src/lyxfind.cpp

   1 /**
   2  * \file lyxfind.cpp
   3  * This file is part of LyX, the document processor.
   4  * License details can be found in the file COPYING.
   5  *
   6  * \author Lars Gullik Bjønnes
   7  * \author John Levon
   8  * \author Jürgen Vigna
   9  * \author Alfredo Braunstein
  10  * \author Tommaso Cucinotta
  11  * \author Kornel Benko
  12  *
  13  * Full author contact details are available in file CREDITS.
  14  */
  15
  16 #include <config.h>
  17
  18 #include "lyxfind.h"
  19
  20 #include "Buffer.h"
  21 #include "BufferList.h"
  22 #include "BufferParams.h"
  23 #include "BufferView.h"
  24 #include "Changes.h"
  25 #include "Cursor.h"
  26 #include "CutAndPaste.h"
  27 #include "FuncRequest.h"
  28 #include "LyX.h"
  29 #include "output_latex.h"
  30 #include "OutputParams.h"
  31 #include "Paragraph.h"
  32 #include "Text.h"
  33 #include "Encoding.h"
  34 #include "Language.h"
  35
  36 #include "frontends/Application.h"
  37 #include "frontends/alert.h"
  38
  39 #include "mathed/InsetMath.h"
  40 #include "mathed/InsetMathHull.h"
  41 #include "mathed/MathData.h"
  42 #include "mathed/MathStream.h"
  43 #include "mathed/MathSupport.h"
  44
  45 #include "support/debug.h"
  46 #include "support/docstream.h"
  47 #include "support/FileName.h"
  48 #include "support/gettext.h"
  49 #include "support/lassert.h"
  50 #include "support/lstrings.h"
  51 #include "support/textutils.h"
  52
  53 #include <unordered_map>
  54 #include <regex>
  55
  56 //#define ResultsDebug
  57 #define USE_QT_FOR_SEARCH
  58 #if defined(USE_QT_FOR_SEARCH)
  59         #include <QtCore>       // sets QT_VERSION
  60         #if (QT_VERSION >= 0x050000)
  61                 #include <QRegularExpression>
  62                 #define QTSEARCH 1
  63         #else
  64                 #define QTSEARCH 0
  65         #endif
  66 #else
  67         #define QTSEARCH 0
  68 #endif
  69
  70 using namespace std;
  71 using namespace lyx::support;
  72
  73 namespace lyx {
  74
  75 typedef unordered_map<string, string> AccentsMap;
  76 typedef unordered_map<string,string>::const_iterator AccentsIterator;
  77 static AccentsMap accents = unordered_map<string, string>();
  78
  79 // Helper class for deciding what should be ignored
  80 class IgnoreFormats {
  81  public:
  82         ///
  83         IgnoreFormats() = default;
  84         ///
  85         bool getFamily() const { return ignoreFamily_; }
  86         ///
  87         bool getSeries() const { return ignoreSeries_; }
  88         ///
  89         bool getShape() const { return ignoreShape_; }
  90         ///
  91         bool getUnderline() const { return ignoreUnderline_; }
  92         ///
  93         bool getMarkUp() const { return ignoreMarkUp_; }
  94         ///
  95         bool getStrikeOut() const { return ignoreStrikeOut_; }
  96         ///
  97         bool getSectioning() const { return ignoreSectioning_; }
  98         ///
  99         bool getFrontMatter() const { return ignoreFrontMatter_; }
 100         ///
 101         bool getColor() const { return ignoreColor_; }
 102         ///
 103         bool getLanguage() const { return ignoreLanguage_; }
 104         ///
 105         bool getDeleted() const { return ignoreDeleted_; }
 106         ///
 107         void setIgnoreDeleted(bool value);
 108         ///
 109         bool getNonContent() const { return searchNonContent_; }
 110         ///
 111         void setIgnoreFormat(string const & type, bool value, bool fromUser = true);
 112
 113 private:
 114         ///
 115         bool ignoreFamily_ = false;
 116         ///
 117         bool ignoreSeries_ = false;
 118         ///
 119         bool ignoreShape_ = false;
 120         ///
 121         bool ignoreUnderline_ = false;
 122         ///
 123         bool ignoreMarkUp_ = false;
 124         ///
 125         bool ignoreStrikeOut_ = false;
 126         ///
 127         bool ignoreSectioning_ = false;
 128         ///
 129         bool ignoreFrontMatter_ = false;
 130         ///
 131         bool ignoreColor_ = false;
 132         ///
 133         bool ignoreLanguage_ = false;
 134         bool userSelectedIgnoreLanguage_ = false;
 135         ///
 136         bool ignoreDeleted_ = true;
 137         ///
 138         bool searchNonContent_ = true;
 139 };
 140
 141 void IgnoreFormats::setIgnoreFormat(string const & type, bool value, bool fromUser)
 142 {
 143         if (type == "color") {
 144                 ignoreColor_ = value;
 145         }
 146         else if (type == "language") {
 147                 if (fromUser) {
 148                         userSelectedIgnoreLanguage_ = value;
 149                         ignoreLanguage_ = value;
 150                 }
 151                 else
 152                         ignoreLanguage_ = (value || userSelectedIgnoreLanguage_);
 153         }
 154         else if (type == "sectioning") {
 155                 ignoreSectioning_ = value;
 156                 ignoreFrontMatter_ = value;
 157         }
 158         else if (type == "font") {
 159                 ignoreSeries_ = value;
 160                 ignoreShape_ = value;
 161                 ignoreFamily_ = value;
 162         }
 163         else if (type == "series") {
 164                 ignoreSeries_ = value;
 165         }
 166         else if (type == "shape") {
 167                 ignoreShape_ = value;
 168         }
 169         else if (type == "family") {
 170                 ignoreFamily_ = value;
 171         }
 172         else if (type == "markup") {
 173                 ignoreMarkUp_ = value;
 174         }
 175         else if (type == "underline") {
 176                 ignoreUnderline_ = value;
 177         }
 178         else if (type == "strike") {
 179                 ignoreStrikeOut_ = value;
 180         }
 181         else if (type == "deleted") {
 182                 ignoreDeleted_ = value;
 183         }
 184         else if (type == "non-output-content") {
 185                 searchNonContent_ = !value;
 186         }
 187 }
 188
 189 // The global variable that can be changed from outside
 190 IgnoreFormats ignoreFormats;
 191
 192
 193 void setIgnoreFormat(string const & type, bool value, bool fromUser)
 194 {
 195         ignoreFormats.setIgnoreFormat(type, value, fromUser);
 196 }
 197
 198
 199 namespace {
 200
 201 bool parse_bool(docstring & howto, bool const defvalue = false)
 202 {
 203         if (howto.empty())
 204                 return defvalue;
 205         docstring var;
 206         howto = split(howto, var, ' ');
 207         return var == "1";
 208 }
 209
 210
 211 class MatchString
 212 {
 213 public:
 214         MatchString(docstring const & s, bool cs, bool mw)
 215                 : str(s), case_sens(cs), whole_words(mw)
 216         {}
 217
 218         // returns true if the specified string is at the specified position
 219         // del specifies whether deleted strings in ct mode will be considered
 220         int operator()(Paragraph const & par, pos_type pos, bool del = true) const
 221         {
 222                 return par.find(str, case_sens, whole_words, pos, del);
 223         }
 224
 225 private:
 226         // search string
 227         docstring str;
 228         // case sensitive
 229         bool case_sens;
 230         // match whole words only
 231         bool whole_words;
 232 };
 233
 234
 235 int findForward(DocIterator & cur, DocIterator const endcur,
 236                 MatchString const & match,
 237                 bool find_del = true, bool onlysel = false)
 238 {
 239         for (; cur; cur.forwardChar()) {
 240                 if (onlysel && endcur.pit() == cur.pit()
 241                     && endcur.idx() == cur.idx() && endcur.pos() < cur.pos())
 242                         break;
 243                 if (cur.inTexted()) {
 244                         int len = match(cur.paragraph(), cur.pos(), find_del);
 245                         if (len > 0)
 246                                 return len;
 247                 }
 248         }
 249         return 0;
 250 }
 251
 252
 253 int findBackwards(DocIterator & cur, DocIterator const endcur,
 254                   MatchString const & match,
 255                   bool find_del = true, bool onlysel = false)
 256 {
 257         while (cur) {
 258                 cur.backwardChar();
 259                 if (onlysel && endcur.pit() == cur.pit()
 260                     && endcur.idx() == cur.idx() && endcur.pos() > cur.pos())
 261                         break;
 262                 if (cur.inTexted()) {
 263                         int len = match(cur.paragraph(), cur.pos(), find_del);
 264                         if (len > 0)
 265                                 return len;
 266                 }
 267         }
 268         return 0;
 269 }
 270
 271
 272 bool searchAllowed(docstring const & str)
 273 {
 274         if (str.empty()) {
 275                 frontend::Alert::error(_("Search error"), _("Search string is empty"));
 276                 return false;
 277         }
 278         return true;
 279 }
 280
 281 } // namespace
 282
 283
 284 bool findOne(BufferView * bv, docstring const & searchstr,
 285              bool case_sens, bool whole, bool forward,
 286              bool find_del, bool check_wrap, bool const auto_wrap,
 287              bool instant, bool onlysel)
 288 {
 289         // Clean up previous selections with empty searchstr on instant
 290         if (searchstr.empty() && instant) {
 291                 if (bv->cursor().selection()) {
 292                         bv->setCursor(bv->cursor().selectionBegin());
 293                         bv->clearSelection();
 294                 }
 295                 return true;
 296         }
 297
 298         if (!searchAllowed(searchstr))
 299                 return false;
 300
 301         DocIterator const endcur = forward ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin();
 302
 303         if (onlysel && bv->cursor().selection()) {
 304                 docstring const matchstring = bv->cursor().selectionAsString(false);
 305                 docstring const lcmatchsting = support::lowercase(matchstring);
 306                 if (matchstring == searchstr || (!case_sens && lcmatchsting == lowercase(searchstr))) {
 307                         docstring q = _("The search string matches the selection, and search is limited to selection.\n"
 308                                         "Continue search outside?");
 309                         int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
 310                                 q, 0, 1, _("&Yes"), _("&No"));
 311                         if (search_answer == 0) {
 312                                 bv->clearSelection();
 313                                 if (findOne(bv, searchstr, case_sens, whole, forward,
 314                                             find_del, check_wrap, auto_wrap, false, false))
 315                                         return true;
 316                         }
 317                         return false;
 318                 }
 319         }
 320
 321         DocIterator cur = forward
 322                 ? ((instant || onlysel) ? bv->cursor().selectionBegin() : bv->cursor().selectionEnd())
 323                 : ((instant || onlysel) ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin());
 324
 325         MatchString const match(searchstr, case_sens, whole);
 326
 327         int match_len = forward
 328                 ? findForward(cur, endcur, match, find_del, onlysel)
 329                 : findBackwards(cur, endcur, match, find_del, onlysel);
 330
 331         if (match_len > 0)
 332                 bv->putSelectionAt(cur, match_len, !forward);
 333         else if (onlysel) {
 334                 docstring q = _("The search string was not found within the selection.\n"
 335                                 "Continue search outside?");
 336                 int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
 337                         q, 0, 1, _("&Yes"), _("&No"));
 338                 if (search_answer == 0) {
 339                         bv->clearSelection();
 340                         if (findOne(bv, searchstr, case_sens, whole, forward,
 341                                     find_del, check_wrap, auto_wrap, false, false))
 342                                 return true;
 343                 }
 344                 return false;
 345         }
 346         else if (check_wrap) {
 347                 DocIterator cur_orig(bv->cursor());
 348                 bool wrap = auto_wrap;
 349                 if (!auto_wrap) {
 350                         docstring q;
 351                         if (forward)
 352                                 q = _("End of file reached while searching forward.\n"
 353                                   "Continue searching from the beginning?");
 354                         else
 355                                 q = _("Beginning of file reached while searching backward.\n"
 356                                   "Continue searching from the end?");
 357                         int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 358                                 q, 0, 1, _("&Yes"), _("&No"));
 359                         wrap = wrap_answer == 0;
 360                 }
 361                 if (wrap) {
 362                         if (forward) {
 363                                 bv->cursor().clear();
 364                                 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
 365                         } else {
 366                                 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
 367                                 bv->cursor().backwardPos();
 368                         }
 369                         if (auto_wrap) {
 370                                 docstring const msg = forward
 371                                   ? _("Search reached end of document, continuing from beginning.")
 372                                   : _("Search reached beginning of document, continuing from end.");
 373                                 bv->message(msg);
 374                         }
 375                         bv->clearSelection();
 376                         if (findOne(bv, searchstr, case_sens, whole, forward,
 377                                     find_del, false, false, false, false))
 378                                 return true;
 379                 }
 380                 bv->cursor().setCursor(cur_orig);
 381                 return false;
 382         }
 383
 384         return match_len > 0;
 385 }
 386
 387
 388 namespace {
 389
 390 int replaceAll(BufferView * bv,
 391                docstring const & searchstr, docstring const & replacestr,
 392                bool case_sens, bool whole, bool onlysel)
 393 {
 394         Buffer & buf = bv->buffer();
 395
 396         if (!searchAllowed(searchstr) || buf.isReadonly())
 397                 return 0;
 398
 399         DocIterator startcur = bv->cursor().selectionBegin();
 400         DocIterator endcur = bv->cursor().selectionEnd();
 401         bool const had_selection = bv->cursor().selection();
 402
 403         MatchString const match(searchstr, case_sens, whole);
 404         int num = 0;
 405
 406         int const rsize = replacestr.size();
 407         int const ssize = searchstr.size();
 408
 409         Cursor cur(*bv);
 410         cur.setCursor(doc_iterator_begin(&buf));
 411         int match_len = findForward(cur, endcur, match, false, onlysel);
 412         while (match_len > 0) {
 413                 // Backup current cursor position and font.
