(end > start && end <= size() + 1);
358
359         for (pos_type pos = start; pos < end; ++pos) {
360                 switch (lookupChange(pos).type) {
361                         case Change::UNCHANGED:
362                                 // reject changes in nested inset
363                                 if (pos < size() && isInset(pos)) {
364                                         getInset(pos)->rejectChanges(bparams);
365                                 }
366                                 break;
367
368                         case Change::INSERTED:
369                                 // Suppress access to non-existent
370                                 // "end-of-paragraph char"
371                                 if (pos < size()) {
372                                         eraseChar(pos, false);
373                                         --end;
374                                         --pos;
375                                 }
376                                 break;
377
378                         case Change::DELETED:
379                                 d->changes_.set(Change(Change::UNCHANGED), pos);
380
381                                 // Do NOT reject changes within a deleted inset!
382                                 // There may be insertions of a co-author inside of it!
383
384                                 break;
385                 }
386         }
387 }
388
389
390 void Paragraph::Private::insertChar(pos_type pos, char_type c,
391                 Change const & change)
392 {
393         BOOST_ASSERT(pos >= 0 && pos <= int(text_.size()));
394
395         // track change
396         changes_.insert(change, pos);
397
398         // This is actually very common when parsing buffers (and
399         // maybe inserting ascii text)
400         if (pos == pos_type(text_.size())) {
401                 // when appending characters, no need to update tables
402                 text_.push_back(c);
403                 return;
404         }
405
406         text_.insert(text_.begin() + pos, c);
407
408         // Update the font table.
409         fontlist_.increasePosAfterPos(pos);
410
411         // Update the insets
412         insetlist_.increasePosAfterPos(pos);
413 }
414
415
416 void Paragraph::insertInset(pos_type pos, Inset * inset,
417                                    Change const & change)
418 {
419         BOOST_ASSERT(inset);
420         BOOST_ASSERT(pos >= 0 && pos <= size());
421
422         d->insertChar(pos, META_INSET, change);
423         BOOST_ASSERT(d->text_[pos] == META_INSET);
424
425         // Add a new entry in the insetlist_.
426         d->insetlist_.insert(inset, pos);
427 }
428
429
430 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
431 {
432         BOOST_ASSERT(pos >= 0 && pos <= size());
433
434         // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
435
436         if (trackChanges) {
437                 Change change = d->changes_.lookup(pos);
438
439                 // set the character to DELETED if
440                 //  a) it was previously unchanged or
441                 //  b) it was inserted by a co-author
442
443                 if (change.type == Change::UNCHANGED ||
444                     (change.type == Change::INSERTED && change.author != 0)) {
445                         setChange(pos, Change(Change::DELETED));
446                         return false;
447                 }
448
449                 if (change.type == Change::DELETED)
450                         return false;
451         }
452
453         // Don't physically access the imaginary end-of-paragraph character.
454         // eraseChar() can only mark it as DELETED. A physical deletion of
455         // end-of-par must be handled externally.
456         if (pos == size()) {
457                 return false;
458         }
459
460         // track change
461         d->changes_.erase(pos);
462
463         // if it is an inset, delete the inset entry
464         if (d->text_[pos] == META_INSET)
465                 d->insetlist_.erase(pos);
466
467         d->text_.erase(d->text_.begin() + pos);
468
469         // Update the fontlist_
470         d->fontlist_.erase(pos);
471
472         // Update the insetlist_
473         d->insetlist_.decreasePosAfterPos(pos);
474
475         return true;
476 }
477
478
479 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
480 {
481         BOOST_ASSERT(start >= 0 && start <= size());
482         BOOST_ASSERT(end >= start && end <= size() + 1);
483
484         pos_type i = start;
485         for (pos_type count = end - start; count; --count) {
486                 if (!eraseChar(i, trackChanges))
487                         ++i;
488         }
489         return end - i;
490 }
491
492
493 int Paragraph::Private::latexSurrogatePair(odocstream & os, char_type c,
494                 char_type next, Encoding const & encoding)
495 {
496         // Writing next here may circumvent a possible font change between
497         // c and next. Since next is only output if it forms a surrogate pair
498         // with c we can ignore this:
499         // A font change inside a surrogate pair does not make sense and is
500         // hopefully impossible to input.
501         // FIXME: change tracking
502         // Is this correct WRT change tracking?
503         docstring const latex1 = encoding.latexChar(next);
504         docstring const latex2 = encoding.latexChar(c);
505         os << latex1 << '{' << latex2 << '}';
506         return latex1.length() + latex2.length() + 2;
507 }
508
509
510 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
511                                        odocstream & os, TexRow & texrow,
512                                        pos_type i,
513                                        unsigned int & column,
514                                        Font const & font,
515                                        Layout const & style)
516 {
517         if (style.pass_thru || runparams.verbatim)
518                 return false;
519
520         if (i + 1 < int(text_.size())) {
521                 char_type next = text_[i + 1];
522                 if (Encodings::isCombiningChar(next)) {
523                         Encoding const & encoding = *(runparams.encoding);
524                         // This space has an accent, so we must always output it.
525                         column += latexSurrogatePair(os, ' ', next, encoding) - 1;
526                         return true;
527                 }
528         }
529
530         if (lyxrc.plaintext_linelen > 0
531             && column > lyxrc.plaintext_linelen
532             && i
533             && text_[i - 1] != ' '
534             && (i + 1 < int(text_.size()))
535             // same in FreeSpacing mode
536             && !owner_->isFreeSpacing()
537             // In typewriter mode, we want to avoid
538             // ! . ? : at the end of a line
539             && !(font.fontInfo().family() == TYPEWRITER_FAMILY
540                  && (text_[i - 1] == '.'
541                      || text_[i - 1] == '?'
542                      || text_[i - 1] == ':'
543                      || text_[i - 1] == '!'))) {
544                 os << '\n';
545                 texrow.newline();
546                 texrow.start(owner_->id(), i + 1);
547                 column = 0;
548         } else if (style.free_spacing) {
549                 os << '~';
550         } else {
551                 os << ' ';
552         }
553         return false;
554 }
555
556
557 int Paragraph::Private::writeScriptChars(odocstream & os,
558                                          docstring const & ltx,
559                                          Change & runningChange,
560                                          Encoding const & encoding,
561                                          pos_type & i)
562 {
563         // FIXME: modifying i here is not very nice...
564
565         // We only arrive here when a proper language for character text_[i] has
566         // not been specified (i.e., it could not be translated in the current
567         // latex encoding) and it belongs to a known script.
568         // Parameter ltx contains the latex translation of text_[i] as specified in
569         // the unicodesymbols file and is something like "\textXXX{<spec>}".
570         // The latex macro name "textXXX" specifies the script to which text_[i]
571         // belongs and we use it in order to check whether characters from the
572         // same script immediately follow, such that we can collect them in a
573         // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
574         // for the first char but only "<spec>" for all subsequent chars.
575         docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
576         docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
577         string script = to_ascii(ltx.substr(1, brace1 - 1));
578         int length = ltx.substr(0, brace2).length();
579         os << ltx.substr(0, brace2);
580         int size = text_.size();
581         while (i + 1 < size) {
582                 char_type const next = text_[i + 1];
583                 // Stop here if next character belongs to another script
584                 // or there is a change in change tracking status.
585                 if (!Encodings::isKnownScriptChar(next, script) ||
586                     runningChange != owner_->lookupChange(i + 1))
587                         break;
588                 Font prev_font;
589                 bool found = false;
590                 FontList::const_iterator cit = fontlist_.begin();
591                 FontList::const_iterator end = fontlist_.end();
592                 for (; cit != end; ++cit) {
593                         if (cit->pos() >= i && !found) {
594                                 prev_font = cit->font();
595                                 found = true;
596                         }
597                         if (cit->pos() >= i + 1)
598                                 break;
599                 }
600                 // Stop here if there is a font attribute or encoding change.
601                 if (found && cit != end && prev_font != cit->font())
602                         break;
603                 docstring const latex = encoding.latexChar(next);
604                 docstring::size_type const b1 =
605                                         latex.find_first_of(from_ascii("{"));
606                 docstring::size_type const b2 =
607                                         latex.find_last_of(from_ascii("}"));
608                 int const len = b2 - b1 - 1;
609                 os << latex.substr(b1 + 1, len);
610                 length += len;
611                 ++i;
612         }
613         os << '}';
614         ++length;
615         return length;
616 }
617
618
619 bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const
620 {
621         pos_type const len = str.length();
622
623         // is the paragraph large enough?
624         if (pos + len > int(text_.size()))
625                 return false;
626
627         // does the wanted text start at point?
628         for (string::size_type i = 0; i < str.length(); ++i) {
629                 // Caution: direct comparison of characters works only
630                 // because str is pure ASCII.
