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