 414                 pos_type const pos = cur.pos();
 415                 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
 416                 cur.recordUndo();
 417                 int ct_deleted_text = ssize -
 418                         cur.paragraph().eraseChars(pos, pos + match_len,
 419                                                    buf.params().track_changes);
 420                 cur.paragraph().insert(pos, replacestr, font,
 421                                        Change(buf.params().track_changes
 422                                               ? Change::INSERTED
 423                                               : Change::UNCHANGED));
 424                 for (int i = 0; i < rsize + ct_deleted_text
 425                      && cur.pos() < cur.lastpos(); ++i)
 426                         cur.forwardPos();
 427                 if (onlysel && cur.pit() == endcur.pit() && cur.idx() == endcur.idx()) {
 428                         // Adjust end of selection for replace-all in selection
 429                         if (rsize > ssize) {
 430                                 int const offset = rsize - ssize;
 431                                 for (int i = 0; i < offset + ct_deleted_text
 432                                      && endcur.pos() < endcur.lastpos(); ++i)
 433                                         endcur.forwardPos();
 434                         } else {
 435                                 int const offset = ssize - rsize;
 436                                 for (int i = 0; i < offset && endcur.pos() > 0; ++i)
 437                                         endcur.backwardPos();
 438                                 for (int i = 0; i < ct_deleted_text
 439                                      && endcur.pos() < endcur.lastpos(); ++i)
 440                                         endcur.forwardPos();
 441                         }
 442                 }
 443                 ++num;
 444                 match_len = findForward(cur, endcur, match, false, onlysel);
 445         }
 446
 447         bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
 448
 449         startcur.fixIfBroken();
 450         bv->setCursor(startcur);
 451
 452         // Reset selection, accounting for changes in selection
 453         if (had_selection) {
 454                 endcur.fixIfBroken();
 455                 bv->cursor().resetAnchor();
 456                 bv->setCursorSelectionTo(endcur);
 457         }
 458
 459         return num;
 460 }
 461
 462
 463 // the idea here is that we are going to replace the string that
 464 // is selected IF it is the search string.
 465 // if there is a selection, but it is not the search string, then
 466 // we basically ignore it. (FIXME We ought to replace only within
 467 // the selection.)
 468 // if there is no selection, then:
 469 //  (i) if some search string has been provided, then we find it.
 470 //      (think of how the dialog works when you hit "replace" the
 471 //      first time.)
 472 // (ii) if no search string has been provided, then we treat the
 473 //      word the cursor is in as the search string. (why? i have no
 474 //      idea.) but this only works in text?
 475 //
 476 // returns the number of replacements made (one, if any) and
 477 // whether anything at all was done.
 478 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
 479                            docstring const & replacestr, bool case_sens,
 480                            bool whole, bool forward, bool findnext, bool wrap,
 481                            bool onlysel)
 482 {
 483         Cursor & cur = bv->cursor();
 484         if (!cur.selection() || onlysel) {
 485                 // no selection, non-empty search string: find it
 486                 if (!searchstr.empty()) {
 487                         bool const found = findOne(bv, searchstr, case_sens, whole,
 488                                                    forward, true, findnext, wrap, false, onlysel);
 489                         return make_pair(found, 0);
 490                 }
 491                 // empty search string
 492                 if (!cur.inTexted())
 493                         // bail in math
 494                         return make_pair(false, 0);
 495                 // select current word and treat it as the search string.
 496                 // This causes a minor bug as undo will restore this selection,
 497                 // which the user did not create (#8986).
 498                 cur.innerText()->selectWord(cur, WHOLE_WORD);
 499                 searchstr = cur.selectionAsString(false, true);
 500         }
 501
 502         // if we still don't have a search string, report the error
 503         // and abort.
 504         if (!searchAllowed(searchstr))
 505                 return make_pair(false, 0);
 506
 507         bool have_selection = cur.selection();
 508         docstring const selected = cur.selectionAsString(false, true);
 509         bool match =
 510                 case_sens
 511                 ? searchstr == selected
 512                 : compare_no_case(searchstr, selected) == 0;
 513
 514         // no selection or current selection is not search word:
 515         // just find the search word
 516         if (!have_selection || !match) {
 517                 bool const found = findOne(bv, searchstr, case_sens, whole, forward,
 518                                            true, findnext, wrap, false, onlysel);
 519                 return make_pair(found, 0);
 520         }
 521
 522         // we're now actually ready to replace. if the buffer is
 523         // read-only, we can't, though.
 524         if (bv->buffer().isReadonly())
 525                 return make_pair(false, 0);
 526
 527         cap::replaceSelectionWithString(cur, replacestr);
 528         if (forward) {
 529                 cur.pos() += replacestr.length();
 530                 LASSERT(cur.pos() <= cur.lastpos(),
 531                         cur.pos() = cur.lastpos());
 532         }
 533         if (findnext)
 534                 findOne(bv, searchstr, case_sens, whole,
 535                         forward, false, findnext, wrap, false, onlysel);
 536
 537         return make_pair(true, 1);
 538 }
 539
 540 } // namespace
 541
 542
 543 docstring const find2string(docstring const & search,
 544                             bool casesensitive, bool matchword,
 545                             bool forward, bool wrap, bool instant,
 546                             bool onlysel)
 547 {
 548         odocstringstream ss;
 549         ss << search << '\n'
 550            << int(casesensitive) << ' '
 551            << int(matchword) << ' '
 552            << int(forward) << ' '
 553            << int(wrap) << ' '
 554            << int(instant) << ' '
 555            << int(onlysel);
 556         return ss.str();
 557 }
 558
 559
 560 docstring const replace2string(docstring const & replace,
 561                                docstring const & search,
 562                                bool casesensitive, bool matchword,
 563                                bool all, bool forward, bool findnext,
 564                                bool wrap, bool onlysel)
 565 {
 566         odocstringstream ss;
 567         ss << replace << '\n'
 568            << search << '\n'
 569            << int(casesensitive) << ' '
 570            << int(matchword) << ' '
 571            << int(all) << ' '
 572            << int(forward) << ' '
 573            << int(findnext) << ' '
 574            << int(wrap) << ' '
 575            << int(onlysel);
 576         return ss.str();
 577 }
 578
 579
 580 docstring const string2find(docstring const & argument,
 581                               bool &casesensitive,
 582                               bool &matchword,
 583                               bool &forward,
 584                               bool &wrap,
 585                               bool &instant,
 586                               bool &onlysel)
 587 {
 588         // data is of the form
 589         // "<search>
 590         //  <casesensitive> <matchword> <forward> <wrap> <onlysel>"
 591         docstring search;
 592         docstring howto = split(argument, search, '\n');
 593
 594         casesensitive = parse_bool(howto);
 595         matchword     = parse_bool(howto);
 596         forward       = parse_bool(howto, true);
 597         wrap          = parse_bool(howto);
 598         instant       = parse_bool(howto);
 599         onlysel       = parse_bool(howto);
 600
 601         return search;
 602 }
 603
 604
 605 bool lyxfind(BufferView * bv, FuncRequest const & ev)
 606 {
 607         if (!bv || ev.action() != LFUN_WORD_FIND)
 608                 return false;
 609
 610         //lyxerr << "find called, cmd: " << ev << endl;
 611         bool casesensitive;
 612         bool matchword;
 613         bool forward;
 614         bool wrap;
 615         bool instant;
 616         bool onlysel;
 617
 618         docstring search = string2find(ev.argument(), casesensitive,
 619                                        matchword, forward, wrap, instant, onlysel);
 620
 621         return findOne(bv, search, casesensitive, matchword, forward,
 622                        false, true, wrap, instant, onlysel);
 623 }
 624
 625
 626 bool lyxreplace(BufferView * bv, FuncRequest const & ev)
 627 {
 628         if (!bv || ev.action() != LFUN_WORD_REPLACE)
 629                 return false;
 630
 631         // data is of the form
 632         // "<search>
 633         //  <replace>
 634         //  <casesensitive> <matchword> <all> <forward> <findnext> <wrap> <onlysel>"
 635         docstring search;
 636         docstring rplc;
 637         docstring howto = split(ev.argument(), rplc, '\n');
 638         howto = split(howto, search, '\n');
 639
 640         bool casesensitive = parse_bool(howto);
 641         bool matchword     = parse_bool(howto);
 642         bool all           = parse_bool(howto);
 643         bool forward       = parse_bool(howto, true);
 644         bool findnext      = parse_bool(howto, true);
 645         bool wrap          = parse_bool(howto);
 646         bool onlysel       = parse_bool(howto);
 647
 648         if (!bv->cursor().selection())
 649                 // only selection only makes sense with selection
 650                 onlysel = false;
 651
 652         bool update = false;
 653
 654         int replace_count = 0;
 655         if (all) {
 656                 replace_count = replaceAll(bv, search, rplc, casesensitive,
 657                                            matchword, onlysel);
 658                 update = replace_count > 0;
 659         } else {
 660                 pair<bool, int> rv =
 661                         replaceOne(bv, search, rplc, casesensitive, matchword,
 662                                    forward, findnext, wrap, onlysel);
 663                 update = rv.first;
 664                 replace_count = rv.second;
 665         }
 666
 667         Buffer const & buf = bv->buffer();
 668         if (!update) {
 669                 // emit message signal.
 670                 if (onlysel)
 671                         buf.message(_("String not found in selection."));
 672                 else
 673                         buf.message(_("String not found."));
 674         } else {
 675                 if (replace_count == 0) {
 676                         buf.message(_("String found."));
 677                 } else if (replace_count == 1) {
 678                         buf.message(_("String has been replaced."));
 679                 } else {
 680                         docstring const str = onlysel
 681                                         ? bformat(_("%1$d strings have been replaced in the selection."), replace_count)
 682                                         : bformat(_("%1$d strings have been replaced."), replace_count);
 683                         buf.message(str);
 684                 }
 685         }
 686         return update;
 687 }
 688
 689
 690 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
 691 {
 692         for (; cur; cur.forwardPos())
 693                 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
 694                         return true;
 695
 696         if (check_wrap) {
 697                 DocIterator cur_orig(bv->cursor());
 698                 docstring q = _("End of file reached while searching forward.\n"
 699                           "Continue searching from the beginning?");
 700                 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 701                         q, 0, 1, _("&Yes"), _("&No"));
 702                 if (wrap_answer == 0) {
 703                         bv->cursor().clear();
 704                         bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
 705                         bv->clearSelection();
 706                         cur.setCursor(bv->cursor().selectionBegin());
 707                         if (findNextChange(bv, cur, false))
 708                                 return true;
 709                 }
 710                 bv->cursor().setCursor(cur_orig);
 711         }
 712
 713         return false;
 714 }
 715
 716
 717 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
 718 {
 719         for (cur.backwardPos(); cur; cur.backwardPos()) {
 720                 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
 721                         return true;
 722         }
 723
 724         if (check_wrap) {
 725                 DocIterator cur_orig(bv->cursor());
 726                 docstring q = _("Beginning of file reached while searching backward.\n"
 727                           "Continue searching from the end?");
 728                 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 729                         q, 0, 1, _("&Yes"), _("&No"));
 730                 if (wrap_answer == 0) {
 731                         bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
 732                         bv->cursor().backwardPos();
 733                         bv->clearSelection();
 734                         cur.setCursor(bv->cursor().selectionBegin());
 735                         if (findPreviousChange(bv, cur, false))
 736                                 return true;
 737                 }
 738                 bv->cursor().setCursor(cur_orig);
 739         }
 740
 741         return false;
 742 }
 743
 744
 745 bool selectChange(Cursor & cur, bool forward)
 746 {
 747         if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
 748                 return false;
 749         Change ch = cur.paragraph().lookupChange(cur.pos());
 750
 751         CursorSlice tip1 = cur.top();
 752         for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
 753                 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
 754                 if (!ch2.isSimilarTo(ch))
 755                         break;
 756         }
 757         CursorSlice tip2 = cur.top();
 758         for (; tip2.pit() > 0 || tip2.pos() > 0;) {
 759                 tip2.backwardPos();
 760                 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
 761                 if (!ch2.isSimilarTo(ch)) {
 762                         // take a step forward to correctly set the selection
 763                         tip2.forwardPos();
 764                         break;
 765                 }
 766         }
 767         if (forward)
 768                 swap(tip1, tip2);
 769         cur.top() = tip1;
 770         cur.bv().mouseSetCursor(cur, false);
 771         cur.top() = tip2;
 772         cur.bv().mouseSetCursor(cur, true);
 773         return true;
 774 }
 775
 776
 777 namespace {
 778
 779
 780 bool findChange(BufferView * bv, bool forward)
 781 {
 782         Cursor cur(*bv);
 783         cur.setCursor(forward ? bv->cursor().selectionEnd()
 784                       : bv->cursor().selectionBegin());
 785         forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
 786         return selectChange(cur, forward);
 787 }
 788
 789 } // namespace
 790
 791 bool findNextChange(BufferView * bv)
 792 {
 793         return findChange(bv, true);
 794 }
 795
 796
 797 bool findPreviousChange(BufferView * bv)
 798 {
 799         return findChange(bv, false);
 800 }
 801
 802
 803
 804 namespace {
 805
 806 typedef vector<pair<string, string> > Escapes;
 807
 808 string string2regex(string in)
 809 {
 810         static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\$\\])" };
 811         string temp = std::regex_replace(in, specialChars,  R"(\$&)" );
 812         string temp2("");
 813         size_t lastpos = 0;
 814         size_t fl_pos = 0;
 815         int offset = 1;
 816         while (fl_pos < temp.size()) {
 817                 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
 818                 if (fl_pos == string::npos)
 819                         break;
 820                 offset = 16;
 821                 temp2 += temp.substr(lastpos, fl_pos - lastpos);
 822                 temp2 += "\\n";
 823                 lastpos = fl_pos;
 824         }
 825         if (lastpos == 0)
 826                 return(temp);
 827         if (lastpos < temp.size()) {
 828                 temp2 += temp.substr(lastpos, temp.size() - lastpos);
 829         }
 830         return temp2;
 831 }
 832
 833 static void buildAccentsMap();
 834
 835 string correctRegex(string t, bool withformat)
 836 {
 837         /* Convert \backslash => \
 838          * and \{, \}, \[, \] => {, }, [, ]
 839          */
 840         string s("");
 841         regex wordre("(\\\\)*(\\\\(([A-Za-z]+|[\\{\\}])( |\\{\\})?|[\\[\\]\\{\\}]))");
 842         size_t lastpos = 0;
 843         smatch sub;
 844         bool backslashed = false;
 845         if (accents.empty())
 846                 buildAccentsMap();
 847
 848         //LYXERR0("correctRegex input '" << t << "'");
 849         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
 850                 sub = *it;
 851                 string replace;
 852                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
 853                         continue;
 854                 }
 855                 else {
 856                         if (sub.str(4) == "backslash") {
 857                                 replace = "\\";
 858                                 {
 859                                         // transforms '\backslash \{' into '\{'
 860                                         string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
 861                                         if ((next == "\\{") || (next == "\\}")) {
 862                                                 replace = "";
 863                                                 backslashed = true;
 864                                         }
 865                                 }
 866                         }
 867                         else if (sub.str(4) == "mathcircumflex")
 868                                 replace = "^";
 869                         else if (backslashed) {
 870                                 backslashed = false;
 871                                 if (withformat) {
 872                                         if (sub.str(3) == "{")
 873                                                 replace = accents["braceleft"];