631                 if (str[i] != text_[pos + i])
632                         return false;
633         }
634
635         return fontlist_.hasChangeInRange(pos, len);
636 }
637
638
639 void Paragraph::Private::latexInset(Buffer const & buf,
640                                              BufferParams const & bparams,
641                                              odocstream & os,
642                                              TexRow & texrow,
643                                              OutputParams & runparams,
644                                              Font & running_font,
645                                              Font & basefont,
646                                              Font const & outerfont,
647                                              bool & open_font,
648                                              Change & running_change,
649                                              Layout const & style,
650                                              pos_type & i,
651                                              unsigned int & column)
652 {
653         Inset * inset = owner_->getInset(i);
654         BOOST_ASSERT(inset);
655
656         if (style.pass_thru) {
657                 inset->plaintext(buf, os, runparams);
658                 return;
659         }
660
661         // FIXME: move this to InsetNewline::latex
662         if (inset->lyxCode() == NEWLINE_CODE) {
663                 // newlines are handled differently here than
664                 // the default in simpleTeXSpecialChars().
665                 if (!style.newline_allowed) {
666                         os << '\n';
667                 } else {
668                         if (open_font) {
669                                 column += running_font.latexWriteEndChanges(
670                                         os, bparams, runparams,
671                                         basefont, basefont);
672                                 open_font = false;
673                         }
674
675                         if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
676                                 os << '~';
677
678                         basefont = owner_->getLayoutFont(bparams, outerfont);
679                         running_font = basefont;
680
681                         if (runparams.moving_arg)
682                                 os << "\\protect ";
683
684                 }
685                 texrow.newline();
686                 texrow.start(owner_->id(), i + 1);
687                 column = 0;
688         }
689
690         if (owner_->lookupChange(i).type == Change::DELETED) {
691                 if( ++runparams.inDeletedInset == 1)
692                         runparams.changeOfDeletedInset = owner_->lookupChange(i);
693         }
694
695         if (inset->canTrackChanges()) {
696                 column += Changes::latexMarkChange(os, bparams, running_change,
697                         Change(Change::UNCHANGED));
698                 running_change = Change(Change::UNCHANGED);
699         }
700
701         bool close = false;
702         odocstream::pos_type const len = os.tellp();
703
704         if (inset->forceLTR() 
705             && running_font.isRightToLeft()
706                 // ERT is an exception, it should be output with no decorations at all
707                 && inset->lyxCode() != ERT_CODE) {
708                 if (running_font.language()->lang() == "farsi")
709                         os << "\\beginL{}";
710                 else
711                         os << "\\L{";
712                 close = true;
713         }
714
715         // FIXME: Bug: we can have an empty font change here!
716         // if there has just been a font change, we are going to close it
717         // right now, which means stupid latex code like \textsf{}. AFAIK,
718         // this does not harm dvi output. A minor bug, thus (JMarc)
719
720         // Some insets cannot be inside a font change command.
721         // However, even such insets *can* be placed in \L or \R
722         // or their equivalents (for RTL language switches), so we don't
723         // close the language in those cases.
724         // ArabTeX, though, cannot handle this special behavior, it seems.
725         bool arabtex = basefont.language()->lang() == "arabic_arabtex"
726                 || running_font.language()->lang() == "arabic_arabtex";
727         if (open_font && inset->noFontChange()) {
728                 bool closeLanguage = arabtex
729                         || basefont.isRightToLeft() == running_font.isRightToLeft();
730                 unsigned int count = running_font.latexWriteEndChanges(os,
731                         bparams, runparams, basefont, basefont, closeLanguage);
732                 column += count;
733                 // if any font properties were closed, update the running_font, 
734                 // making sure, however, to leave the language as it was
735                 if (count > 0) {
736                         // FIXME: probably a better way to keep track of the old 
737                         // language, than copying the entire font?
738                         Font const copy_font(running_font);
739                         basefont = owner_->getLayoutFont(bparams, outerfont);
740                         running_font = basefont;
741                         if (!closeLanguage)
742                                 running_font.setLanguage(copy_font.language());
743                         // leave font open if language is still open
744                         open_font = (running_font.language() == basefont.language());
745                         if (closeLanguage)
746                                 runparams.local_font = &basefont;
747                 }
748         }
749
750         int tmp = inset->latex(buf, os, runparams);
751
752         if (close) {
753         if (running_font.language()->lang() == "farsi")
754                         os << "\\endL{}";
755                 else
756                         os << '}';
757         }
758
759         if (tmp) {
760                 for (int j = 0; j < tmp; ++j)
761                         texrow.newline();
762
763                 texrow.start(owner_->id(), i + 1);
764                 column = 0;
765         } else {
766                 column += os.tellp() - len;
767         }
768
769         if (owner_->lookupChange(i).type == Change::DELETED)
770                 --runparams.inDeletedInset;
771 }
772
773
774 void Paragraph::Private::latexSpecialChar(
775                                              odocstream & os,
776                                              OutputParams & runparams,
777                                              Font & running_font,
778                                              Change & running_change,
779                                              Layout const & style,
780                                              pos_type & i,
781                                              unsigned int & column)
782 {
783         char_type const c = text_[i];
784
785         if (style.pass_thru) {
786                 if (c != '\0')
787                         // FIXME UNICODE: This can fail if c cannot
788                         // be encoded in the current encoding.
789                         os.put(c);
790                 return;
791         }
792
793         if (runparams.verbatim) {
794                 os.put(c);
795                 return;
796         }
797
798         if (lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column))
799                 return;
800
801         if (running_font.fontInfo().family() == TYPEWRITER_FAMILY
802                 && latexSpecialTypewriter(c, os, i, column))
803                 return;
804
805         // Otherwise, we use what LaTeX provides us.
806         switch (c) {
807         case '\\':
808                 os << "\\textbackslash{}";
809                 column += 15;
810                 break;
811         case '<':
812                 os << "\\textless{}";
813                 column += 10;
814                 break;
815         case '>':
816                 os << "\\textgreater{}";
817                 column += 13;
818                 break;
819         case '|':
820                 os << "\\textbar{}";
821                 column += 9;
822                 break;
823         case '-':
824                 os << '-';
825                 break;
826         case '\"':
827                 os << "\\char`\\\"{}";
828                 column += 9;
829                 break;
830
831         case '$': case '&':
832         case '%': case '#': case '{':
833         case '}': case '_':
834                 os << '\\';
835                 os.put(c);
836                 column += 1;
837                 break;
838
839         case '~':
840                 os << "\\textasciitilde{}";
841                 column += 16;
842                 break;
843
844         case '^':
845                 os << "\\textasciicircum{}";
846                 column += 17;
847                 break;
848
849         case '*': case '[':
850                 // avoid being mistaken for optional arguments
851                 os << '{';
852                 os.put(c);
853                 os << '}';
854                 column += 2;
855                 break;
856
857         case ' ':
858                 // Blanks are printed before font switching.
859                 // Sure? I am not! (try nice-latex)
860                 // I am sure it's correct. LyX might be smarter
861                 // in the future, but for now, nothing wrong is
862                 // written. (Asger)
863                 break;
864
865         default:
866
867                 // LyX, LaTeX etc.
868                 if (latexSpecialPhrase(os, i, column, runparams))
869                         return;
870
871                 if (c == '\0')
872                         return;
873
874                 Encoding const & encoding = *(runparams.encoding);
875                 if (i + 1 < int(text_.size())) {
876                         char_type next = text_[i + 1];
877                         if (Encodings::isCombiningChar(next)) {
878                                 column += latexSurrogatePair(os, c, next, encoding) - 1;
879                                 ++i;
880                                 break;
881                         }
882                 }
883                 string script;
884                 docstring const latex = encoding.latexChar(c);
885                 if (Encodings::isKnownScriptChar(c, script)
886                     && prefixIs(latex, from_ascii("\\" + script)))
887                         column += writeScriptChars(os, latex,
888                                         running_change, encoding, i) - 1;
889                 else if (latex.length() > 1 && latex[latex.length() - 1] != '}') {
890                         // Prevent eating of a following
891                         // space or command corruption by
892                         // following characters
893                         column += latex.length() + 1;
894                         os << latex << "{}";
895                 } else {
896                         column += latex.length() - 1;
897                         os << latex;
898                 }
899                 break;
900         }
901 }
902
903
904 bool Paragraph::Private::latexSpecialT1(char_type const c, odocstream & os,
905         pos_type & i, unsigned int & column)
906 {
907         switch (c) {
908         case '>':
909         case '<':
910                 os.put(c);
911                 // In T1 encoding, these characters exist
912                 // but we should avoid ligatures
913                 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
914                         return true;
915                 os << "\\,{}";
916                 column += 3;
917                 // Alternative code:
918                 //os << "\\textcompwordmark{}";
919                 //column += 19;
920                 return true;
921         case '|':
922                 os.put(c);
923                 return true;
924         default:
925                 return false;
926         }
927 }
928
929
930 bool Paragraph::Private::latexSpecialTypewriter(char_type const c, odocstream & os,
931         pos_type & i, unsigned int & column)
932 {
933         switch (c) {
934         case '-':
935                 if (i + 1 < int(text_.size()) && text_[i + 1] == '-') {
936                         // "--" in Typewriter mode -> "-{}-"
937                         os << "-{}";
938                         column += 2;
939                 } else
940                         os << '-';
941                 return true;
942
943         // I assume this is hack treating typewriter as verbatim
944         // FIXME UNICODE: This can fail if c cannot be encoded
945         // in the current encoding.
946
947         case '\0':
948                 return true;
949
950         // Those characters are not directly supported.
