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