 874                                         else if (sub.str(3) == "}")
 875                                                 replace = accents["braceright"];
 876                                         else {
 877                                                 // else part should not exist
 878                                                 LASSERT(1, /**/);
 879                                         }
 880                                 }
 881                                 else {
 882                                         if (sub.str(3) == "{")
 883                                                 replace = "\\{";
 884                                         else if (sub.str(3) == "}")
 885                                                 replace = "\\}";
 886                                         else {
 887                                                 // else part should not exist
 888                                                 LASSERT(1, /**/);
 889                                         }
 890                                 }
 891                         }
 892                         else if (sub.str(4) == "{") // transforms '\{' into '{'
 893                                 replace = "{";
 894                         else if (sub.str(4) == "}")
 895                                 replace = "}";
 896                         else {
 897                                 AccentsIterator it_ac = accents.find(sub.str(4));
 898                                 if (it_ac == accents.end()) {
 899                                         replace = sub.str(2);
 900                                 }
 901                                 else {
 902                                         replace = it_ac->second;
 903                                 }
 904                         }
 905                 }
 906                 if (lastpos < (size_t) sub.position(2))
 907                         s += t.substr(lastpos, sub.position(2) - lastpos);
 908                 s += replace;
 909                 lastpos = sub.position(2) + sub.length(2);
 910         }
 911         if (lastpos == 0)
 912                 s = t;
 913         else if (lastpos < t.length())
 914                 s += t.substr(lastpos, t.length() - lastpos);
 915         // Handle quotes in regex
 916         // substitute all '„', '“', '»', '«' with '"'
 917         // and all '‚', '‘', '›', '‹' with "\'"
 918         static std::regex plainquotes { R"(„|“|»|«)" };
 919         static std::regex innerquotes { R"(‚|‘|›|‹)" };
 920         t = std::regex_replace(s, plainquotes, R"(")");
 921         s = std::regex_replace(t, innerquotes, R"(')");
 922         //LYXERR0("correctRegex output '" << s << "'");
 923         return s;
 924 }
 925
 926 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
 927 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
 928 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
 929 string escape_for_regex(string s, bool withformat)
 930 {
 931         size_t lastpos = 0;
 932         string result = "";
 933         while (lastpos < s.size()) {
 934                 size_t regex_pos = s.find("\\regexp{", lastpos);
 935                 if (regex_pos == string::npos) {
 936                         regex_pos = s.size();
 937                 }
 938                 if (regex_pos > lastpos) {
 939                         result += string2regex(s.substr(lastpos, regex_pos-lastpos));
 940                         lastpos = regex_pos;
 941                         if (lastpos == s.size())
 942                                 break;
 943                 }
 944                 size_t end_pos = s.find("\\endregexp", regex_pos + 8);
 945                 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
 946                 lastpos = end_pos + 13;
 947         }
 948         return result;
 949 }
 950
 951
 952 /// Wrapper for lyx::regex_replace with simpler interface
 953 bool regex_replace(string const & s, string & t, string const & searchstr,
 954                    string const & replacestr)
 955 {
 956         regex e(searchstr, regex_constants::ECMAScript);
 957         ostringstream oss;
 958         ostream_iterator<char, char> it(oss);
 959         regex_replace(it, s.begin(), s.end(), e, replacestr);
 960         // tolerate t and s be references to the same variable
 961         bool rv = (s != oss.str());
 962         t = oss.str();
 963         return rv;
 964 }
 965
 966 class MatchResult {
 967 public:
 968         enum range {
 969                 newIsTooFar,
 970                 newIsBetter,
 971                 newIsInvalid
 972         };
 973         int match_len;
 974         int match_prefix;
 975         int match2end;
 976         int pos;
 977         int leadsize;
 978         int pos_len;
 979         int searched_size;
 980         vector <string> result = vector <string>();
 981         MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {}
 982 };
 983
 984 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
 985 {
 986         if (newres.match2end < oldres.match2end)
 987                 return MatchResult::newIsTooFar;
 988         if (newres.match_len < oldres.match_len)
 989                 return MatchResult::newIsTooFar;
 990
 991         if (newres.match_len == oldres.match_len) {
 992                 if (newres.match2end == oldres.match2end)
 993                         return MatchResult::newIsBetter;
 994         }
 995         return MatchResult::newIsInvalid;
 996 }
 997
 998 /** The class performing a match between a position in the document and the FindAdvOptions.
 999  **/
1000
1001 class MatchStringAdv {
1002 public:
1003         MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
1004
1005         /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
1006          ** constructor as opt.search, under the opt.* options settings.
1007          **
1008          ** @param at_begin
1009          **     If set, then match is searched only against beginning of text starting at cur.
1010          **     If unset, then match is searched anywhere in text starting at cur.
1011          **
1012          ** @return
1013          ** The length of the matching text, or zero if no match was found.
1014          **/
1015         MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1016 #if QTSEARCH
1017         bool regexIsValid;
1018         string regexError;
1019 #endif
1020
1021 public:
1022         /// buffer
1023         lyx::Buffer * p_buf;
1024         /// first buffer on which search was started
1025         lyx::Buffer * const p_first_buf;
1026         /// options
1027         FindAndReplaceOptions const & opt;
1028
1029 private:
1030         /// Auxiliary find method (does not account for opt.matchword)
1031         MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1032         void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
1033
1034         /** Normalize a stringified or latexified LyX paragraph.
1035          **
1036          ** Normalize means:
1037          ** <ul>
1038          **   <li>if search is not casesensitive, then lowercase the string;
1039          **   <li>remove any newline at begin or end of the string;
1040          **   <li>replace any newline in the middle of the string with a simple space;
1041          **   <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
1042          ** </ul>
1043          **
1044          ** @todo Normalization should also expand macros, if the corresponding
1045          ** search option was checked.
1046          **/
1047         string normalize(docstring const & s, bool ignore_fomat) const;
1048         // normalized string to search
1049         string par_as_string;
1050         // regular expression to use for searching
1051         // regexp2 is same as regexp, but prefixed with a ".*?"
1052 #if QTSEARCH
1053         QRegularExpression regexp;
1054         QRegularExpression regexp2;
1055 #else
1056         regex regexp;
1057         regex regexp2;
1058 #endif
1059         // leading format material as string
1060         string lead_as_string;
1061         // par_as_string after removal of lead_as_string
1062         string par_as_string_nolead;
1063         // unmatched open braces in the search string/regexp
1064         int open_braces;
1065         // number of (.*?) subexpressions added at end of search regexp for closing
1066         // environments, math mode, styles, etc...
1067         int close_wildcards;
1068 public:
1069         // Are we searching with regular expressions ?
1070         bool use_regexp = false;
1071         static int valid_matches;
1072         static vector <string> matches;
1073         void FillResults(MatchResult &found_mr);
1074 };
1075
1076 int MatchStringAdv::valid_matches = 0;
1077 vector <string> MatchStringAdv::matches = vector <string>(10);
1078
1079 void MatchStringAdv::FillResults(MatchResult &found_mr)
1080 {
1081         if (found_mr.match_len > 0) {
1082                 valid_matches = found_mr.result.size();
1083                 for (size_t i = 0; i < found_mr.result.size(); i++)
1084                         matches[i] = found_mr.result[i];
1085         } else
1086                 valid_matches = 0;
1087 }
1088
1089 static docstring buffer_to_latex(Buffer & buffer)
1090 {
1091         //OutputParams runparams(&buffer.params().encoding());
1092         OutputParams runparams(encodings.fromLyXName("utf8"));
1093         odocstringstream ods;
1094         otexstream os(ods);
1095         runparams.nice = true;
1096         runparams.flavor = Flavor::XeTeX;
1097         runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1098         // No side effect of file copying and image conversion
1099         runparams.dryrun = true;
1100         if (ignoreFormats.getDeleted())
1101                 runparams.for_search = OutputParams::SearchWithoutDeleted;
1102         else
1103                 runparams.for_search = OutputParams::SearchWithDeleted;
1104         if (ignoreFormats.getNonContent()) {
1105                 runparams.for_search |= OutputParams::SearchNonOutput;
1106         }
1107         pit_type const endpit = buffer.paragraphs().size();
1108         for (pit_type pit = 0; pit != endpit; ++pit) {
1109                 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
1110                 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
1111         }
1112         return ods.str();
1113 }
1114
1115 static string latexNamesToUtf8(docstring strIn)
1116 {
1117         string addtmp = to_utf8(strIn);
1118         static regex const rmAcc("(\\\\)*("
1119                                          "\\\\([A-Za-z]+\\{.\\})"       // e.g. "ddot{A}" == sub.str(3)
1120                                         "|\\\\([A-Za-z]+)( |\\{\\})?"   // e.g. "LyX", "LyX{}", "LyX " == sub.str(4)
1121                                         ")"
1122                                 );
1123         size_t lastpos = 0;
1124         smatch sub;
1125         string replace;
1126         string add("");
1127         if (accents.empty())
1128                 buildAccentsMap();
1129         for (sregex_iterator it_add(addtmp.begin(), addtmp.end(), rmAcc), end; it_add != end; ++it_add) {
1130                 sub = *it_add;
1131                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
1132                         continue;
1133                 }
1134                 else {
1135                         string key;
1136                         if (sub.length(3) > 0)
1137                                 key = sub.str(3);
1138                         else
1139                                 key = sub.str(4);
1140                         AccentsIterator it_ac = accents.find(key);
1141                         if (it_ac == accents.end()) {
1142                                 replace = sub.str(2);
1143                         }
1144                         else {
1145                                 replace = it_ac->second;
1146                         }
1147                 }
1148                 if (lastpos < (size_t) sub.position(2))
1149                         add += addtmp.substr(lastpos, sub.position(2) - lastpos);
1150                 add += replace;
1151                 lastpos = sub.position(2) + sub.length(2);
1152         }
1153         if (lastpos == 0)
1154                 add = addtmp;
1155         else if (addtmp.length() > lastpos)
1156                 add += addtmp.substr(lastpos, addtmp.length() - lastpos);
1157         LYXERR(Debug::FIND, "Adding to search string: '"
1158                         << add << "'");
1159         return add;
1160 }
1161
1162 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1163 {
1164         docstring str;
1165         if (!opt.ignoreformat) {
1166                 str = buffer_to_latex(buffer);
1167         } else {
1168                 // OutputParams runparams(&buffer.params().encoding());
1169                 OutputParams runparams(encodings.fromLyXName("utf8"));
1170                 runparams.nice = true;
1171                 runparams.flavor = Flavor::XeTeX;
1172                 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1173                 runparams.dryrun = true;
1174                 int option = AS_STR_INSETS |AS_STR_PLAINTEXT;
1175                 if (ignoreFormats.getDeleted()) {
1176                         option |= AS_STR_SKIPDELETE;
1177                         runparams.for_search = OutputParams::SearchWithoutDeleted;
1178                 }
1179                 else {
1180                         runparams.for_search = OutputParams::SearchWithDeleted;
1181                 }
1182                 if (ignoreFormats.getNonContent()) {
1183                         runparams.for_search |= OutputParams::SearchNonOutput;
1184                 }
1185                 string t("");
1186                 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1187                         Paragraph const & par = buffer.paragraphs().at(pit);
1188                         string add = latexNamesToUtf8(par.asString(pos_type(0), par.size(),
1189                                                                 option,
1190                                                                 &runparams));
1191                         LYXERR(Debug::FIND, "Adding to search string: '"
1192                                 << add << "'");
1193                         t += add;
1194                 }
1195                 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1196                 while (regex_replace(t, t, "\\\\(text|lyxmathsym|ensuremath)\\{([^\\}]*)\\}", "$2"));
1197                 str = from_utf8(t);
1198         }
1199         return str;
1200 }
1201
1202
1203 /// Return separation pos between the leading material and the rest
1204 static size_t identifyLeading(string const & s)
1205 {
1206         string t = s;
1207         // @TODO Support \item[text]
1208         // Kornel: Added textsl, textsf, textit, texttt and noun
1209         // + allow to search for colored text too
1210         while (regex_replace(t, t, "^\\\\(("
1211                              "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1212                              "lyxaddress|lyxrightaddress|"
1213                              "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1214                              "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1215                              "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1216                              "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1217                || regex_replace(t, t, "^\\$", "")
1218                || regex_replace(t, t, "^\\\\\\[", "")
1219                || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1220                || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1221                ;
1222         LYXERR(Debug::FIND, "  after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1223         return s.find(t);
1224 }
1225
1226 /*
1227  * Given a latexified string, retrieve some handled features
1228  * The features of the regex will later be compared with the features
1229  * of the searched text. If the regex features are not a
1230  * subset of the analized, then, in not format ignoring search
1231  * we can early stop the search in the relevant inset.
1232  */
1233 typedef map<string, bool> Features;
1234
1235 static Features identifyFeatures(string const & s)
1236 {
1237         static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1238         static regex const valid("^("
1239                 "("
1240                         "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1241                                 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1242                                 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1243                         "(u|uu)line|(s|x)out|uwave|"
1244                         "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1245                 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1246         smatch sub;
1247         bool displ = true;
1248         Features info;
1249
1250         for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1251                 sub = *it;
1252                 if (displ) {
1253                         if (sub.str(1).compare("regexp") == 0) {
1254                                 displ = false;
1255                                 continue;
1256                         }
1257                         string token = sub.str(1);
1258                         smatch sub2;
1259                         if (regex_match(token, sub2, valid)) {
1260                                 info[token] = true;
1261                         }
1262                         else {
1263                                 // ignore
1264                         }
1265                 }
1266                 else {
1267                         if (sub.str(1).compare("endregexp") == 0) {
1268                                 displ = true;
1269                                 continue;
1270                         }
1271                 }
1272         }
1273         return info;
1274 }
1275
1276 /*
1277  * defines values features of a key "\\[a-z]+{"
1278  */
1279 class KeyInfo {
1280 public:
1281         enum KeyType {
1282                 /* Char type with content discarded
1283                  * like \hspace{1cm} */
1284                 noContent,
1285                 /* Char, like \backslash */
1286                 isChar,
1287                 /* replace starting backslash with '#' */
1288                 isText,
1289                 /* \part, \section*, ... */
1290                 isSectioning,
1291                 /* title, author etc */
1292                 isTitle,
1293                 /* \foreignlanguage{ngerman}, ... */
1294                 isMain,
1295                 /* inside \code{}
1296                  * to discard language in content */
1297                 noMain,
1298                 isRegex,
1299                 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1300                 isMath,
1301                 /* fonts, colors, markups, ... */
1302                 isStandard,
1303                 /* footnotesize, ... large, ...