951         case '\\':
952         case '\"':
953         case '$': case '&':
954         case '%': case '#': case '{':
955         case '}': case '_':
956         case '~':
957         case '^':
958         case '*': case '[':
959         case ' ':
960                 return false;
961
962         default:
963                 // With Typewriter font, these characters exist.
964                 os.put(c);
965                 return true;
966         }
967 }
968
969
970 bool Paragraph::Private::latexSpecialPhrase(odocstream & os, pos_type & i,
971         unsigned int & column, OutputParams & runparams)
972 {
973         // FIXME: if we have "LaTeX" with a font
974         // change in the middle (before the 'T', then
975         // the "TeX" part is still special cased.
976         // Really we should only operate this on
977         // "words" for some definition of word
978
979         for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
980                 if (!isTextAt(special_phrases[pnr].phrase, i))
981                         continue;
982                 if (runparams.moving_arg)
983                         os << "\\protect";
984                 os << special_phrases[pnr].macro;
985                 i += special_phrases[pnr].phrase.length() - 1;
986                 column += special_phrases[pnr].macro.length() - 1;
987                 return true;
988         }
989         return false;
990 }
991
992
993 void Paragraph::Private::validate(LaTeXFeatures & features,
994                                 Layout const & layout) const
995 {
996         // check the params.
997         if (!params_.spacing().isDefault())
998                 features.require("setspace");
999
1000         // then the layouts
1001         features.useLayout(layout.name());
1002
1003         // then the fonts
1004         fontlist_.validate(features);
1005
1006         // then the indentation
1007         if (!params_.leftIndent().zero())
1008                 features.require("ParagraphLeftIndent");
1009
1010         // then the insets
1011         InsetList::const_iterator icit = insetlist_.begin();
1012         InsetList::const_iterator iend = insetlist_.end();
1013         for (; icit != iend; ++icit) {
1014                 if (icit->inset) {
1015                         icit->inset->validate(features);
1016                         if (layout.needprotect &&
1017                             icit->inset->lyxCode() == FOOT_CODE)
1018                                 features.require("NeedLyXFootnoteCode");
1019                 }
1020         }
1021
1022         // then the contents
1023         for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1024                 for (size_t pnr = 0; pnr < phrases_nr; ++pnr) {
1025                         if (!special_phrases[pnr].builtin
1026                             && isTextAt(special_phrases[pnr].phrase, i)) {
1027                                 features.require(special_phrases[pnr].phrase);
1028                                 break;
1029                         }
1030                 }
1031                 Encodings::validate(text_[i], features);
1032         }
1033 }
1034
1035 /////////////////////////////////////////////////////////////////////
1036 //
1037 // Paragraph
1038 //
1039 /////////////////////////////////////////////////////////////////////
1040
1041 Paragraph::Paragraph()
1042         : d(new Paragraph::Private(this))
1043 {
1044         itemdepth = 0;
1045         d->params_.clear();
1046 }
1047
1048
1049 Paragraph::Paragraph(Paragraph const & par)
1050         : itemdepth(par.itemdepth),
1051         d(new Paragraph::Private(*par.d, this))
1052 {
1053 }
1054
1055
1056 Paragraph & Paragraph::operator=(Paragraph const & par)
1057 {
1058         // needed as we will destroy the private part before copying it
1059         if (&par != this) {
1060                 itemdepth = par.itemdepth;
1061
1062                 delete d;
1063                 d = new Private(*par.d, this);
1064         }
1065         return *this;
1066 }
1067
1068
1069 Paragraph::~Paragraph()
1070 {
1071         delete d;
1072 }
1073
1074
1075 void Paragraph::write(Buffer const & buf, ostream & os,
1076                           BufferParams const & bparams,
1077                           depth_type & dth) const
1078 {
1079         // The beginning or end of a deeper (i.e. nested) area?
1080         if (dth != params().depth()) {
1081                 if (params().depth() > dth) {
1082                         while (params().depth() > dth) {
1083                                 os << "\n\\begin_deeper";
1084                                 ++dth;
1085                         }
1086                 } else {
1087                         while (params().depth() < dth) {
1088                                 os << "\n\\end_deeper";
1089                                 --dth;
1090                         }
1091                 }
1092         }
1093
1094         // First write the layout
1095         os << "\n\\begin_layout " << to_utf8(layout()->name()) << '\n';
1096
1097         params().write(os);
1098
1099         Font font1(inherit_font, bparams.language);
1100
1101         Change running_change = Change(Change::UNCHANGED);
1102
1103         int column = 0;
1104         for (pos_type i = 0; i <= size(); ++i) {
1105
1106                 Change change = lookupChange(i);
1107                 Changes::lyxMarkChange(os, column, running_change, change);
1108                 running_change = change;
1109
1110                 if (i == size())
1111                         break;
1112
1113                 // Write font changes
1114                 Font font2 = getFontSettings(bparams, i);
1115                 if (font2 != font1) {
1116                         font2.lyxWriteChanges(font1, os);
1117                         column = 0;
1118                         font1 = font2;
1119                 }
1120
1121                 char_type const c = d->text_[i];
1122                 switch (c) {
1123                 case META_INSET:
1124                 {
1125                         Inset const * inset = getInset(i);
1126                         if (inset)
1127                                 if (inset->directWrite()) {
1128                                         // international char, let it write
1129                                         // code directly so it's shorter in
1130                                         // the file
1131                                         inset->write(buf, os);
1132                                 } else {
1133                                         if (i)
1134                                                 os << '\n';
1135                                         os << "\\begin_inset ";
1136                                         inset->write(buf, os);
1137                                         os << "\n\\end_inset\n\n";
1138                                         column = 0;
1139                                 }
1140                 }
1141                 break;
1142                 case '\\':
1143                         os << "\n\\backslash\n";
1144                         column = 0;
1145                         break;
1146                 case '.':
1147                         if (i + 1 < size() && d->text_[i + 1] == ' ') {
1148                                 os << ".\n";
1149                                 column = 0;
1150                         } else
1151                                 os << '.';
1152                         break;
1153                 default:
1154                         if ((column > 70 && c == ' ')
1155                             || column > 79) {
1156                                 os << '\n';
1157                                 column = 0;
1158                         }
1159                         // this check is to amend a bug. LyX sometimes
1160                         // inserts '\0' this could cause problems.
1161                         if (c != '\0') {
1162                                 vector<char> tmp = ucs4_to_utf8(c);
1163                                 tmp.push_back('\0');
1164                                 os << &tmp[0];
1165                         } else
1166                                 lyxerr << "ERROR (Paragraph::writeFile):"
1167                                         " NULL char in structure." << endl;
1168                         ++column;
1169                         break;
1170                 }
1171         }
1172
1173         os << "\n\\end_layout\n";
1174 }
1175
1176
1177 void Paragraph::validate(LaTeXFeatures & features) const
1178 {
1179         d->validate(features, *layout());
1180 }
1181
1182
1183 void Paragraph::insert(pos_type start, docstring const & str,
1184                        Font const & font, Change const & change)
1185 {
1186         for (size_t i = 0, n = str.size(); i != n ; ++i)
1187                 insertChar(start + i, str[i], font, change);
1188 }
1189
1190
1191 void Paragraph::appendChar(char_type c, Font const & font,
1192                 Change const & change)
1193 {
1194         // track change
1195         d->changes_.insert(change, d->text_.size());
1196         // when appending characters, no need to update tables
1197         d->text_.push_back(c);
1198         setFont(d->text_.size() - 1, font);
1199 }
1200
1201
1202 void Paragraph::appendString(docstring const & s, Font const & font,
1203                 Change const & change)
1204 {
1205         pos_type end = s.size();
1206         size_t oldsize = d->text_.size();
1207         size_t newsize = oldsize + end;
1208         size_t capacity = d->text_.capacity();
1209         if (newsize >= capacity)
1210                 d->text_.reserve(max(capacity + 100, newsize));
1211
1212         // when appending characters, no need to update tables
1213         d->text_.append(s);
1214
1215         // FIXME: Optimize this!
1216         for (pos_type i = 0; i != end; ++i) {
1217                 // track change
1218                 d->changes_.insert(change, i);
1219         }
1220         d->fontlist_.set(oldsize, font);
1221         d->fontlist_.set(newsize - 1, font);
1222 }
1223
1224
1225 void Paragraph::insertChar(pos_type pos, char_type c,
1226                            bool trackChanges)
1227 {
1228         d->insertChar(pos, c, Change(trackChanges ?
1229                            Change::INSERTED : Change::UNCHANGED));
1230 }
1231
1232
1233 void Paragraph::insertChar(pos_type pos, char_type c,
1234                            Font const & font, bool trackChanges)
1235 {
1236         d->insertChar(pos, c, Change(trackChanges ?
1237                            Change::INSERTED : Change::UNCHANGED));
1238         setFont(pos, font);
1239 }
1240
1241
1242 void Paragraph::insertChar(pos_type pos, char_type c,
1243                            Font const & font, Change const & change)
1244 {
1245         d->insertChar(pos, c, change);
1246         setFont(pos, font);
1247 }
1248
1249
1250 void Paragraph::insertInset(pos_type pos, Inset * inset,
1251                             Font const & font, Change const & change)
1252 {
1253         insertInset(pos, inset, change);
1254         // Set the font/language of the inset...
