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