1304                  * Ignore all of them */
1305                 isSize,
1306                 invalid,
1307                 /* inputencoding, ...
1308                  * Discard also content, because they do not help in search */
1309                 doRemove,
1310                 /* twocolumns, ...
1311                  * like remove, but also all arguments */
1312                 removeWithArg,
1313                 /* item, listitem */
1314                 isList,
1315                 /* tex, latex, ... like isChar */
1316                 isIgnored,
1317                 /* like \lettrine[lines=5]{}{} */
1318                 cleanToStart,
1319                 // like isStandard, but always remove head
1320                 headRemove,
1321                 /* End of arguments marker for lettrine,
1322                  * so that they can be ignored */
1323                 endArguments
1324         };
1325         KeyInfo() = default;
1326         KeyInfo(KeyType type, int parcount, bool disable)
1327                 : keytype(type),
1328                   parenthesiscount(parcount),
1329                   disabled(disable) {}
1330         KeyType keytype = invalid;
1331         string head;
1332         int _tokensize = -1;
1333         int _tokenstart = -1;
1334         int _dataStart = -1;
1335         int _dataEnd = -1;
1336         int parenthesiscount = 1;
1337         bool disabled = false;
1338         bool used = false; /* by pattern */
1339 };
1340
1341 class Border {
1342 public:
1343         Border(int l=0, int u=0) : low(l), upper(u) {}
1344         int low;
1345         int upper;
1346 };
1347
1348 #define MAXOPENED 30
1349 class Intervall {
1350         bool isPatternString_;
1351 public:
1352         explicit Intervall(bool isPattern, string const & p)
1353                 : isPatternString_(isPattern), par(p), ignoreidx(-1),
1354                   actualdeptindex(0), hasTitle(false), langcount(0)
1355         {
1356                 depts[0] = 0;
1357                 closes[0] = 0;
1358         }
1359
1360         string par;
1361         int ignoreidx;
1362         static vector<Border> borders;
1363         int depts[MAXOPENED];
1364         int closes[MAXOPENED];
1365         int actualdeptindex;
1366         int previousNotIgnored(int) const;
1367         int nextNotIgnored(int) const;
1368         void handleOpenP(int i);
1369         void handleCloseP(int i, bool closingAllowed);
1370         void resetOpenedP(int openPos);
1371         void addIntervall(int upper);
1372         void addIntervall(int low, int upper); /* if explicit */
1373         void removeAccents();
1374         void setForDefaultLang(KeyInfo const & defLang) const;
1375         int findclosing(int start, int end, char up, char down, int repeat);
1376         void handleParentheses(int lastpos, bool closingAllowed);
1377         bool hasTitle;
1378         // Number of disabled language specs up
1379         // to current position in actual interval
1380         int langcount;
1381         int isOpeningPar(int pos) const;
1382         string titleValue;
1383         void output(ostringstream &os, int lastpos);
1384         // string show(int lastpos);
1385 };
1386
1387 vector<Border> Intervall::borders = vector<Border>(30);
1388
1389 int Intervall::isOpeningPar(int pos) const
1390 {
1391         if ((pos < 0) || (size_t(pos) >= par.size()))
1392                 return 0;
1393         if (par[pos] != '{')
1394                 return 0;
1395         if (size_t(pos) + 2 >= par.size())
1396                 return 1;
1397         if (par[pos+2] != '}')
1398                 return 1;
1399         if (par[pos+1] == '[' || par[pos+1] == ']')
1400                 return 3;
1401         return 1;
1402 }
1403
1404 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1405 {
1406         // Enable the use of first token again
1407         if (ignoreidx >= 0) {
1408                 int value = defLang._tokenstart + defLang._tokensize;
1409                 int borderidx = 0;
1410                 if (hasTitle)
1411                         borderidx = 1;
1412                 if (value > 0) {
1413                         if (borders[borderidx].low < value)
1414                                 borders[borderidx].low = value;
1415                         if (borders[borderidx].upper < value)
1416                                 borders[borderidx].upper = value;
1417                 }
1418         }
1419 }
1420
1421 static void checkDepthIndex(int val)
1422 {
1423         static int maxdepthidx = MAXOPENED-2;
1424         static int lastmaxdepth = 0;
1425         if (val > lastmaxdepth) {
1426                 LYXERR(Debug::INFO, "Depth reached " << val);
1427                 lastmaxdepth = val;
1428         }
1429         if (val > maxdepthidx) {
1430                 maxdepthidx = val;
1431                 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1432         }
1433 }
1434
1435 #if 0
1436 // Not needed, because borders are now dynamically expanded
1437 static void checkIgnoreIdx(int val)
1438 {
1439         static int lastmaxignore = -1;
1440         if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1441                 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1442                 lastmaxignore = val;
1443         }
1444 }
1445 #endif
1446
1447 /*
1448  * Expand the region of ignored parts of the input latex string
1449  * The region is only relevant in output()
1450  */
1451 void Intervall::addIntervall(int low, int upper)
1452 {
1453         int idx;
1454         if (low == upper) return;
1455         for (idx = ignoreidx+1; idx > 0; --idx) {
1456                 if (low > borders[idx-1].upper) {
1457                         break;
1458                 }
1459         }
1460         Border br(low, upper);
1461         if (idx > ignoreidx) {
1462                 if (borders.size() <= size_t(idx)) {
1463                         borders.push_back(br);
1464                 }
1465                 else {
1466                         borders[idx] = br;
1467                 }
1468                 ignoreidx = idx;
1469                 // checkIgnoreIdx(ignoreidx);
1470                 return;
1471         }
1472         else {
1473                 // Expand only if one of the new bound is inside the interwall
1474                 // We know here that br.low > borders[idx-1].upper
1475                 if (br.upper < borders[idx].low) {
1476                         // We have to insert at this pos
1477                         if (size_t(ignoreidx+1) >= borders.size()) {
1478                                 borders.push_back(borders[ignoreidx]);
1479                         }
1480                         else {
1481                                 borders[ignoreidx+1] = borders[ignoreidx];
1482                         }
1483                         for (int i = ignoreidx; i > idx; --i) {
1484                                 borders[i] = borders[i-1];
1485                         }
1486                         borders[idx] = br;
1487                         ignoreidx += 1;
1488                         // checkIgnoreIdx(ignoreidx);
1489                         return;
1490                 }
1491                 // Here we know, that we are overlapping
1492                 if (br.low > borders[idx].low)
1493                         br.low = borders[idx].low;
1494                 // check what has to be concatenated
1495                 int count = 0;
1496                 for (int i = idx; i <= ignoreidx; i++) {
1497                         if (br.upper >= borders[i].low) {
1498                                 count++;
1499                                 if (br.upper < borders[i].upper)
1500                                         br.upper = borders[i].upper;
1501                         }
1502                         else {
1503                                 break;
1504                         }
1505                 }
1506                 // count should be >= 1 here
1507                 borders[idx] = br;
1508                 if (count > 1) {
1509                         for (int i = idx + count; i <= ignoreidx; i++) {
1510                                 borders[i-count+1] = borders[i];
1511                         }
1512                         ignoreidx -= count - 1;
1513                         return;
1514                 }
1515         }
1516 }
1517
1518 static void buildaccent(string n, string param, string values)
1519 {
1520         stringstream s(n);
1521         string name;
1522         const char delim = '|';
1523         while (getline(s, name, delim)) {
1524                 size_t start = 0;
1525                 for (char c : param) {
1526                         string key = name + "{" + c + "}";
1527                         // get the corresponding utf8-value
1528                         if ((values[start] & 0xc0) != 0xc0) {
1529                                 // should not happen, utf8 encoding starts at least with 11xxxxxx
1530                                 // but value for '\dot{i}' is 'i', which is ascii
1531                                 if ((values[start] & 0x80) == 0) {
1532                                         // is ascii
1533                                         accents[key] = values.substr(start, 1);
1534                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1535                                 }
1536                                 start++;
1537                                 continue;
1538                         }
1539                         for (int j = 1; ;j++) {
1540                                 if (start + j >= values.size()) {
1541                                         accents[key] = values.substr(start, j);
1542                                         start = values.size() - 1;
1543                                         break;
1544                                 }
1545                                 else if ((values[start+j] & 0xc0) != 0x80) {
1546                                         // This is the first byte of following utf8 char
1547                                         accents[key] = values.substr(start, j);
1548                                         start += j;
1549                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1550                                         break;
1551                                 }
1552                         }
1553                 }
1554         }
1555 }
1556
1557 // Helper function
1558 static string getutf8(unsigned uchar)
1559 {
1560 #define maxc 5
1561         string ret = string();
1562         char c[maxc] = {0};
1563         if (uchar <= 0x7f) {
1564                 c[maxc-1] = uchar & 0x7f;
1565         }
1566         else {
1567                 unsigned char rest = 0x40;
1568                 unsigned char first = 0x80;
1569                 int start = maxc-1;
1570                 for (int i = start; i >=0; --i) {
1571                         if (uchar < rest) {
1572                                 c[i] = first + uchar;
1573                                 break;
1574                         }
1575                         c[i] = 0x80 | (uchar &  0x3f);
1576                         uchar >>= 6;
1577                         rest >>= 1;
1578                         first >>= 1;
1579                         first |= 0x80;
1580                 }
1581         }
1582         for (int i = 0; i < maxc; i++) {
1583                 if (c[i] == 0) continue;
1584                 ret += c[i];
1585         }
1586         return(ret);
1587 }
1588
1589 static void addAccents(string latex_in, string unicode_out)
1590 {
1591         latex_in = latex_in.substr(1);
1592         AccentsIterator it_ac = accents.find(latex_in);
1593         if (it_ac == accents.end()) {
1594                 accents[latex_in] = unicode_out;
1595         }
1596         else {
1597                 LYXERR0("Key " << latex_in  << " already set");
1598         }
1599 }
1600
1601 void static fillMissingUnicodesymbols()
1602 {
1603         addAccents("\\pounds", getutf8(0x00a3));
1604         addAccents("\\textsterling", getutf8(0x00a3));
1605         addAccents("\\textyen", getutf8(0x00a5));
1606         addAccents("\\yen", getutf8(0x00a5));
1607         addAccents("\\textsection", getutf8(0x00a7));
1608         addAccents("\\mathsection", getutf8(0x00a7));
1609         addAccents("\\textcopyright", getutf8(0x00a9));
1610         addAccents("\\copyright", getutf8(0x00a9));
1611         addAccents("\\textlnot", getutf8(0x00ac));
1612         addAccents("\\neg", getutf8(0x00ac));
1613         addAccents("\\textregistered", getutf8(0x00ae));
1614         addAccents("\\circledR", getutf8(0x00ae));
1615         addAccents("\\textpm", getutf8(0x00b1));
1616         addAccents("\\pm", getutf8(0x00b1));
1617         addAccents("\\textparagraph", getutf8(0x00b6));
1618         addAccents("\\mathparagraph", getutf8(0x00b6));
1619         addAccents("\\textperiodcentered", getutf8(0x00b7));
1620         addAccents("\\texttimes", getutf8(0x00d7));
1621         addAccents("\\times", getutf8(0x00d7));
1622         addAccents("\\O", getutf8(0x00d8));
1623         addAccents("\\dh", getutf8(0x00f0));
1624         addAccents("\\eth", getutf8(0x00f0));
1625         addAccents("\\textdiv", getutf8(0x00f7));
1626         addAccents("\\div", getutf8(0x00f7));
1627         addAccents("\\o", getutf8(0x00f8));
1628         addAccents("\\textcrlambda", getutf8(0x019b));
1629         addAccents("\\j", getutf8(0x0237));
1630         addAccents("\\textrevepsilon", getutf8(0x025c));
1631         addAccents("\\textbaru", getutf8(0x0289));
1632         addAccents("\\textquoteleft", getutf8(0x02bb));
1633         addAccents("\\textGamma", getutf8(0x0393));
1634         addAccents("\\Gamma", getutf8(0x0393));
1635         addAccents("\\textDelta", getutf8(0x0394));
1636         addAccents("\\Delta", getutf8(0x0394));
1637         addAccents("\\textTheta", getutf8(0x0398));
1638         addAccents("\\Theta", getutf8(0x0398));
1639         addAccents("\\textLambda", getutf8(0x039b));
1640         addAccents("\\Lambda", getutf8(0x039b));
1641         addAccents("\\textXi", getutf8(0x039e));
1642         addAccents("\\Xi", getutf8(0x039e));
1643         addAccents("\\textPi", getutf8(0x03a0));
1644         addAccents("\\Pi", getutf8(0x03a0));
1645         addAccents("\\textSigma", getutf8(0x03a3));
1646         addAccents("\\Sigma", getutf8(0x03a3));
1647         addAccents("\\textUpsilon", getutf8(0x03a5));
1648         addAccents("\\Upsilon", getutf8(0x03a5));
1649         addAccents("\\textPhi", getutf8(0x03a6));
1650         addAccents("\\Phi", getutf8(0x03a6));
1651         addAccents("\\textPsi", getutf8(0x03a8));
1652         addAccents("\\Psi", getutf8(0x03a8));
1653         addAccents("\\textOmega", getutf8(0x03a9));
1654         addAccents("\\Omega", getutf8(0x03a9));
1655         addAccents("\\textalpha", getutf8(0x03b1));
1656         addAccents("\\alpha", getutf8(0x03b1));
1657         addAccents("\\textbeta", getutf8(0x03b2));
1658         addAccents("\\beta", getutf8(0x03b2));
1659         addAccents("\\textgamma", getutf8(0x03b3));
1660         addAccents("\\gamma", getutf8(0x03b3));
1661         addAccents("\\textdelta", getutf8(0x03b4));
1662         addAccents("\\delta", getutf8(0x03b4));
1663         addAccents("\\textepsilon", getutf8(0x03b5));
1664         addAccents("\\varepsilon", getutf8(0x03b5));
1665         addAccents("\\textzeta", getutf8(0x03b6));
1666         addAccents("\\zeta", getutf8(0x03b6));
1667         addAccents("\\texteta", getutf8(0x03b7));
1668         addAccents("\\eta", getutf8(0x03b7));
1669         addAccents("\\texttheta", getutf8(0x03b8));
1670         addAccents("\\theta", getutf8(0x03b8));
1671         addAccents("\\textiota", getutf8(0x03b9));
1672         addAccents("\\iota", getutf8(0x03b9));
1673         addAccents("\\textkappa", getutf8(0x03ba));
1674         addAccents("\\kappa", getutf8(0x03ba));
1675         addAccents("\\textlambda", getutf8(0x03bb));
1676         addAccents("\\lambda", getutf8(0x03bb));
1677         addAccents("\\textmu", getutf8(0x03bc));
1678         addAccents("\\mu", getutf8(0x03bc));
1679         addAccents("\\textnu", getutf8(0x03bd));
1680         addAccents("\\nu", getutf8(0x03bd));
1681         addAccents("\\textxi", getutf8(0x03be));
1682         addAccents("\\xi", getutf8(0x03be));
1683         addAccents("\\textpi", getutf8(0x03c0));
1684         addAccents("\\pi", getutf8(0x03c0));
1685         addAccents("\\textrho", getutf8(0x03c1));
1686         addAccents("\\rho", getutf8(0x03c1));
1687         addAccents("\\textfinalsigma", getutf8(0x03c2));
1688         addAccents("\\varsigma", getutf8(0x03c2));
1689         addAccents("\\textsigma", getutf8(0x03c3));
1690         addAccents("\\sigma", getutf8(0x03c3));
1691         addAccents("\\texttau", getutf8(0x03c4));
1692         addAccents("\\tau", getutf8(0x03c4));
1693         addAccents("\\textupsilon", getutf8(0x03c5));
1694         addAccents("\\upsilon", getutf8(0x03c5));
1695         addAccents("\\textphi", getutf8(0x03c6));
1696         addAccents("\\varphi", getutf8(0x03c6));
1697         addAccents("\\textchi", getutf8(0x03c7));