1255         setFont(pos, font);
1256 }
1257
1258
1259 bool Paragraph::insetAllowed(InsetCode code)
1260 {
1261         return !d->inset_owner_ || d->inset_owner_->insetAllowed(code);
1262 }
1263
1264
1265 void Paragraph::resetFonts(Font const & font)
1266 {
1267         d->fontlist_.clear();
1268         d->fontlist_.set(0, font);
1269         d->fontlist_.set(d->text_.size() - 1, font);
1270 }
1271
1272 // Gets uninstantiated font setting at position.
1273 Font const Paragraph::getFontSettings(BufferParams const & bparams,
1274                                          pos_type pos) const
1275 {
1276         if (pos > size()) {
1277                 lyxerr << " pos: " << pos << " size: " << size() << endl;
1278                 BOOST_ASSERT(pos <= size());
1279         }
1280
1281         FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1282         if (cit != d->fontlist_.end())
1283                 return cit->font();
1284
1285         if (pos == size() && !empty())
1286                 return getFontSettings(bparams, pos - 1);
1287
1288         return Font(inherit_font, getParLanguage(bparams));
1289 }
1290
1291
1292 FontSpan Paragraph::fontSpan(pos_type pos) const
1293 {
1294         BOOST_ASSERT(pos <= size());
1295         pos_type start = 0;
1296
1297         FontList::const_iterator cit = d->fontlist_.begin();
1298         FontList::const_iterator end = d->fontlist_.end();
1299         for (; cit != end; ++cit) {
1300                 if (cit->pos() >= pos) {
1301                         if (pos >= beginOfBody())
1302                                 return FontSpan(max(start, beginOfBody()),
1303                                                 cit->pos());
1304                         else
1305                                 return FontSpan(start,
1306                                                 min(beginOfBody() - 1,
1307                                                          cit->pos()));
1308                 }
1309                 start = cit->pos() + 1;
1310         }
1311
1312         // This should not happen, but if so, we take no chances.
1313         //lyxerr << "Paragraph::getEndPosOfFontSpan: This should not happen!"
1314         //      << endl;
1315         return FontSpan(pos, pos);
1316 }
1317
1318
1319 // Gets uninstantiated font setting at position 0
1320 Font const Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1321 {
1322         if (!empty() && !d->fontlist_.empty())
1323                 return d->fontlist_.begin()->font();
1324
1325         return Font(inherit_font, bparams.language);
1326 }
1327
1328
1329 // Gets the fully instantiated font at a given position in a paragraph
1330 // This is basically the same function as Text::GetFont() in text2.cpp.
1331 // The difference is that this one is used for generating the LaTeX file,
1332 // and thus cosmetic "improvements" are disallowed: This has to deliver
1333 // the true picture of the buffer. (Asger)
1334 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1335                                  Font const & outerfont) const
1336 {
1337         BOOST_ASSERT(pos >= 0);
1338
1339         Font font = getFontSettings(bparams, pos);
1340
1341         pos_type const body_pos = beginOfBody();
1342         if (pos < body_pos)
1343                 font.fontInfo().realize(d->layout_->labelfont);
1344         else
1345                 font.fontInfo().realize(d->layout_->font);
1346
1347         font.fontInfo().realize(outerfont.fontInfo());
1348         font.fontInfo().realize(bparams.getFont().fontInfo());
1349
1350         return font;
1351 }
1352
1353
1354 Font const Paragraph::getLabelFont
1355         (BufferParams const & bparams, Font const & outerfont) const
1356 {
1357         FontInfo tmpfont = layout()->labelfont;
1358         tmpfont.realize(outerfont.fontInfo());
1359         tmpfont.realize(bparams.getFont().fontInfo());
1360         return Font(tmpfont, getParLanguage(bparams));
1361 }
1362
1363
1364 Font const Paragraph::getLayoutFont
1365         (BufferParams const & bparams, Font const & outerfont) const
1366 {
1367         FontInfo tmpfont = layout()->font;
1368         tmpfont.realize(outerfont.fontInfo());
1369         tmpfont.realize(bparams.getFont().fontInfo());
1370         return Font(tmpfont, getParLanguage(bparams));
1371 }
1372
1373
1374 /// Returns the height of the highest font in range
1375 FontSize Paragraph::highestFontInRange
1376         (pos_type startpos, pos_type endpos, FontSize def_size) const
1377 {
1378         return d->fontlist_.highestInRange(startpos, endpos, def_size);
1379 }
1380
1381
1382 char_type
1383 Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1384 {
1385         char_type c = d->text_[pos];
1386         if (!lyxrc.rtl_support)
1387                 return c;
1388
1389         char_type uc = c;
1390         switch (c) {
1391         case '(':
1392                 uc = ')';
1393                 break;
1394         case ')':
1395                 uc = '(';
1396                 break;
1397         case '[':
1398                 uc = ']';
1399                 break;
1400         case ']':
1401                 uc = '[';
1402                 break;
1403         case '{':
1404                 uc = '}';
1405                 break;
1406         case '}':
1407                 uc = '{';
1408                 break;
1409         case '<':
1410                 uc = '>';
1411                 break;
1412         case '>':
1413                 uc = '<';
1414                 break;
1415         }
1416         if (uc != c && getFontSettings(bparams, pos).isRightToLeft())
1417                 return uc;
1418         else
1419                 return c;
1420 }
1421
1422
1423 void Paragraph::setFont(pos_type pos, Font const & font)
1424 {
1425         BOOST_ASSERT(pos <= size());
1426
1427         // First, reduce font against layout/label font
1428         // Update: The setCharFont() routine in text2.cpp already
1429         // reduces font, so we don't need to do that here. (Asger)
1430         
1431         d->fontlist_.set(pos, font);
1432 }
1433
1434
1435 void Paragraph::makeSameLayout(Paragraph const & par)
1436 {
1437         layout(par.layout());
1438         // move to pimpl?
1439         d->params_ = par.params();
1440 }
1441
1442
1443 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1444 {
1445         if (isFreeSpacing())
1446                 return false;
1447
1448         int pos = 0;
1449         int count = 0;
1450
1451         while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1452                 if (eraseChar(pos, trackChanges))
1453                         ++count;
1454                 else
1455                         ++pos;
1456         }
1457
1458         return count > 0 || pos > 0;
1459 }
1460
1461
1462 bool Paragraph::hasSameLayout(Paragraph const & par) const
1463 {
1464         return par.layout() == layout() && d->params_.sameLayout(par.params());
1465 }
1466
1467
1468 depth_type Paragraph::getDepth() const
1469 {
1470         return params().depth();
1471 }
1472
1473
1474 depth_type Paragraph::getMaxDepthAfter() const
1475 {
1476         if (layout()->isEnvironment())
1477                 return params().depth() + 1;
1478         else
1479                 return params().depth();
1480 }
1481
1482
1483 char Paragraph::getAlign() const
1484 {
1485         if (params().align() == LYX_ALIGN_LAYOUT)
1486                 return layout()->align;
1487         else
1488                 return params().align();
1489 }
1490
1491
1492 docstring const & Paragraph::getLabelstring() const
1493 {
1494         return params().labelString();
1495 }
1496
1497
1498 // the next two functions are for the manual labels
1499 docstring const Paragraph::getLabelWidthString() const
1500 {
1501         if (layout()->margintype == MARGIN_MANUAL)
1502                 return params().labelWidthString();
1503         else
1504                 return _("Senseless with this layout!");
1505 }
1506
1507
1508 void Paragraph::setLabelWidthString(docstring const & s)
1509 {
1510         params().labelWidthString(s);
1511 }
1512
1513
1514 docstring const Paragraph::translateIfPossible(docstring const & s,
1515                 BufferParams const & bparams) const
1516 {
1517         if (!isAscii(s) || s.empty()) {
1518                 // This must be a user defined layout. We cannot translate
1519                 // this, since gettext accepts only ascii keys.