1698         addAccents("\\chi", getutf8(0x03c7));
1699         addAccents("\\textpsi", getutf8(0x03c8));
1700         addAccents("\\psi", getutf8(0x03c8));
1701         addAccents("\\textomega", getutf8(0x03c9));
1702         addAccents("\\omega", getutf8(0x03c9));
1703         addAccents("\\textdigamma", getutf8(0x03dd));
1704         addAccents("\\digamma", getutf8(0x03dd));
1705         addAccents("\\hebalef", getutf8(0x05d0));
1706         addAccents("\\aleph", getutf8(0x05d0));
1707         addAccents("\\hebbet", getutf8(0x05d1));
1708         addAccents("\\beth", getutf8(0x05d1));
1709         addAccents("\\hebgimel", getutf8(0x05d2));
1710         addAccents("\\gimel", getutf8(0x05d2));
1711         addAccents("\\hebdalet", getutf8(0x05d3));
1712         addAccents("\\daleth", getutf8(0x05d3));
1713         addAccents("\\hebhe", getutf8(0x05d4));
1714         addAccents("\\hebvav", getutf8(0x05d5));
1715         addAccents("\\hebzayin", getutf8(0x05d6));
1716         addAccents("\\hebhet", getutf8(0x05d7));
1717         addAccents("\\hebtet", getutf8(0x05d8));
1718         addAccents("\\hebyod", getutf8(0x05d9));
1719         addAccents("\\hebfinalkaf", getutf8(0x05da));
1720         addAccents("\\hebkaf", getutf8(0x05db));
1721         addAccents("\\heblamed", getutf8(0x05dc));
1722         addAccents("\\hebfinalmem", getutf8(0x05dd));
1723         addAccents("\\hebmem", getutf8(0x05de));
1724         addAccents("\\hebfinalnun", getutf8(0x05df));
1725         addAccents("\\hebnun", getutf8(0x05e0));
1726         addAccents("\\hebsamekh", getutf8(0x05e1));
1727         addAccents("\\hebayin", getutf8(0x05e2));
1728         addAccents("\\hebfinalpe", getutf8(0x05e3));
1729         addAccents("\\hebpe", getutf8(0x05e4));
1730         addAccents("\\hebfinaltsadi", getutf8(0x05e5));
1731         addAccents("\\hebtsadi", getutf8(0x05e6));
1732         addAccents("\\hebqof", getutf8(0x05e7));
1733         addAccents("\\hebresh", getutf8(0x05e8));
1734         addAccents("\\hebshin", getutf8(0x05e9));
1735         addAccents("\\hebtav", getutf8(0x05ea));
1736
1737         // Thai characters
1738         addAccents("\\thaiKoKai", getutf8(0x0e01));
1739         addAccents("\\thaiKhoKhai", getutf8(0x0e02));
1740         addAccents("\\thaiKhoKhuat", getutf8(0x0e03));
1741         addAccents("\\thaiKhoKhwai", getutf8(0x0e04));
1742         addAccents("\\thaiKhoKhon", getutf8(0x0e05));
1743         addAccents("\\thaiKhoRakhang", getutf8(0x0e06));
1744         addAccents("\\thaiNgoNgu", getutf8(0x0e07));
1745         addAccents("\\thaiChoChan", getutf8(0x0e08));
1746         addAccents("\\thaiChoChing", getutf8(0x0e09));
1747         addAccents("\\thaiChoChang", getutf8(0x0e0a));
1748         addAccents("\\thaiSoSo", getutf8(0x0e0b));
1749         addAccents("\\thaiChoChoe", getutf8(0x0e0c));
1750         addAccents("\\thaiYoYing", getutf8(0x0e0d));
1751         addAccents("\\thaiDoChada", getutf8(0x0e0e));
1752         addAccents("\\thaiToPatak", getutf8(0x0e0f));
1753         addAccents("\\thaiThoThan", getutf8(0x0e10));
1754         addAccents("\\thaiThoNangmontho", getutf8(0x0e11));
1755         addAccents("\\thaiThoPhuthao", getutf8(0x0e12));
1756         addAccents("\\thaiNoNen", getutf8(0x0e13));
1757         addAccents("\\thaiDoDek", getutf8(0x0e14));
1758         addAccents("\\thaiToTao", getutf8(0x0e15));
1759         addAccents("\\thaiThoThung", getutf8(0x0e16));
1760         addAccents("\\thaiThoThahan", getutf8(0x0e17));
1761         addAccents("\\thaiThoThong", getutf8(0x0e18));
1762         addAccents("\\thaiNoNu", getutf8(0x0e19));
1763         addAccents("\\thaiBoBaimai", getutf8(0x0e1a));
1764         addAccents("\\thaiPoPla", getutf8(0x0e1b));
1765         addAccents("\\thaiPhoPhung", getutf8(0x0e1c));
1766         addAccents("\\thaiFoFa", getutf8(0x0e1d));
1767         addAccents("\\thaiPhoPhan", getutf8(0x0e1e));
1768         addAccents("\\thaiFoFan", getutf8(0x0e1f));
1769         addAccents("\\thaiPhoSamphao", getutf8(0x0e20));
1770         addAccents("\\thaiMoMa", getutf8(0x0e21));
1771         addAccents("\\thaiYoYak", getutf8(0x0e22));
1772         addAccents("\\thaiRoRua", getutf8(0x0e23));
1773         addAccents("\\thaiRu", getutf8(0x0e24));
1774         addAccents("\\thaiLoLing", getutf8(0x0e25));
1775         addAccents("\\thaiLu", getutf8(0x0e26));
1776         addAccents("\\thaiWoWaen", getutf8(0x0e27));
1777         addAccents("\\thaiSoSala", getutf8(0x0e28));
1778         addAccents("\\thaiSoRusi", getutf8(0x0e29));
1779         addAccents("\\thaiSoSua", getutf8(0x0e2a));
1780         addAccents("\\thaiHoHip", getutf8(0x0e2b));
1781         addAccents("\\thaiLoChula", getutf8(0x0e2c));
1782         addAccents("\\thaiOAng", getutf8(0x0e2d));
1783         addAccents("\\thaiHoNokhuk", getutf8(0x0e2e));
1784         addAccents("\\thaiPaiyannoi", getutf8(0x0e2f));
1785         addAccents("\\thaiSaraA", getutf8(0x0e30));
1786         addAccents("\\thaiMaiHanakat", getutf8(0x0e31));
1787         addAccents("\\thaiSaraAa", getutf8(0x0e32));
1788         addAccents("\\thaiSaraAm", getutf8(0x0e33));
1789         addAccents("\\thaiSaraI", getutf8(0x0e34));
1790         addAccents("\\thaiSaraIi", getutf8(0x0e35));
1791         addAccents("\\thaiSaraUe", getutf8(0x0e36));
1792         addAccents("\\thaiSaraUee", getutf8(0x0e37));
1793         addAccents("\\thaiSaraU", getutf8(0x0e38));
1794         addAccents("\\thaiSaraUu", getutf8(0x0e39));
1795         addAccents("\\thaiPhinthu", getutf8(0x0e3a));
1796         addAccents("\\thaiSaraE", getutf8(0x0e40));
1797         addAccents("\\thaiSaraAe", getutf8(0x0e41));
1798         addAccents("\\thaiSaraO", getutf8(0x0e42));
1799         addAccents("\\thaiSaraAiMaimuan", getutf8(0x0e43));
1800         addAccents("\\thaiSaraAiMaimalai", getutf8(0x0e44));
1801         addAccents("\\thaiLakkhangyao", getutf8(0x0e45));
1802         addAccents("\\thaiMaiyamok", getutf8(0x0e46));
1803         addAccents("\\thaiMaitaikhu", getutf8(0x0e47));
1804         addAccents("\\thaiMaiEk", getutf8(0x0e48));
1805         addAccents("\\thaiMaiTho", getutf8(0x0e49));
1806         addAccents("\\thaiMaiTri", getutf8(0x0e4a));
1807         addAccents("\\thaiMaiChattawa", getutf8(0x0e4b));
1808         addAccents("\\thaiThanthakhat", getutf8(0x0e4c));
1809         addAccents("\\thaiNikhahit", getutf8(0x0e4d));
1810         addAccents("\\thaiYamakkan", getutf8(0x0e4e));
1811         addAccents("\\thaiFongman", getutf8(0x0e4f));
1812         addAccents("\\thaizero", getutf8(0x0e50));
1813         addAccents("\\thaione", getutf8(0x0e51));
1814         addAccents("\\thaitwo", getutf8(0x0e52));
1815         addAccents("\\thaithree", getutf8(0x0e53));
1816         addAccents("\\thaifour", getutf8(0x0e54));
1817         addAccents("\\thaifive", getutf8(0x0e55));
1818         addAccents("\\thaisix", getutf8(0x0e56));
1819         addAccents("\\thaiseven", getutf8(0x0e57));
1820         addAccents("\\thaieight", getutf8(0x0e58));
1821         addAccents("\\thainine", getutf8(0x0e59));
1822         addAccents("\\thaiAngkhankhu", getutf8(0x0e5a));
1823         addAccents("\\thaiKhomut", getutf8(0x0e5b));
1824         addAccents("\\dag", getutf8(0x2020));
1825         addAccents("\\dagger", getutf8(0x2020));
1826         addAccents("\\textdagger", getutf8(0x2020));
1827         addAccents("\\ddag", getutf8(0x2021));
1828         addAccents("\\ddagger", getutf8(0x2021));
1829         addAccents("\\textdaggerdbl", getutf8(0x2021));
1830         addAccents("\\textbullet", getutf8(0x2022));
1831         addAccents("\\bullet", getutf8(0x2022));
1832         addAccents("\\dots", getutf8(0x2026));
1833         addAccents("\\ldots", getutf8(0x2026));
1834         addAccents("\\textellipsis", getutf8(0x2026));
1835         addAccents("\\textasciiacute", getutf8(0x2032));
1836         addAccents("\\prime", getutf8(0x2032));
1837         addAccents("\\textacutedbl", getutf8(0x2033));
1838         addAccents("\\dprime", getutf8(0x2033));
1839         addAccents("\\textasciigrave", getutf8(0x2035));
1840         addAccents("\\backprime", getutf8(0x2035));
1841         addAccents("\\textsubcircum{ }", getutf8(0x2038));
1842         addAccents("\\caretinsert", getutf8(0x2038));
1843         addAccents("\\textasteriskcentered", getutf8(0x204e));
1844         addAccents("\\ast", getutf8(0x204e));
1845         addAccents("\\textmho", getutf8(0x2127));
1846         addAccents("\\mho", getutf8(0x2127));
1847         addAccents("\\textleftarrow", getutf8(0x2190));
1848         addAccents("\\leftarrow", getutf8(0x2190));
1849         addAccents("\\textuparrow", getutf8(0x2191));
1850         addAccents("\\uparrow", getutf8(0x2191));
1851         addAccents("\\textrightarrow", getutf8(0x2192));
1852         addAccents("\\rightarrow", getutf8(0x2192));
1853         addAccents("\\textdownarrow", getutf8(0x2193));
1854         addAccents("\\downarrow", getutf8(0x2193));
1855         addAccents("\\textglobrise", getutf8(0x2197));
1856         addAccents("\\nearrow", getutf8(0x2197));
1857         addAccents("\\textglobfall", getutf8(0x2198));
1858         addAccents("\\searrow", getutf8(0x2198));
1859         addAccents("\\textsurd", getutf8(0x221a));
1860         addAccents("\\surd", getutf8(0x221a));
1861         addAccents("\\textbigcircle", getutf8(0x25ef));
1862         addAccents("\\bigcirc", getutf8(0x25ef));
1863         addAccents("\\FiveStar", getutf8(0x2605));
1864         addAccents("\\bigstar", getutf8(0x2605));
1865         addAccents("\\FiveStarOpen", getutf8(0x2606));
1866         addAccents("\\bigwhitestar", getutf8(0x2606));
1867         addAccents("\\Checkmark", getutf8(0x2713));
1868         addAccents("\\checkmark", getutf8(0x2713));
1869         addAccents("\\CrossMaltese", getutf8(0x2720));
1870         addAccents("\\maltese", getutf8(0x2720));
1871         addAccents("\\textlangle", getutf8(0x27e8));
1872         addAccents("\\langle", getutf8(0x27e8));
1873         addAccents("\\textrangle", getutf8(0x27e9));
1874         addAccents("\\rangle", getutf8(0x27e9));
1875 }
1876
1877 static void buildAccentsMap()
1878 {
1879         accents["imath"] = "ı";
1880         accents["i"] = "ı";
1881         accents["jmath"] = "ȷ";
1882         accents["cdot"] = "·";
1883         accents["textasciicircum"] = "^";
1884         accents["mathcircumflex"] = "^";
1885         accents["sim"] = "~";
1886         accents["guillemotright"] = "»";
1887         accents["guillemotleft"] = "«";
1888         accents["hairspace"]     = getutf8(0xf0000);    // select from free unicode plane 15
1889         accents["thinspace"]     = getutf8(0xf0002);    // and used _only_ by findadv
1890         accents["negthinspace"]  = getutf8(0xf0003);    // to omit backslashed latex macros
1891         accents["medspace"]      = getutf8(0xf0004);    // See https://en.wikipedia.org/wiki/Private_Use_Areas
1892         accents["negmedspace"]   = getutf8(0xf0005);
1893         accents["thickspace"]    = getutf8(0xf0006);
1894         accents["negthickspace"] = getutf8(0xf0007);
1895         accents["lyx"]           = getutf8(0xf0010);    // Used logos
1896         accents["LyX"]           = getutf8(0xf0010);
1897         accents["tex"]           = getutf8(0xf0011);
1898         accents["TeX"]           = getutf8(0xf0011);
1899         accents["latex"]         = getutf8(0xf0012);
1900         accents["LaTeX"]         = getutf8(0xf0012);
1901         accents["latexe"]        = getutf8(0xf0013);
1902         accents["LaTeXe"]        = getutf8(0xf0013);
1903         accents["lyxarrow"]      = getutf8(0xf0020);
1904         accents["braceleft"]     = getutf8(0xf0030);
1905         accents["braceright"]    = getutf8(0xf0031);
1906         accents["backslash lyx"]           = getutf8(0xf0010);  // Used logos inserted with starting \backslash
1907         accents["backslash LyX"]           = getutf8(0xf0010);
1908         accents["backslash tex"]           = getutf8(0xf0011);
1909         accents["backslash TeX"]           = getutf8(0xf0011);
1910         accents["backslash latex"]         = getutf8(0xf0012);
1911         accents["backslash LaTeX"]         = getutf8(0xf0012);
1912         accents["backslash latexe"]        = getutf8(0xf0013);
1913         accents["backslash LaTeXe"]        = getutf8(0xf0013);
1914         accents["backslash lyxarrow"]      = getutf8(0xf0020);
1915         accents["ddot{\\imath}"] = "ï";
1916         buildaccent("ddot", "aAeEhHiIoOtuUwWxXyY",
1917                     "äÄëËḧḦïÏöÖẗüÜẅẄẍẌÿŸ");   // umlaut
1918         buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1919                     "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ");      // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1920         accents["acute{\\imath}"] = "í";
1921         buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1922                     "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1923         buildaccent("dacute|H|h", "oOuU", "őŐűŰ");  // double acute
1924         buildaccent("mathring|r", "aAuUwy",
1925                     "åÅůŮẘẙ");  // ring
1926         accents["check{\\imath}"] = "ǐ";
1927         accents["check{\\jmath}"] = "ǰ";
1928         buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1929                     "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ");        // caron
1930         accents["hat{\\imath}"] = "î";
1931         accents["hat{\\jmath}"] = "ĵ";
1932         buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1933                     "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ");  // circ
1934         accents["bar{\\imath}"] = "ī";
1935         buildaccent("bar|=", "aAeEiIoOuUyY",
1936                     "āĀēĒīĪōŌūŪȳȲ");        // macron
1937         accents["tilde{\\imath}"] = "ĩ";
1938         buildaccent("tilde", "aAeEiInNoOuUvVyY",
1939                     "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ");  // tilde
1940         accents["breve{\\imath}"] = "ĭ";
1941         buildaccent("breve|u", "aAeEgGiIoOuU",
1942                     "ăĂĕĔğĞĭĬŏŎŭŬ");        // breve
1943         accents["grave{\\imath}"] = "ì";
1944         buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1945                     "àÀèÈìÌòÒùÙǹǸẁẀỳỲ");    // grave
1946         buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1947                     "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ");      // dot below
1948         buildaccent("ogonek|k", "AaEeIiUuOo",
1949                     "ĄąĘęĮįŲųǪǫ");    // ogonek
1950         buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1951                     "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ");        // cedilla
1952         buildaccent("subring|textsubring", "Aa",
1953                     "Ḁḁ");  // subring
1954         buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1955                     "ḒḓḘḙḼḽṊṋṰṱṶṷ");    // subcircum
1956         buildaccent("subtilde|textsubtilde", "EeIiUu",
1957                     "ḚḛḬḭṴṵ");      // subtilde
1958         accents["dgrave{\\imath}"] = "ȉ";
1959         accents["textdoublegrave{\\i}"] = "ȉ";
1960         buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1961                     "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1962         accents["rcap{\\imath}"] = "ȋ";
1963         accents["textroundcap{\\i}"] = "ȋ";
1964         buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1965                     "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1966         buildaccent("slashed", "oO",
1967                     "øØ"); // slashed
1968         fillMissingUnicodesymbols(); // Add some still not handled entries contained in 'unicodesynbols'
1969         // LYXERR0("Number of accents " << accents.size());
1970 }
1971
1972 /*
1973  * Created accents in math or regexp environment
1974  * are macros, but we need the utf8 equivalent
1975  */
1976 void Intervall::removeAccents()
1977 {
1978         if (accents.empty())
1979                 buildAccentsMap();
1980         static regex const accre("\\\\("
1981                                  "([\\S]|[A-Za-z]+)\\{[^\\{\\}]+\\}"
1982                                  "|("
1983                                  "(backslash ([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))"
1984                                  "|[A-Za-z]+"
1985                                  ")"
1986                                  "(?![a-zA-Z]))");
1987         smatch sub;
1988         for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1989                 sub = *itacc;
1990                 string key = sub.str(1);
1991                 AccentsIterator it_ac = accents.find(key);
1992                 if (it_ac != accents.end()) {
1993                         string val = it_ac->second;
1994                         size_t pos = sub.position(size_t(0));
1995                         for (size_t i = 0; i < val.size(); i++) {
1996                                 par[pos+i] = val[i];
1997                         }
1998                         // Remove possibly following space too
1999                         if (par[pos+sub.str(0).size()] == ' ')
2000                                 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
2001                         else
2002                                 addIntervall(pos+val.size(), pos + sub.str(0).size());
2003                         for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
2004                                 // remove traces of any remaining chars
2005                                 par[i] = ' ';
2006                         }
2007                 }
2008                 else {
2009                         LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
2010                 }
2011         }
2012 }
2013
2014 void Intervall::handleOpenP(int i)
2015 {
2016         actualdeptindex++;
2017         depts[actualdeptindex] = i+1;
2018         closes[actualdeptindex] = -1;
2019         checkDepthIndex(actualdeptindex);
2020 }
2021
2022 void Intervall::handleCloseP(int i, bool closingAllowed)
2023 {
2024         if (actualdeptindex <= 0) {
2025                 if (! closingAllowed)
2026                         LYXERR(Debug::FIND, "Bad closing parenthesis in latex");  /* should not happen, but the latex input may be wrong */
2027                 // if we are at the very end
2028                 addIntervall(i, i+1);
2029         }
2030         else {
2031                 closes[actualdeptindex] = i+1;
2032                 actualdeptindex--;
2033         }
2034 }
2035
2036 void Intervall::resetOpenedP(int openPos)
2037 {
2038         // Used as initializer for foreignlanguage entry
2039         actualdeptindex = 1;
2040         depts[1] = openPos+1;
2041         closes[1] = -1;
2042 }
2043
2044 int Intervall::previousNotIgnored(int start) const
2045 {
2046         int idx = 0;                          /* int intervalls */
2047         for (idx = ignoreidx; idx >= 0; --idx) {
2048                 if (start > borders[idx].upper)
2049                         return start;
2050                 if (start >= borders[idx].low)
2051                         start = borders[idx].low-1;
2052         }
2053         return start;
2054 }
2055
2056 int Intervall::nextNotIgnored(int start) const
2057 {
2058         int idx = 0;                          /* int intervalls */
2059         for (idx = 0; idx <= ignoreidx; idx++) {
2060                 if (start < borders[idx].low)
2061                         return start;
2062                 if (start < borders[idx].upper)
2063                         start = borders[idx].upper;
2064         }
2065         return start;
2066 }
2067
2068 typedef unordered_map<string, KeyInfo> KeysMap;
2069 typedef unordered_map<string, KeyInfo>::const_iterator KeysIterator;
2070 typedef vector< KeyInfo> Entries;
2071 static KeysMap keys = unordered_map<string, KeyInfo>();
2072
2073 class LatexInfo {
2074 private:
2075         int entidx_;
2076         Entries entries_;
2077         Intervall interval_;
2078         void buildKeys(bool);
2079         void buildEntries(bool);
2080         void makeKey(const string &, KeyInfo, bool isPatternString);
2081         void processRegion(int start, int region_end); /*  remove {} parts */
2082         void removeHead(KeyInfo const &, int count=0);
2083
2084 public:
2085         LatexInfo(string const & par, bool isPatternString)
2086                 : entidx_(-1), interval_(isPatternString, par)
2087         {
2088                 buildKeys(isPatternString);
2089                 entries_ = vector<KeyInfo>();
2090                 buildEntries(isPatternString);
2091         }
2092         int getFirstKey() {
2093                 entidx_ = 0;
2094                 if (entries_.empty()) {
2095                         return -1;
2096                 }
2097                 if (entries_[0].keytype == KeyInfo::isTitle) {
2098                         interval_.hasTitle = true;
2099                         if (! entries_[0].disabled) {
2100                                 interval_.titleValue = entries_[0].head;
2101                         }
2102                         else {
2103                                 interval_.titleValue = "";
2104                         }
2105                         removeHead(entries_[0]);
2106                         if (entries_.size() > 1)
2107                                 return 1;
2108                         else
2109                                 return -1;
2110                 }
2111                 return 0;
2112         }
2113         int getNextKey() {
2114                 entidx_++;
2115                 if (int(entries_.size()) > entidx_) {
2116                         return entidx_;
2117                 }
2118                 else {
2119                         return -1;
2120                 }
2121         }
2122         bool setNextKey(int idx) {
2123                 if ((idx == entidx_) && (entidx_ >= 0)) {
2124                         entidx_--;
2125                         return true;
2126                 }
2127                 else
2128                         return false;
2129         }
2130         int find(int start, KeyInfo::KeyType keytype) const {
2131                 if (start < 0)
2132                         return -1;
2133                 int tmpIdx = start;
2134                 while (tmpIdx < int(entries_.size())) {
2135                         if (entries_[tmpIdx].keytype == keytype)
2136                                 return tmpIdx;
2137                         tmpIdx++;
2138                 }
2139                 return -1;
2140         }
2141         int process(ostringstream & os, KeyInfo const & actual);
2142         int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
2143         // string show(int lastpos) { return interval.show(lastpos);}
2144         int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);}
2145         KeyInfo &getKeyInfo(int keyinfo) {
2146                 static KeyInfo invalidInfo = KeyInfo();
2147                 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
2148                         return invalidInfo;
2149                 else
2150                         return entries_[keyinfo];
2151         }
2152         void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);}
2153         void addIntervall(int low, int up) { interval_.addIntervall(low, up); }
2154 };
2155
2156
2157 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
2158 {
2159         int skip = 0;
2160         int depth = 0;
2161         for (int i = start; i < end; i += 1 + skip) {
2162                 char c;
2163                 c = par[i];
2164                 skip = 0;
2165                 if (c == '\\') skip = 1;
2166                 else if (c == up) {
2167                         depth++;
2168                 }
2169                 else if (c == down) {
2170                         if (depth == 0) {
2171                                 repeat--;
2172                                 if ((repeat <= 0) || (par[i+1] != up))
2173                                         return i;
2174                         }
2175                         --depth;
2176                 }
2177         }
2178         return end;
2179 }
2180
2181 class MathInfo {
2182         class MathEntry {
2183         public:
2184                 string wait;
2185                 size_t mathEnd;
2186                 size_t mathpostfixsize;
2187                 size_t mathStart;
2188                 size_t mathprefixsize;
2189                 size_t mathSize;
2190         };
2191         size_t actualIdx_;
2192         vector<MathEntry> entries_;
2193 public:
2194         MathInfo() {
2195                 actualIdx_ = 0;
2196         }
2197         void insert(string const & wait, size_t start, size_t prefixsize, size_t end, size_t postfixsize) {
2198                 MathEntry m = MathEntry();
2199                 m.wait = wait;
2200                 m.mathStart = start;
2201                 m.mathprefixsize = prefixsize;
2202                 m.mathEnd = end + postfixsize;
2203                 m.mathpostfixsize = postfixsize;
2204                 m.mathSize = m.mathEnd - m.mathStart;
2205                 entries_.push_back(m);
2206         }
2207         bool empty() const { return entries_.empty(); }
2208         size_t getEndPos() const {
2209                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2210                         return 0;
2211                 }
2212                 return entries_[actualIdx_].mathEnd;
2213         }
2214         size_t getStartPos() const {
2215                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2216                         return 100000;                    /*  definitely enough? */
2217                 }
2218                 return entries_[actualIdx_].mathStart;
2219         }
2220         size_t getPrefixSize() const {
2221                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2222                         return 0;
2223                 }
2224                 return entries_[actualIdx_].mathprefixsize;
2225         }
2226         size_t getPostfixSize() const {
2227                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2228                         return 0;
2229                 }
2230                 return entries_[actualIdx_].mathpostfixsize;
2231         }
2232         size_t getFirstPos() {
2233                 actualIdx_ = 0;
2234                 return getStartPos();
2235         }
2236         size_t getSize() const {
2237                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2238                         return size_t(0);
2239                 }
2240                 return entries_[actualIdx_].mathSize;
2241         }
2242         void incrEntry() { actualIdx_++; }
2243 };
2244
2245 void LatexInfo::buildEntries(bool isPatternString)
2246 {
2247         static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|x?x?alignat)\\*?\\})(\\{[0-9]+\\})?)");
2248         static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
2249         static bool disableLanguageOverride = false;
2250         smatch sub, submath;
2251         bool evaluatingRegexp = false;
2252         MathInfo mi;
2253         bool evaluatingMath = false;
2254         bool evaluatingCode = false;
2255         size_t codeEnd = 0;
2256         bool evaluatingOptional = false;
2257         size_t optionalEnd = 0;
2258         int codeStart = -1;
2259         KeyInfo found;
2260         bool math_end_waiting = false;
2261         size_t math_pos = 10000;
2262         size_t math_prefix_size = 1;
2263         string math_end;
2264         static vector<string> usedText = vector<string>();
2265         static bool removeMathHull = false;
2266
2267         interval_.removeAccents();
2268
2269         for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
2270                 submath = *itmath;
2271                 if ((submath.position(2) - submath.position(0)) %2 == 1) {
2272                         // prefixed by odd count of '\\'
2273                         continue;
2274                 }
2275                 if (math_end_waiting) {
2276                         size_t pos = submath.position(size_t(2));
2277                         if ((math_end == "$") &&
2278                                         (submath.str(2) == "$")) {
2279                                 mi.insert("$", math_pos, 1, pos, 1);
2280                                 math_end_waiting = false;
2281                         }
2282                         else if ((math_end == "\\]") &&
2283                                  (submath.str(2) == "\\]")) {
2284                                 mi.insert("\\]", math_pos, 2, pos, 2);
2285                                 math_end_waiting = false;
2286                         }
2287                         else if ((submath.str(3).compare("end") == 0) &&
2288                                  (submath.str(5).compare(math_end) == 0)) {
2289                                 mi.insert(math_end, math_pos, math_prefix_size, pos, submath.str(2).length());
2290                                 math_end_waiting = false;
2291                         }
2292                         else
2293                                 continue;
2294                 }
2295                 else {
2296                         if (submath.str(3).compare("begin") == 0) {
2297                                 math_end_waiting = true;
2298                                 math_end = submath.str(5);
2299                                 math_pos = submath.position(size_t(2));
2300                                 math_prefix_size = submath.str(2).length();
2301                         }
2302                         else if (submath.str(2).compare("\\[") == 0) {
2303                                 math_end_waiting = true;
2304                                 math_end = "\\]";
2305                                 math_pos = submath.position(size_t(2));
2306                         }
2307                         else if (submath.str(2) == "$") {
2308                                 size_t pos = submath.position(size_t(2));
2309                                 math_end_waiting = true;
2310                                 math_end = "$";
2311                                 math_pos = pos;
2312                         }
2313                 }
2314         }
2315         // Ignore language if there is math somewhere in pattern-string
2316         if (isPatternString) {
2317                 for (auto s: usedText) {
2318                         // Remove entries created in previous search runs
2319                         keys.erase(s);
2320                 }
2321                 usedText = vector<string>();
2322                 if (! mi.empty()) {
2323                         // Disable language
2324                         keys["foreignlanguage"].disabled = true;
2325                         disableLanguageOverride = true;
2326                         removeMathHull = false;
2327                 }
2328                 else {
2329                         removeMathHull = true;  // used later if not isPatternString
2330                         disableLanguageOverride = false;
2331                 }
2332         }
2333         else {
2334                 if (disableLanguageOverride) {
2335                         keys["foreignlanguage"].disabled = true;
2336                 }
2337         }
2338         math_pos = mi.getFirstPos();
2339         for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
2340                 sub = *it;
2341                 if ((sub.position(2) - sub.position(0)) %2 == 1) {
2342                         // prefixed by odd count of '\\'
2343                         continue;
2344                 }
2345                 string key = sub.str(5);
2346                 if (key == "") {
2347                         if (sub.str(2)[0] == '\\')
2348                                 key = sub.str(2)[1];
2349                         else {
2350                                 key = sub.str(2);
2351                         }
2352                 }
2353                 KeysIterator it_key = keys.find(key);
2354                 if (it_key != keys.end()) {
2355                         if (it_key->second.keytype == KeyInfo::headRemove) {
2356                                 KeyInfo found1 = it_key->second;
2357                                 found1.disabled = true;
2358                                 found1.head = "\\" + key + "{";
2359                                 found1._tokenstart = sub.position(size_t(2));
2360                                 found1._tokensize = found1.head.length();
2361                                 found1._dataStart = found1._tokenstart + found1.head.length();
2362                                 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
2363                                 found1._dataEnd = endpos;
2364                                 removeHead(found1);
2365                                 continue;
2366                         }
2367                 }
2368                 if (evaluatingRegexp) {
2369                         if (sub.str(3).compare("endregexp") == 0) {
2370                                 evaluatingRegexp = false;
2371                                 // found._tokenstart already set
2372                                 found._dataEnd = sub.position(size_t(2)) + 13;
2373                                 found._dataStart = found._dataEnd;
2374                                 found._tokensize = found._dataEnd - found._tokenstart;
2375                                 found.parenthesiscount = 0;
2376                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2377                         }
2378                         else {
2379                                 continue;
2380                         }
2381                 }
2382                 else {
2383                         if (evaluatingMath) {
2384                                 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
2385                                         continue;
2386                                 evaluatingMath = false;
2387                                 mi.incrEntry();
2388                                 math_pos = mi.getStartPos();
2389                         }
2390                         if (it_key == keys.end()) {