1520                 return s;
1521         }
1522         // Probably standard layout, try to translate
1523         Messages & m = getMessages(getParLanguage(bparams)->code());
1524         return m.get(to_ascii(s));
1525 }
1526
1527
1528 docstring Paragraph::expandLabel(LayoutPtr const & layout,
1529                 BufferParams const & bparams, bool process_appendix) const
1530 {
1531         TextClass const & tclass = bparams.getTextClass();
1532
1533         docstring fmt;
1534         if (process_appendix && params().appendix())
1535                 fmt = translateIfPossible(layout->labelstring_appendix(),
1536                         bparams);
1537         else
1538                 fmt = translateIfPossible(layout->labelstring(), bparams);
1539
1540         if (fmt.empty() && layout->labeltype == LABEL_COUNTER 
1541             && !layout->counter.empty())
1542                 fmt = "\\the" + layout->counter;
1543
1544         // handle 'inherited level parts' in 'fmt',
1545         // i.e. the stuff between '@' in   '@Section@.\arabic{subsection}'
1546         size_t const i = fmt.find('@', 0);
1547         if (i != docstring::npos) {
1548                 size_t const j = fmt.find('@', i + 1);
1549                 if (j != docstring::npos) {
1550                         docstring parent(fmt, i + 1, j - i - 1);
1551                         docstring label = from_ascii("??");
1552                         if (tclass.hasLayout(parent))
1553                                 docstring label = expandLabel(tclass[parent], bparams,
1554                                                       process_appendix);
1555                         fmt = docstring(fmt, 0, i) + label 
1556                                 + docstring(fmt, j + 1, docstring::npos);
1557                 }
1558         }
1559
1560         return tclass.counters().counterLabel(fmt);
1561 }
1562
1563
1564 void Paragraph::applyLayout(LayoutPtr const & new_layout)
1565 {
1566         layout(new_layout);
1567         LyXAlignment const oldAlign = params().align();
1568         
1569         if (!(oldAlign & layout()->alignpossible)) {
1570                 frontend::Alert::warning(_("Alignment not permitted"), 
1571                         _("The new layout does not permit the alignment previously used.\nSetting to default."));
1572                 params().align(LYX_ALIGN_LAYOUT);
1573         }
1574 }
1575
1576
1577 pos_type Paragraph::beginOfBody() const
1578 {
1579         return d->begin_of_body_;
1580 }
1581
1582
1583 void Paragraph::setBeginOfBody()
1584 {
1585         if (layout()->labeltype != LABEL_MANUAL) {
1586                 d->begin_of_body_ = 0;
1587                 return;
1588         }
1589
1590         // Unroll the first two cycles of the loop
1591         // and remember the previous character to
1592         // remove unnecessary getChar() calls
1593         pos_type i = 0;
1594         pos_type end = size();
1595         if (i < end && !isNewline(i)) {
1596                 ++i;
1597                 char_type previous_char = 0;
1598                 char_type temp = 0;
1599                 if (i < end) {
1600                         previous_char = d->text_[i];
1601                         if (!isNewline(i)) {
1602                                 ++i;
1603                                 while (i < end && previous_char != ' ') {
1604                                         temp = d->text_[i];
1605                                         if (isNewline(i))
1606                                                 break;
1607                                         ++i;
1608                                         previous_char = temp;
1609                                 }
1610                         }
1611                 }
1612         }
1613
1614         d->begin_of_body_ = i;
1615 }
1616
1617
1618 bool Paragraph::forceDefaultParagraphs() const
1619 {
1620         return inInset() && inInset()->forceDefaultParagraphs(0);
1621 }
1622
1623
1624 namespace {
1625
1626 // paragraphs inside floats need different alignment tags to avoid
1627 // unwanted space
1628
1629 bool noTrivlistCentering(InsetCode code)
1630 {
1631         return code == FLOAT_CODE || code == WRAP_CODE;
1632 }
1633
1634
1635 string correction(string const & orig)
1636 {
1637         if (orig == "flushleft")
1638                 return "raggedright";
1639         if (orig == "flushright")
1640                 return "raggedleft";
1641         if (orig == "center")
1642                 return "centering";
1643         return orig;
1644 }
1645
1646
1647 string const corrected_env(string const & suffix, string const & env,
1648         InsetCode code)
1649 {
1650         string output = suffix + "{";
1651         if (noTrivlistCentering(code))
1652                 output += correction(env);
1653         else
1654                 output += env;
1655         output += "}";
1656         if (suffix == "\\begin")
1657                 output += "\n";
1658         return output;
1659 }
1660
1661
1662 void adjust_row_column(string const & str, TexRow & texrow, int & column)
1663 {
1664         if (!contains(str, "\n"))
1665                 column += str.size();
1666         else {
1667                 string tmp;
1668                 texrow.newline();
1669                 column = rsplit(str, tmp, '\n').size();
1670         }
1671 }
1672
1673 } // namespace anon
1674
1675
1676 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
1677                                  odocstream & os, TexRow & texrow,
1678                                  bool moving_arg) const
1679 {
1680         int column = 0;
1681
1682         if (params_.noindent()) {
1683                 os << "\\noindent ";
1684                 column += 10;
1685         }
1686         
1687         LyXAlignment const curAlign = params_.align();
1688
1689         if (curAlign == layout_->align)
1690                 return column;
1691
1692         switch (curAlign) {
1693         case LYX_ALIGN_NONE:
1694         case LYX_ALIGN_BLOCK:
1695         case LYX_ALIGN_LAYOUT:
1696         case LYX_ALIGN_SPECIAL:
1697                 break;
1698         case LYX_ALIGN_LEFT:
1699         case LYX_ALIGN_RIGHT:
1700         case LYX_ALIGN_CENTER:
1701                 if (moving_arg) {
1702                         os << "\\protect";
1703                         column += 8;
1704                 }
1705                 break;
1706         }
1707
1708         switch (curAlign) {
1709         case LYX_ALIGN_NONE:
1710         case LYX_ALIGN_BLOCK:
1711         case LYX_ALIGN_LAYOUT:
1712         case LYX_ALIGN_SPECIAL:
1713                 break;
1714         case LYX_ALIGN_LEFT: {
1715                 string output;
1716                 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1717                         output = corrected_env("\\begin", "flushleft", owner_->ownerCode());
1718                 else
1719                         output = corrected_env("\\begin", "flushright", owner_->ownerCode());
1720                 os << from_ascii(output);
1721                 adjust_row_column(output, texrow, column);
1722                 break;
1723         } case LYX_ALIGN_RIGHT: {
1724                 string output;
1725                 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1726                         output = corrected_env("\\begin", "flushright", owner_->ownerCode());
1727                 else
1728                         output = corrected_env("\\begin", "flushleft", owner_->ownerCode());
1729                 os << from_ascii(output);
1730                 adjust_row_column(output, texrow, column);
1731                 break;
1732         } case LYX_ALIGN_CENTER: {
1733                 string output;
1734                 output = corrected_env("\\begin", "center", owner_->ownerCode());
1735                 os << from_ascii(output);
1736                 adjust_row_column(output, texrow, column);
1737                 break;
1738         }
1739         }
1740
1741         return column;
1742 }
1743
1744
1745 int Paragraph::Private::endTeXParParams(BufferParams const & bparams,
1746                                odocstream & os, TexRow & texrow,
1747                                bool moving_arg) const
1748 {
1749         int column = 0;
1750
1751         switch (params_.align()) {
1752         case LYX_ALIGN_NONE:
1753         case LYX_ALIGN_BLOCK:
1754         case LYX_ALIGN_LAYOUT:
1755         case LYX_ALIGN_SPECIAL:
1756                 break;
1757         case LYX_ALIGN_LEFT:
1758         case LYX_ALIGN_RIGHT:
1759         case LYX_ALIGN_CENTER:
1760                 if (moving_arg) {
1761                         os << "\\protect";
1762                         column = 8;
1763                 }
1764                 break;
1765         }
1766
1767         switch (params_.align()) {
1768         case LYX_ALIGN_NONE:
1769         case LYX_ALIGN_BLOCK:
1770         case LYX_ALIGN_LAYOUT:
1771         case LYX_ALIGN_SPECIAL:
1772                 break;
1773         case LYX_ALIGN_LEFT: {
1774                 string output;
1775                 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1776                         output = corrected_env("\n\\par\\end", "flushleft", owner_->ownerCode());
1777                 else
1778                         output = corrected_env("\n\\par\\end", "flushright", owner_->ownerCode());
1779                 os << from_ascii(output);
1780                 adjust_row_column(output, texrow, column);
1781                 break;
1782         } case LYX_ALIGN_RIGHT: {
1783                 string output;
1784                 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
1785                         output = corrected_env("\n\\par\\end", "flushright", owner_->ownerCode());
1786                 else
1787                         output = corrected_env("\n\\par\\end", "flushleft", owner_->ownerCode());
1788                 os << from_ascii(output);
1789                 adjust_row_column(output, texrow, column);
1790                 break;
1791         } case LYX_ALIGN_CENTER: {
1792                 string output;
1793                 output = corrected_env("\n\\par\\end", "center", owner_->ownerCode());
1794                 os << from_ascii(output);
1795                 adjust_row_column(output, texrow, column);
1796                 break;
1797         }
1798         }
1799
1800         return column;
1801 }
1802
1803
1804 // This one spits out the text of the paragraph
1805 bool Paragraph::latex(Buffer const & buf,
1806                                 BufferParams const & bparams,
1807                                 Font const & outerfont,
1808                                 odocstream & os, TexRow & texrow,
1809                                 OutputParams const & runparams) const
1810 {
1811         LYXERR(Debug::LATEX, "SimpleTeXOnePar...     " << this);
1812
1813         bool return_value = false;
1814
1815         LayoutPtr style;
1816
1817         // well we have to check if we are in an inset with unlimited
1818         // length (all in one row) if that is true then we don't allow
1819         // any special options in the paragraph and also we don't allow
1820         // any environment other than the default layout of the text class
1821         // to be valid!
1822         bool asdefault = forceDefaultParagraphs();
1823
1824         if (asdefault) {
1825                 style = bparams.getTextClass().defaultLayout();
1826         } else {
1827                 style = layout();
1828         }
1829
1830         // Current base font for all inherited font changes, without any
1831         // change caused by an individual character, except for the language:
1832         // It is set to the language of the first character.
1833         // As long as we are in the label, this font is the base font of the
1834         // label. Before the first body character it is set to the base font
1835         // of the body.
1836         Font basefont;
1837
1838         // Maybe we have to create a optional argument.
1839         pos_type body_pos = beginOfBody();
1840         unsigned int column = 0;
1841
1842         if (body_pos > 0) {
1843                 // the optional argument is kept in curly brackets in
1844                 // case it contains a ']'
1845                 os << "[{";
1846                 column += 2;
1847                 basefont = getLabelFont(bparams, outerfont);
1848         } else {
1849                 basefont = getLayoutFont(bparams, outerfont);
1850         }
1851
1852         // Which font is currently active?
1853         Font running_font(basefont);
1854         // Do we have an open font change?
1855         bool open_font = false;
1856
1857         Change runningChange = Change(Change::UNCHANGED);
1858
1859         texrow.start(id(), 0);
1860
1861         // if the paragraph is empty, the loop will not be entered at all
1862         if (empty()) {
1863                 if (style->isCommand()) {
1864                         os << '{';
1865                         ++column;
1866                 }
1867                 if (!asdefault)
1868                         column += d->startTeXParParams(bparams, os, texrow,
1869                                                     runparams.moving_arg);
1870         }
1871
1872         for (pos_type i = 0; i < size(); ++i) {
1873                 // First char in paragraph or after label?
1874                 if (i == body_pos) {
1875                         if (body_pos > 0) {
1876                                 if (open_font) {
1877                                         column += running_font.latexWriteEndChanges(
1878                                                 os, bparams, runparams,
1879                                                 basefont, basefont);
1880                                         open_font = false;
1881                                 }
1882                                 basefont = getLayoutFont(bparams, outerfont);
1883                                 running_font = basefont;
1884
1885                                 column += Changes::latexMarkChange(os, bparams,
1886                                                 runningChange, Change(Change::UNCHANGED));
1887                                 runningChange = Change(Change::UNCHANGED);
1888
1889                                 os << "}] ";
1890                                 column +=3;
1891                         }
1892                         if (style->isCommand()) {
1893                                 os << '{';
1894                                 ++column;
1895                         }
1896
1897                         if (!asdefault)
1898                                 column += d->startTeXParParams(bparams, os,
1899                                                             texrow,
1900                                                             runparams.moving_arg);
1901                 }
1902
1903                 Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset
1904                                                                  : lookupChange(i);
1905
1906                 if (bparams.outputChanges && runningChange != change) {
1907                         if (open_font) {
1908                                 column += running_font.latexWriteEndChanges(
1909                                                 os, bparams, runparams, basefont, basefont);
1910                                 open_font = false;
1911                         }
1912                         basefont = getLayoutFont(bparams, outerfont);
1913                         running_font = basefont;
1914
1915                         column += Changes::latexMarkChange(os, bparams, runningChange, change);
1916                         runningChange = change;
1917                 }
1918
1919                 // do not output text which is marked deleted
1920                 // if change tracking output is disabled
1921                 if (!bparams.outputChanges && change.type == Change::DELETED) {
1922                         continue;
1923                 }
1924
1925                 ++column;
1926
1927                 // Fully instantiated font
1928                 Font const font = getFont(bparams, i, outerfont);
1929
1930                 Font const last_font = running_font;
1931
1932                 // Do we need to close the previous font?
1933                 if (open_font &&
1934                     (font != running_font ||
1935                      font.language() != running_font.language()))
1936                 {
1937                         column += running_font.latexWriteEndChanges(
1938                                         os, bparams, runparams, basefont,
1939                                         (i == body_pos-1) ? basefont : font);
1940                         running_font = basefont;
1941                         open_font = false;
1942                 }
1943
1944                 // close babel's font environment before opening CJK.
1945                 if (!running_font.language()->babel().empty() &&
1946                     font.language()->encoding()->package() == Encoding::CJK) {
1947                                 string end_tag = subst(lyxrc.language_command_end,
1948                                                         "$$lang",
1949                                                         running_font.language()->babel());
1950                                 os << from_ascii(end_tag);
1951                                 column += end_tag.length();
1952                 }
1953
1954                 // Switch file encoding if necessary (and allowed)
1955                 if (!runparams.verbatim && 
1956                     runparams.encoding->package() == Encoding::none &&
1957                     font.language()->encoding()->package() == Encoding::none) {
1958                         pair<bool, int> const enc_switch = switchEncoding(os, bparams,
1959                                         runparams, *(font.language()->encoding()));
1960                         if (enc_switch.first) {
1961                                 column += enc_switch.second;
1962                                 runparams.encoding = font.language()->encoding();
1963                         }
1964                 }
1965
1966                 char_type const c = d->text_[i];
1967
1968                 // Do we need to change font?
1969                 if ((font != running_font ||
1970                      font.language() != running_font.language()) &&
1971                         i != body_pos - 1)
1972                 {
1973                         odocstringstream ods;
1974                         column += font.latexWriteStartChanges(ods, bparams,
1975                                                               runparams, basefont,
1976                                                               last_font);
1977                         running_font = font;
1978                         open_font = true;
1979                         docstring fontchange = ods.str();
1980                         // check if the fontchange ends with a trailing blank
1981                         // (like "\small " (see bug 3382)
1982                         if (suffixIs(fontchange, ' ') && c == ' ')
1983                                 os << fontchange.substr(0, fontchange.size() - 1) 
1984                                    << from_ascii("{}");
1985                         else
1986                                 os << fontchange;
1987                 }
1988
1989                 if (c == ' ') {
1990                         // FIXME: integrate this case in latexSpecialChar
1991                         // Do not print the separation of the optional argument
1992                         // if style->pass_thru is false. This works because
1993                         // latexSpecialChar ignores spaces if
1994                         // style->pass_thru is false.
1995                         if (i != body_pos - 1) {
1996                                 if (d->simpleTeXBlanks(
1997                                                 runparams, os, texrow,
1998                                                 i, column, font, *style)) {
1999                                         // A surrogate pair was output. We
2000                                         // must not call latexSpecialChar
2001                                         // in this iteration, since it would output
2002                                         // the combining character again.
2003                                         ++i;
2004                                         continue;
2005                                 }
2006                         }
2007                 }
2008
2009                 OutputParams rp = runparams;
2010                 rp.free_spacing = style->free_spacing;
2011                 rp.local_font = &font;
2012                 rp.intitle = style->intitle;
2013
2014                 // Two major modes:  LaTeX or plain
2015                 // Handle here those cases common to both modes
2016                 // and then split to handle the two modes separately.
2017                 if (c == META_INSET)
2018                         d->latexInset(buf, bparams, os,
2019                                         texrow, rp, running_font,
2020                                         basefont, outerfont, open_font,
2021                                         runningChange, *style, i, column);
2022                 else {
2023                         try {
2024                                 d->latexSpecialChar(os, rp, running_font, runningChange,
2025                                         *style, i, column);
2026                         } catch (EncodingException & e) {
2027                                 if (runparams.dryrun) {
2028                                         os << _("<LyX Warning: uncodable character>");
2029                                         os.put(c);
2030                                         os << _("</LyX Warning>");
2031                                 } else {
2032                                         // add location information and throw again.
2033                                         e.par_id = id();
2034                                         e.pos = i;
2035                                         throw(e);
2036                                 }
2037                         }
2038                 }
2039
2040                 // Set the encoding to that returned from simpleTeXSpecialChars (see
2041                 // comment for encoding member in OutputParams.h)
2042                 runparams.encoding = rp.encoding;
2043         }
2044
2045         // If we have an open font definition, we have to close it
2046         if (open_font) {
2047 #ifdef FIXED_LANGUAGE_END_DETECTION
2048                 if (next_) {
2049                         running_font
2050                                 .latexWriteEndChanges(os, bparams, runparams,
2051                                         basefont,
2052                                         next_->getFont(bparams, 0, outerfont));
2053                 } else {
2054                         running_font.latexWriteEndChanges(os, bparams,
2055                                         runparams, basefont, basefont);
2056                 }
2057 #else
2058 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2059 //FIXME: there as we start another \selectlanguage with the next paragraph if
2060 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2061                 running_font.latexWriteEndChanges(os, bparams, runparams,
2062                                 basefont, basefont);
2063 #endif
2064         }
2065
2066         column += Changes::latexMarkChange(os, bparams, runningChange, Change(Change::UNCHANGED));
2067
2068         // Needed if there is an optional argument but no contents.