2391                                 found = KeyInfo(KeyInfo::isStandard, 0, true);
2392                                 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
2393                                 found = KeyInfo(KeyInfo::isText, 0, false);
2394                                 if (isPatternString) {
2395                                         found.keytype = KeyInfo::isChar;
2396                                         found.disabled = false;
2397                                         found.used = true;
2398                                 }
2399                                 keys[key] = found;
2400                                 usedText.push_back(key);
2401                         }
2402                         else
2403                                 found = keys[key];
2404                         if (key.compare("regexp") == 0) {
2405                                 evaluatingRegexp = true;
2406                                 found._tokenstart = sub.position(size_t(2));
2407                                 found._tokensize = 0;
2408                                 continue;
2409                         }
2410                 }
2411                 // Handle the other params of key
2412                 if (found.keytype == KeyInfo::isIgnored)
2413                         continue;
2414                 else if (found.keytype == KeyInfo::isMath) {
2415                         if (size_t(sub.position(size_t(2))) == math_pos) {
2416                                 found = keys[key];
2417                                 found._tokenstart = sub.position(size_t(2));
2418                                 found._tokensize = mi.getSize();
2419                                 found._dataEnd = found._tokenstart + found._tokensize;
2420                                 found._dataStart = found._dataEnd;
2421                                 found.parenthesiscount = 0;
2422                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2423                                 if (removeMathHull) {
2424                                         interval_.addIntervall(found._tokenstart, found._tokenstart + mi.getPrefixSize());
2425                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize(), found._dataEnd);
2426                                 }
2427                                 else {
2428                                         // Treate all math constructs as simple math
2429                                         interval_.par[found._tokenstart] = '$';
2430                                         interval_.par[found._dataEnd - mi.getPostfixSize()] = '$';
2431                                         interval_.addIntervall(found._tokenstart + 1, found._tokenstart + mi.getPrefixSize());
2432                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize() + 1, found._dataEnd);
2433                                 }
2434                                 evaluatingMath = true;
2435                         }
2436                         else {
2437                                 // begin|end of unknown env, discard
2438                                 // First handle tables
2439                                 // longtable|tabular
2440                                 bool discardComment;
2441                                 found = keys[key];
2442                                 found.keytype = KeyInfo::doRemove;
2443                                 if ((sub.str(7).compare("longtable") == 0) ||
2444                                                 (sub.str(7).compare("tabular") == 0)) {
2445                                         discardComment = true;        /* '%' */
2446                                 }
2447                                 else {
2448                                         discardComment = false;
2449                                         static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
2450                                         smatch sub2;
2451                                         string token = sub.str(7);
2452                                         if (regex_match(token, sub2, removeArgs)) {
2453                                                 found.keytype = KeyInfo::removeWithArg;
2454                                         }
2455                                 }
2456                                 // discard spaces before pos(2)
2457                                 int pos = sub.position(size_t(2));
2458                                 int count;
2459                                 for (count = 0; pos - count > 0; count++) {
2460                                         char c = interval_.par[pos-count-1];
2461                                         if (discardComment) {
2462                                                 if ((c != ' ') && (c != '%'))
2463                                                         break;
2464                                         }
2465                                         else if (c != ' ')
2466                                                 break;
2467                                 }
2468                                 found._tokenstart = pos - count;
2469                                 if (sub.str(3).compare(0, 5, "begin") == 0) {
2470                                         size_t pos1 = pos + sub.str(2).length();
2471                                         if (sub.str(7).compare("cjk") == 0) {
2472                                                 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2473                                                 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2474                                                         pos1 += 2;
2475                                                 found.keytype = KeyInfo::isMain;
2476                                                 found._dataStart = pos1;
2477                                                 found._dataEnd = interval_.par.length();
2478                                                 found.disabled = keys["foreignlanguage"].disabled;
2479                                                 found.used = keys["foreignlanguage"].used;
2480                                                 found._tokensize = pos1 - found._tokenstart;
2481                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2482                                         }
2483                                         else {
2484                                                 // Swallow possible optional params
2485                                                 while (interval_.par[pos1] == '[') {
2486                                                         pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2487                                                 }
2488                                                 // Swallow also the eventual parameter
2489                                                 if (interval_.par[pos1] == '{') {
2490                                                         found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2491                                                 }
2492                                                 else {
2493                                                         found._dataEnd = pos1;
2494                                                 }
2495                                                 found._dataStart = found._dataEnd;
2496                                                 found._tokensize = count + found._dataEnd - pos;
2497                                                 found.parenthesiscount = 0;
2498                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2499                                                 found.disabled = true;
2500                                         }
2501                                 }
2502                                 else {
2503                                         // Handle "\end{...}"
2504                                         found._dataStart = pos + sub.str(2).length();
2505                                         found._dataEnd = found._dataStart;
2506                                         found._tokensize = count + found._dataEnd - pos;
2507                                         found.parenthesiscount = 0;
2508                                         found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2509                                         found.disabled = true;
2510                                 }
2511                         }
2512                 }
2513                 else if (found.keytype != KeyInfo::isRegex) {
2514                         found._tokenstart = sub.position(size_t(2));
2515                         if (found.parenthesiscount == 0) {
2516                                 // Probably to be discarded
2517                                 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2518                                 char following = interval_.par[following_pos];
2519                                 if (following == ' ')
2520                                         found.head = "\\" + sub.str(5) + " ";
2521                                 else if (following == '=') {
2522                                         // like \uldepth=1000pt
2523                                         found.head = sub.str(2);
2524                                 }
2525                                 else
2526                                         found.head = "\\" + key;
2527                                 found._tokensize = found.head.length();
2528                                 found._dataEnd = found._tokenstart + found._tokensize;
2529                                 found._dataStart = found._dataEnd;
2530                         }
2531                         else {
2532                                 int params = found._tokenstart + key.length() + 1;
2533                                 if (evaluatingOptional) {
2534                                         if (size_t(found._tokenstart) > optionalEnd) {
2535                                                 evaluatingOptional = false;
2536                                         }
2537                                         else {
2538                                                 found.disabled = true;
2539                                         }
2540                                 }
2541                                 int optend = params;
2542                                 while (interval_.par[optend] == '[') {
2543                                         // discard optional parameters
2544                                         optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2545                                 }
2546                                 if (optend > params) {
2547                                         key += interval_.par.substr(params, optend-params);
2548                                         evaluatingOptional = true;
2549                                         optionalEnd = optend;
2550                                         if (found.keytype == KeyInfo::isSectioning) {
2551                                                 // Remove optional values (but still keep in header)
2552                                                 interval_.addIntervall(params, optend);
2553                                         }
2554                                 }
2555                                 string token = sub.str(7);
2556                                 int closings;
2557                                 if (interval_.par[optend] != '{') {
2558                                         closings = 0;
2559                                         found.parenthesiscount = 0;
2560                                         found.head = "\\" + key;
2561                                 }
2562                                 else
2563                                         closings = found.parenthesiscount;
2564                                 if (found.parenthesiscount == 1) {
2565                                         found.head = "\\" + key + "{";
2566                                 }
2567                                 else if (found.parenthesiscount > 1) {
2568                                         if (token != "") {
2569                                                 found.head = sub.str(2) + "{";
2570                                                 closings = found.parenthesiscount - 1;
2571                                         }
2572                                         else {
2573                                                 found.head = "\\" + key + "{";
2574                                         }
2575                                 }
2576                                 found._tokensize = found.head.length();
2577                                 found._dataStart = found._tokenstart + found.head.length();
2578                                 if (found.keytype == KeyInfo::doRemove) {
2579                                         if (closings > 0) {
2580                                                 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2581                                                 if (endpar >= interval_.par.length())
2582                                                         found._dataStart = interval_.par.length();
2583                                                 else
2584                                                         found._dataStart = endpar;
2585                                                 found._tokensize = found._dataStart - found._tokenstart;
2586                                         }
2587                                         else {
2588                                                 found._dataStart = found._tokenstart + found._tokensize;
2589                                         }
2590                                         closings = 0;
2591                                 }
2592                                 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2593                                         found._dataStart += 15;
2594                                 }
2595                                 size_t endpos;
2596                                 if (closings < 1)
2597                                         endpos = found._dataStart - 1;
2598                                 else
2599                                         endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2600                                 if (found.keytype == KeyInfo::isList) {
2601                                         // Check if it really is list env
2602                                         static regex const listre("^([a-z]+)$");
2603                                         smatch sub2;
2604                                         if (!regex_match(token, sub2, listre)) {
2605                                                 // Change the key of this entry. It is not in a list/item environment
2606                                                 found.keytype = KeyInfo::endArguments;
2607                                         }
2608                                 }
2609                                 if (found.keytype == KeyInfo::noMain) {
2610                                         evaluatingCode = true;
2611                                         codeEnd = endpos;
2612                                         codeStart = found._dataStart;
2613                                 }
2614                                 else if (evaluatingCode) {
2615                                         if (size_t(found._dataStart) > codeEnd)
2616                                                 evaluatingCode = false;
2617                                         else if (found.keytype == KeyInfo::isMain) {
2618                                                 // Disable this key, treate it as standard
2619                                                 found.keytype = KeyInfo::isStandard;
2620                                                 found.disabled = true;
2621                                                 if ((codeEnd +1 >= interval_.par.length()) &&
2622                                                                 (found._tokenstart == codeStart)) {
2623                                                         // trickery, because the code inset starts
2624                                                         // with \selectlanguage ...
2625                                                         codeEnd = endpos;
2626                                                         if (entries_.size() > 1) {
2627                                                                 entries_[entries_.size()-1]._dataEnd = codeEnd;
2628                                                         }
2629                                                 }
2630                                         }
2631                                 }
2632                                 if ((endpos == interval_.par.length()) &&
2633                                                 (found.keytype == KeyInfo::doRemove)) {
2634                                         // Missing closing => error in latex-input?