2069         if (body_pos > 0 && body_pos == size()) {
2070                 os << "}]~";
2071                 return_value = false;
2072         }
2073
2074         if (!asdefault) {
2075                 column += d->endTeXParParams(bparams, os, texrow,
2076                                           runparams.moving_arg);
2077         }
2078
2079         LYXERR(Debug::LATEX, "SimpleTeXOnePar...done " << this);
2080         return return_value;
2081 }
2082
2083
2084 bool Paragraph::emptyTag() const
2085 {
2086         for (pos_type i = 0; i < size(); ++i) {
2087                 if (isInset(i)) {
2088                         Inset const * inset = getInset(i);
2089                         InsetCode lyx_code = inset->lyxCode();
2090                         if (lyx_code != TOC_CODE &&
2091                             lyx_code != INCLUDE_CODE &&
2092                             lyx_code != GRAPHICS_CODE &&
2093                             lyx_code != ERT_CODE &&
2094                             lyx_code != LISTINGS_CODE &&
2095                             lyx_code != FLOAT_CODE &&
2096                             lyx_code != TABULAR_CODE) {
2097                                 return false;
2098                         }
2099                 } else {
2100                         char_type c = d->text_[i];
2101                         if (c != ' ' && c != '\t')
2102                                 return false;
2103                 }
2104         }
2105         return true;
2106 }
2107
2108
2109 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams) const
2110 {
2111         for (pos_type i = 0; i < size(); ++i) {
2112                 if (isInset(i)) {
2113                         Inset const * inset = getInset(i);
2114                         InsetCode lyx_code = inset->lyxCode();
2115                         if (lyx_code == LABEL_CODE) {
2116                                 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2117                                 docstring const & id = il->getParam("name");
2118                                 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2119                         }
2120                 }
2121
2122         }
2123         return string();
2124 }
2125
2126
2127 pos_type Paragraph::getFirstWord(Buffer const & buf, odocstream & os, OutputParams const & runparams) const
2128 {
2129         pos_type i;
2130         for (i = 0; i < size(); ++i) {
2131                 if (isInset(i)) {
2132                         Inset const * inset = getInset(i);
2133                         inset->docbook(buf, os, runparams);
2134                 } else {
2135                         char_type c = d->text_[i];
2136                         if (c == ' ')
2137                                 break;
2138                         os << sgml::escapeChar(c);
2139                 }
2140         }
2141         return i;
2142 }
2143
2144
2145 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2146 {
2147         Font font_old;
2148         pos_type size = text_.size();
2149         for (pos_type i = initial; i < size; ++i) {
2150                 Font font = owner_->getFont(buf.params(), i, outerfont);
2151                 if (text_[i] == META_INSET)
2152                         return false;
2153                 if (i != initial && font != font_old)
2154                         return false;
2155                 font_old = font;
2156         }
2157
2158         return true;
2159 }
2160
2161
2162 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2163                                     odocstream & os,
2164                                     OutputParams const & runparams,
2165                                     Font const & outerfont,
2166                                     pos_type initial) const
2167 {
2168         bool emph_flag = false;
2169
2170         LayoutPtr const & style = layout();
2171         FontInfo font_old =
2172                 style->labeltype == LABEL_MANUAL ? style->labelfont : style->font;
2173
2174         if (style->pass_thru && !d->onlyText(buf, outerfont, initial))
2175                 os << "]]>";
2176
2177         // parsing main loop
2178         for (pos_type i = initial; i < size(); ++i) {
2179                 Font font = getFont(buf.params(), i, outerfont);
2180
2181                 // handle <emphasis> tag
2182                 if (font_old.emph() != font.fontInfo().emph()) {
2183                         if (font.fontInfo().emph() == FONT_ON) {
2184                                 os << "<emphasis>";
2185                                 emph_flag = true;
2186                         } else if (i != initial) {
2187                                 os << "</emphasis>";
2188                                 emph_flag = false;
2189                         }
2190                 }
2191
2192                 if (isInset(i)) {
2193                         Inset const * inset = getInset(i);
2194                         inset->docbook(buf, os, runparams);
2195                 } else {
2196                         char_type c = d->text_[i];
2197
2198                         if (style->pass_thru)
2199                                 os.put(c);
2200                         else
2201                                 os << sgml::escapeChar(c);
2202                 }
2203                 font_old = font.fontInfo();
2204         }
2205
2206         if (emph_flag) {
2207                 os << "</emphasis>";
2208         }
2209
2210         if (style->free_spacing)
2211                 os << '\n';
2212         if (style->pass_thru && !d->onlyText(buf, outerfont, initial))
2213                 os << "<![CDATA[";
2214 }
2215
2216
2217 bool Paragraph::isHfill(pos_type pos) const
2218 {
2219         return isInset(pos)
2220                 && getInset(pos)->lyxCode() == HFILL_CODE;
2221 }
2222
2223
2224 bool Paragraph::isNewline(pos_type pos) const
2225 {
2226         return isInset(pos)
2227                 && getInset(pos)->lyxCode() == NEWLINE_CODE;
2228 }
2229
2230
2231 bool Paragraph::isLineSeparator(pos_type pos) const
2232 {
2233         char_type const c = d->text_[pos];
2234         return isLineSeparatorChar(c)
2235                 || (c == META_INSET && getInset(pos) &&
2236                 getInset(pos)->isLineSeparator());
2237 }
2238
2239
2240 /// Used by the spellchecker
2241 bool Paragraph::isLetter(pos_type pos) const
2242 {
2243         if (isInset(pos))
2244                 return getInset(pos)->isLetter();
2245         else {
2246                 char_type const c = d->text_[pos];
2247                 return isLetterChar(c) || isDigit(c);
2248         }
2249 }
2250
2251
2252 Language const *
2253 Paragraph::getParLanguage(BufferParams const & bparams) const
2254 {
2255         if (!empty())
2256                 return getFirstFontSettings(bparams).language();
2257         // FIXME: we should check the prev par as well (Lgb)
2258         return bparams.language;
2259 }
2260
2261
2262 bool Paragraph::isRTL(BufferParams const & bparams) const
2263 {
2264         return lyxrc.rtl_support
2265                 && getParLanguage(bparams)->rightToLeft()
2266                 && ownerCode() != ERT_CODE
2267                 && ownerCode() != LISTINGS_CODE;
2268 }
2269
2270
2271 void Paragraph::changeLanguage(BufferParams const & bparams,
2272                                Language const * from, Language const * to)
2273 {
2274         // change language including dummy font change at the end
2275         for (pos_type i = 0; i <= size(); ++i) {
2276                 Font font = getFontSettings(bparams, i);
2277                 if (font.language() == from) {
2278                         font.setLanguage(to);
2279                         setFont(i, font);
2280                 }
2281         }
2282 }
2283
2284
2285 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
2286 {
2287         Language const * doc_language = bparams.language;
2288         FontList::const_iterator cit = d->fontlist_.begin();
2289         FontList::const_iterator end = d->fontlist_.end();
2290
2291         for (; cit != end; ++cit)
2292                 if (cit->font().language() != ignore_language &&
2293                     cit->font().language() != latex_language &&
2294                     cit->font().language() != doc_language)
2295                         return true;
2296         return false;
2297 }
2298
2299
2300 // Convert the paragraph to a string.
2301 // Used for building the table of contents
2302 docstring const Paragraph::asString(Buffer const & buffer, bool label) const
2303 {
2304         return asString(buffer, 0, size(), label);
2305 }
2306
2307
2308 docstring const Paragraph::asString(Buffer const & buffer,
2309                                  pos_type beg, pos_type end, bool label) const
2310 {
2311
2312         odocstringstream os;
2313
2314         if (beg == 0 && label && !params().labelString().empty())
2315                 os << params().labelString() << ' ';
2316
2317         for (pos_type i = beg; i < end; ++i) {
2318                 char_type const c = d->text_[i];
2319                 if (isPrintable(c))
2320                         os.put(c);
2321                 else if (c == META_INSET)
2322                         getInset(i)->textString(buffer, os);
2323         }
2324
2325         return os.str();
2326 }
2327
2328
2329 void Paragraph::setInsetOwner(Inset * inset)
2330 {
2331         d->inset_owner_ = inset;
2332 }
2333
2334
2335 int Paragraph::id() const
2336 {
2337         return d->id_;
2338 }
2339
2340
2341 LayoutPtr const & Paragraph::layout() const
2342 {
2343         return d->layout_;
2344 }
2345
2346
2347 void Paragraph::layout(LayoutPtr const & new_layout)
2348 {
2349         d->layout_ = new_layout;
2350 }
2351
2352
2353 Inset * Paragraph::inInset() const
2354 {
2355         return d->inset_owner_;
2356 }
2357
2358
2359 InsetCode Paragraph::ownerCode() const
2360 {
2361         return d->inset_owner_ ?
2362                 d->inset_owner_->lyxCode() : NO_CODE;
2363 }
2364
2365
2366 ParagraphParameters & Paragraph::params()
2367 {
2368         return d->params_;
2369 }
2370
2371
2372 ParagraphParameters const & Paragraph::params() const
2373 {
2374         return d->params_;
2375 }
2376
2377
2378 bool Paragraph::isFreeSpacing() const
2379 {
2380         if (layout()->free_spacing)
2381                 return true;
2382         return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
2383 }
2384
2385
2386 bool Paragraph::allowEmpty() const
2387 {
2388         if (layout()->keepempty)
2389                 return true;
2390         return d->inset_owner_ && d->inset_owner_->allowEmpty();
2391 }
2392
2393
2394 char_type Paragraph::transformChar(char_type c, pos_type pos) const
2395 {
2396         if (!Encodings::is_arabic(c))
2397                 return c;
2398
2399         char_type prev_char = ' ';
2400         char_type next_char = ' ';
2401
2402         for (pos_type i = pos - 1; i >= 0; --i) {
2403                 char_type const par_char = d->text_[i];
2404                 if (!Encodings::isComposeChar_arabic(par_char)) {
2405                         prev_char = par_char;
2406                         break;
2407                 }
2408         }
2409
2410         for (pos_type i = pos + 1, end = size(); i < end; ++i) {
2411                 char_type const par_char = d->text_[i];
2412                 if (!Encodings::isComposeChar_arabic(par_char)) {
2413                         next_char = par_char;
2414                         break;
2415                 }
2416         }
2417
2418         if (Encodings::is_arabic(next_char)) {
2419                 if (Encodings::is_arabic(prev_char) &&
2420                         !Encodings::is_arabic_special(prev_char))
2421                         return Encodings::transformChar(c, Encodings::FORM_MEDIAL);
2422                 else
2423                         return Encodings::transformChar(c, Encodings::FORM_INITIAL);
2424         } else {
2425                 if (Encodings::is_arabic(prev_char) &&
2426                         !Encodings::is_arabic_special(prev_char))
2427                         return Encodings::transformChar(c, Encodings::FORM_FINAL);
2428                 else
2429                         return Encodings::transformChar(c, Encodings::FORM_ISOLATED);
2430         }
2431 }
2432
2433
2434 int Paragraph::checkBiblio(bool track_changes)
2435 {
2436         //FIXME From JS:
2437         //This is getting more and more a mess. ...We really should clean
2438         //up this bibitem issue for 1.6. See also bug 2743.
2439
2440         // Add bibitem insets if necessary
2441         if (layout()->labeltype != LABEL_BIBLIO)
2442                 return 0;
2443
2444         bool hasbibitem = !d->insetlist_.empty()
2445                 // Insist on it being in pos 0
2446                 && d->text_[0] == META_INSET
2447                 && d->insetlist_.begin()->inset->lyxCode() == BIBITEM_CODE;
2448
2449         docstring oldkey;
2450         docstring oldlabel;
2451
2452         // remove a bibitem in pos != 0
2453         // restore it later in pos 0 if necessary
2454         // (e.g. if a user inserts contents _before_ the item)
2455         // we're assuming there's only one of these, which there
2456         // should be.
2457         int erasedInsetPosition = -1;
2458         InsetList::iterator it = d->insetlist_.begin();
2459         InsetList::iterator end = d->insetlist_.end();
2460         for (; it != end; ++it)
2461                 if (it->inset->lyxCode() == BIBITEM_CODE
2462                     && it->pos > 0) {
2463                         InsetBibitem * olditem = static_cast<InsetBibitem *>(it->inset);
2464                         oldkey = olditem->getParam("key");
2465                         oldlabel = olditem->getParam("label");
2466                         erasedInsetPosition = it->pos;
2467                         eraseChar(erasedInsetPosition, track_changes);
2468                         break;
2469         }
2470
2471         //There was an InsetBibitem at the beginning, and we didn't
2472         //have to erase one.
2473         if (hasbibitem && erasedInsetPosition < 0)
2474                         return 0;
2475
2476         //There was an InsetBibitem at the beginning and we did have to
2477         //erase one. So we give its properties to the beginning inset.
2478         if (hasbibitem) {
2479                 InsetBibitem * inset =
2480                         static_cast<InsetBibitem *>(d->insetlist_.begin()->inset);
2481                 if (!oldkey.empty())
2482                         inset->setParam("key", oldkey);
2483                 inset->setParam("label", oldlabel);
2484                 return -erasedInsetPosition;
2485         }
2486
2487         //There was no inset at the beginning, so we need to create one with
2488         //the key and label of the one we erased.
2489         InsetBibitem * inset(new InsetBibitem(InsetCommandParams(BIBITEM_CODE)));
2490         // restore values of previously deleted item in this par.
2491         if (!oldkey.empty())
2492                 inset->setParam("key", oldkey);
2493         inset->setParam("label", oldlabel);
2494         insertInset(0, static_cast<Inset *>(inset),
2495                     Change(track_changes ? Change::INSERTED : Change::UNCHANGED));
2496
2497         return 1;
2498 }
2499
2500
2501 void Paragraph::checkAuthors(AuthorList const & authorList)
2502 {
2503         d->changes_.checkAuthors(authorList);
2504 }
2505
2506
2507 bool Paragraph::isUnchanged(pos_type pos) const
2508 {
2509         return lookupChange(pos).type == Change::UNCHANGED;
2510 }
2511
2512
2513 bool Paragraph::isInserted(pos_type pos) const
2514 {
2515         return lookupChange(pos).type == Change::INSERTED;
2516 }
2517
2518
2519 bool Paragraph::isDeleted(pos_type pos) const
2520 {
2521         return lookupChange(pos).type == Change::DELETED;
2522 }
2523
2524
2525 InsetList const & Paragraph::insetList() const
2526 {
2527         return d->insetlist_;
2528 }
2529
2530
2531 Inset * Paragraph::releaseInset(pos_type pos)
2532 {
2533         Inset * inset = d->insetlist_.release(pos);
2534         /// does not honour change tracking!
2535         eraseChar(pos, false);
2536         return inset;
2537 }
2538
2539
2540 Inset * Paragraph::getInset(pos_type pos)
2541 {
2542         return d->insetlist_.get(pos);
2543 }
2544
2545
2546 Inset const * Paragraph::getInset(pos_type pos) const
2547 {
2548         return d->insetlist_.get(pos);
2549 }
2550
2551
2552 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
2553                 pos_type right, TextCase action)
2554 {
2555         // process sequences of modified characters; in change
2556         // tracking mode, this approach results in much better
2557         // usability than changing case on a char-by-char basis
2558         docstring changes;
2559
2560         bool const trackChanges = bparams.trackChanges;
2561
2562         bool capitalize = true;
2563
2564         for (; pos < right; ++pos) {
2565                 char_type oldChar = d->text_[pos];
2566                 char_type newChar = oldChar;
2567
2568                 // ignore insets and don't play with deleted text!
2569                 if (oldChar != META_INSET && !isDeleted(pos)) {
2570                         switch (action) {
2571                                 case text_lowercase:
2572                                         newChar = lowercase(oldChar);
2573                                         break;
2574                                 case text_capitalization:
2575                                         if (capitalize) {
2576                                                 newChar = uppercase(oldChar);
2577                                                 capitalize = false;
2578                                         }
2579                                         break;
2580                                 case text_uppercase:
2581                                         newChar = uppercase(oldChar);
2582                                         break;
2583                         }
2584                 }
2585
2586                 if (!isLetter(pos) || isDeleted(pos)) {
2587                         // permit capitalization again
2588                         capitalize = true;
2589                 }
2590
2591                 if (oldChar != newChar)
2592                         changes += newChar;
2593
2594                 if (oldChar == newChar || pos == right - 1) {
2595                         if (oldChar != newChar) {
2596                                 // step behind the changing area
2597                                 pos++;
2598                         }
2599                         int erasePos = pos - changes.size();
2600                         for (size_t i = 0; i < changes.size(); i++) {
2601                                 insertChar(pos, changes[i],
2602                                         getFontSettings(bparams,
2603                                         erasePos),
2604                                         trackChanges);
2605                                 if (!eraseChar(erasePos, trackChanges)) {
2606                                         ++erasePos;
2607                                         ++pos; // advance
2608                                         ++right; // expand selection
2609                                 }
2610                         }
2611                         changes.clear();
2612                 }
2613         }
2614 }
2615
2616
2617 bool Paragraph::find(docstring const & str, bool cs, bool mw,
2618                 pos_type pos, bool del) const
2619 {
2620         int const strsize = str.length();
2621         int i = 0;
2622         pos_type const parsize = d->text_.size();
2623         for (i = 0; pos + i < parsize; ++i) {
2624                 if (i >= strsize)
2625                         break;
2626                 if (cs && str[i] != d->text_[pos + i])
2627                         break;
2628                 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos + i]))
2629                         break;
2630                 if (!del && isDeleted(pos + i))
2631                         break;
2632         }
2633
2634         if (i != strsize)
2635                 return false;
2636
2637         // if necessary, check whether string matches word
2638         if (mw) {
2639                 if (pos > 0 && isLetter(pos - 1))
2640                         return false;
2641                 if (pos + strsize < parsize
2642                         && isLetter(pos + strsize))
2643                         return false;
2644         }
2645
2646         return true;
2647 }
2648
2649
2650 char_type Paragraph::getChar(pos_type pos) const
2651 {
2652         return d->text_[pos];
2653 }
2654
2655
2656 pos_type Paragraph::size() const
2657 {
2658         return d->text_.size();
2659 }
2660
2661
2662 bool Paragraph::empty() const
2663 {
2664         return d->text_.empty();
2665 }
2666
2667
2668 bool Paragraph::isInset(pos_type pos) const
2669 {
2670         return d->text_[pos] == META_INSET;
2671 }
2672
2673
2674 bool Paragraph::isSeparator(pos_type pos) const
2675 {
2676         //FIXME: Are we sure this can be the only separator?
2677         return d->text_[pos] == ' ';
2678 }
2679
2680
2681 } // namespace lyx