2635                                         // therefore do not delete remaining data
2636                                         found._dataStart -= 1;
2637                                         found._dataEnd = found._dataStart;
2638                                 }
2639                                 else
2640                                         found._dataEnd = endpos;
2641                         }
2642                         if (isPatternString) {
2643                                 keys[key].used = true;
2644                         }
2645                 }
2646                 entries_.push_back(found);
2647         }
2648 }
2649
2650 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2651 {
2652         stringstream s(keysstring);
2653         string key;
2654         const char delim = '|';
2655         while (getline(s, key, delim)) {
2656                 KeyInfo keyII(keyI);
2657                 if (isPatternString) {
2658                         keyII.used = false;
2659                 }
2660                 else if ( !keys[key].used)
2661                         keyII.disabled = true;
2662                 keys[key] = keyII;
2663         }
2664 }
2665
2666 void LatexInfo::buildKeys(bool isPatternString)
2667 {
2668
2669         static bool keysBuilt = false;
2670         if (keysBuilt && !isPatternString) return;
2671
2672         // Keys to ignore in any case
2673         makeKey("text|textcyrillic|lyxmathsym|ensuremath", KeyInfo(KeyInfo::headRemove, 1, true), true);
2674         // Known standard keys with 1 parameter.
2675         // Split is done, if not at start of region
2676         makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2677         makeKey("textbf",               KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2678         makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2679         makeKey("uuline|uline|uwave",   KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2680         makeKey("emph|noun",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2681         makeKey("sout|xout",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2682
2683         makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2684                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2685         makeKey("section*|subsection*|subsubsection*|paragraph*",
2686                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2687         makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2688         makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2689         // Regex
2690         makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2691
2692         // Split is done, if not at start of region
2693         makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2694         makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2695
2696         // Split is done always.
2697         makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2698
2699         // Known charaters
2700         // No split
2701         makeKey("backslash|textbackslash|slash",  KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2702         makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2703         makeKey("textasciiacute|texemdash",       KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2704         makeKey("dots|ldots",                     KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2705         // Spaces
2706         makeKey("quad|qquad|hfill|dotfill",               KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2707         makeKey("textvisiblespace|nobreakspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2708         makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2709         makeKey("thickspace|medspace|thinspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2710         // Skip
2711         // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2712         // Custom space/skip, remove the content (== length value)
2713         makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2714         // Found in fr/UserGuide.lyx
2715         makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2716         // quotes
2717         makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2718         makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2719         // Known macros to remove (including their parameter)
2720         // No split
2721         makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2722         makeKey("addtocounter|setlength",                 KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2723         // handle like standard keys with 1 parameter.
2724         makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2725
2726         // Ignore deleted text
2727         makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2728         // but preserve added text
2729         makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2730
2731         // Macros to remove, but let the parameter survive
2732         // No split
2733         makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2734
2735         // Remove language spec from content of these insets
2736         makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2737
2738         // Same effect as previous, parameter will survive (because there is no one anyway)
2739         // No split
2740         makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2741         // Remove table decorations
2742         makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2743         // Discard shape-header.
2744         // For footnote or shortcut too, because of lang settings
2745         // and wrong handling if used 'KeyInfo::noMain'
2746         makeKey("circlepar|diamondpar|heartpar|nutpar",  KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2747         makeKey("trianglerightpar|hexagonpar|starpar",   KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2748         makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2749         makeKey("triangleleftpar|shapepar|dropuppar",    KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2750         makeKey("hphantom|vphantom|note|footnote|shortcut|include|includegraphics",     KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2751         makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2752         // like ('tiny{}' or '\tiny ' ... )
2753         makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2754
2755         // Survives, like known character
2756         // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2757         makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2758         makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2759
2760         makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2761         makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2762         makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2763
2764         makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2765         // Remove RTL/LTR marker
2766         makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2767         makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2768         makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2769         makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2770         makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2771         makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2772         makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2773         if (isPatternString) {
2774                 // Allow the first searched string to rebuild the keys too
2775                 keysBuilt = false;
2776         }
2777         else {
2778                 // no need to rebuild again
2779                 keysBuilt = true;
2780         }
2781 }
2782
2783 /*
2784  * Keep the list of actual opened parentheses actual
2785  * (e.g. depth == 4 means there are 4 '{' not processed yet)
2786  */
2787 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2788 {
2789         int skip = 0;
2790         for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2791                 char c;
2792                 c = par[i];
2793                 skip = 0;
2794                 if (c == '\\') skip = 1;
2795                 else if (c == '{') {
2796                         handleOpenP(i);
2797                 }
2798                 else if (c == '}') {
2799                         handleCloseP(i, closingAllowed);
2800                 }
2801         }
2802 }
2803
2804 #if (0)
2805 string Intervall::show(int lastpos)
2806 {
2807         int idx = 0;                          /* int intervalls */
2808         string s;
2809         int i = 0;
2810         for (idx = 0; idx <= ignoreidx; idx++) {
2811                 while (i < lastpos) {
2812                         int printsize;
2813                         if (i <= borders[idx].low) {
2814                                 if (borders[idx].low > lastpos)
2815                                         printsize = lastpos - i;
2816                                 else
2817                                         printsize = borders[idx].low - i;
2818                                 s += par.substr(i, printsize);
2819                                 i += printsize;
2820                                 if (i >= borders[idx].low)
2821                                         i = borders[idx].upper;
2822                         }
2823                         else {
2824                                 i = borders[idx].upper;
2825                                 break;
2826                         }
2827                 }
2828         }
2829         if (lastpos > i) {
2830                 s += par.substr(i, lastpos-i);
2831         }
2832         return s;
2833 }
2834 #endif
2835
2836 void Intervall::output(ostringstream &os, int lastpos)
2837 {
2838         // get number of chars to output
2839         int idx = 0;                          /* int intervalls */
2840         int i = 0;
2841         int printed = 0;
2842         string startTitle = titleValue;
2843         for (idx = 0; idx <= ignoreidx; idx++) {
2844                 if (i < lastpos) {
2845                         if (i <= borders[idx].low) {
2846                                 int printsize;
2847                                 if (borders[idx].low > lastpos)
2848                                         printsize = lastpos - i;
2849                                 else
2850                                         printsize = borders[idx].low - i;
2851                                 if (printsize > 0) {
2852                                         os << startTitle << par.substr(i, printsize);
2853                                         i += printsize;
2854                                         printed += printsize;
2855                                         startTitle = "";
2856                                 }
2857                                 handleParentheses(i, false);
2858                                 if (i >= borders[idx].low)
2859                                         i = borders[idx].upper;
2860                         }
2861                         else {
2862                                 i = borders[idx].upper;
2863                         }
2864                 }
2865                 else
2866                         break;
2867         }
2868         if (lastpos > i) {
2869                 os << startTitle << par.substr(i, lastpos-i);
2870                 printed += lastpos-i;
2871         }
2872         handleParentheses(lastpos, false);
2873         int startindex;
2874         if (keys["foreignlanguage"].disabled)
2875                 startindex = actualdeptindex-langcount;
2876         else
2877                 startindex = actualdeptindex;
2878         for (int i = startindex; i > 0; --i) {
2879                 os << "}";
2880         }
2881         if (hasTitle && (printed > 0))
2882                 os << "}";
2883         if (! isPatternString_)
2884                 os << "\n";
2885         handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2886 }
2887
2888 void LatexInfo::processRegion(int start, int region_end)
2889 {
2890         while (start < region_end) {          /* Let {[} and {]} survive */
2891                 int cnt = interval_.isOpeningPar(start);
2892                 if (cnt == 1) {
2893                         // Closing is allowed past the region
2894                         int closing = interval_.findclosing(start+1, interval_.par.length());
2895                         interval_.addIntervall(start, start+1);
2896                         interval_.addIntervall(closing, closing+1);
2897                 }
2898                 else if (cnt == 3)
2899                         start += 2;
2900                 start = interval_.nextNotIgnored(start+1);
2901         }
2902 }
2903
2904 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2905 {
2906         if (actual.parenthesiscount == 0) {
2907                 // "{\tiny{} ...}" ==> "{{} ...}"
2908                 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2909         }
2910         else {
2911                 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2912                 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2913                 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2914         }
2915 }
2916
2917 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2918 {
2919         int nextKeyIdx = 0;
2920         switch (actual.keytype)
2921         {
2922         case KeyInfo::isTitle: {
2923                 removeHead(actual);
2924                 nextKeyIdx = getNextKey();
2925                 break;
2926         }
2927         case KeyInfo::cleanToStart: {
2928                 actual._dataEnd = actual._dataStart;
2929                 nextKeyIdx = getNextKey();
2930                 // Search for end of arguments
2931                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2932                 if (tmpIdx > 0) {
2933                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2934                                 entries_[i].disabled = true;
2935                         }
2936                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
2937                 }
2938                 while (interval_.par[actual._dataEnd] == ' ')
2939                         actual._dataEnd++;
2940                 interval_.addIntervall(0, actual._dataEnd+1);
2941                 interval_.actualdeptindex = 0;
2942                 interval_.depts[0] = actual._dataEnd+1;
2943                 interval_.closes[0] = -1;
2944                 break;
2945         }
2946         case KeyInfo::isText:
2947                 interval_.par[actual._tokenstart] = '#';
2948                 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2949                 nextKeyIdx = getNextKey();
2950                 break;
2951         case KeyInfo::noContent: {          /* char like "\hspace{2cm}" */
2952                 if (actual.disabled)
2953                         interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2954                 else
2955                         interval_.addIntervall(actual._dataStart, actual._dataEnd);
2956         }
2957                 // fall through
2958         case KeyInfo::isChar: {
2959                 nextKeyIdx = getNextKey();
2960                 break;
2961         }
2962         case KeyInfo::isSize: {
2963                 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2964                         if (actual.parenthesiscount == 0)
2965                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2966                         else {
2967                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2968                         }
2969                         nextKeyIdx = getNextKey();
2970                 } else {
2971                         // Here _dataStart points to '{', so correct it
2972                         actual._dataStart += 1;
2973                         actual._tokensize += 1;
2974                         actual.parenthesiscount = 1;
2975                         if (interval_.par[actual._dataStart] == '}') {
2976                                 // Determine the end if used like '{\tiny{}...}'
2977                                 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2978                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2979                         }
2980                         else {
2981                                 // Determine the end if used like '\tiny{...}'
2982                                 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2983                         }
2984                         // Split on this key if not at start
2985                         int start = interval_.nextNotIgnored(previousStart);
2986                         if (start < actual._tokenstart) {
2987                                 interval_.output(os, actual._tokenstart);
2988                                 interval_.addIntervall(start, actual._tokenstart);
2989                         }
2990                         // discard entry if at end of actual
2991                         nextKeyIdx = process(os, actual);
2992                 }
2993                 break;
2994         }
2995         case KeyInfo::endArguments: {
2996                 // Remove trailing '{}' too
2997                 actual._dataStart += 1;
2998                 actual._dataEnd += 1;
2999                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3000                 nextKeyIdx = getNextKey();
3001                 break;
3002         }
3003         case KeyInfo::noMain:
3004                 // fall through
3005         case KeyInfo::isStandard: {
3006                 if (actual.disabled) {
3007                         removeHead(actual);
3008                         processRegion(actual._dataStart, actual._dataStart+1);
3009                         nextKeyIdx = getNextKey();
3010                 } else {
3011                         // Split on this key if not at datastart of calling entry
3012                         int start = interval_.nextNotIgnored(previousStart);
3013                         if (start < actual._tokenstart) {
3014                                 interval_.output(os, actual._tokenstart);
3015                                 interval_.addIntervall(start, actual._tokenstart);
3016                         }
3017                         // discard entry if at end of actual
3018                         nextKeyIdx = process(os, actual);
3019                 }
3020                 break;
3021         }
3022         case KeyInfo::removeWithArg: {
3023                 nextKeyIdx = getNextKey();
3024                 // Search for end of arguments
3025                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3026                 if (tmpIdx > 0) {
3027                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3028                                 entries_[i].disabled = true;
3029                         }
3030                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3031                 }
3032                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3033                 break;
3034         }
3035         case KeyInfo::doRemove: {
3036                 // Remove the key with all parameters and following spaces
3037                 size_t pos;
3038                 size_t start;
3039                 if (interval_.par[actual._dataEnd-1] == ' ' || interval_.par[actual._dataEnd-1] == '}')
3040                         start = actual._dataEnd;
3041                 else
3042                         start = actual._dataEnd+1;
3043                 for (pos = start; pos < interval_.par.length(); pos++) {
3044                         if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
3045                                 break;
3046                 }
3047                 // Remove also enclosing parentheses [] and {}
3048                 int numpars = 0;
3049                 int spaces = 0;
3050                 while (actual._tokenstart > numpars) {
3051                         if (pos+numpars >= interval_.par.size())
3052                                 break;
3053                         else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
3054                                 numpars++;
3055                         else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
3056                                 numpars++;
3057                         else
3058                                 break;
3059                 }
3060                 if (numpars > 0) {
3061                         if (interval_.par[pos+numpars] == ' ')
3062                                 spaces++;
3063                 }
3064
3065                 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
3066                 nextKeyIdx = getNextKey();
3067                 break;
3068         }
3069         case KeyInfo::isList: {
3070                 // Discard space before _tokenstart
3071                 int count;
3072                 for (count = 0; count < actual._tokenstart; count++) {
3073                         if (interval_.par[actual._tokenstart-count-1] != ' ')
3074                                 break;
3075                 }
3076                 nextKeyIdx = getNextKey();
3077                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3078                 if (tmpIdx > 0) {
3079                         // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
3080                         // with arguments
3081                         // How else can we catch this one?
3082                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3083                                 entries_[i].disabled = true;
3084                         }
3085                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3086                 }
3087                 else if (nextKeyIdx > 0) {
3088                         // Ignore any lang entries inside data region
3089                         for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
3090                                 if (entries_[i].keytype == KeyInfo::isMain)
3091                                         entries_[i].disabled = true;
3092                         }
3093                 }
3094                 if (actual.disabled) {
3095                         interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
3096                 }
3097                 else {
3098                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3099                 }
3100                 if (interval_.par[actual._dataEnd+1] == '[') {
3101                         int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
3102                         if ((interval_.par[actual._dataEnd+2] == '{') &&
3103                                         (interval_.par[posdown-1] == '}')) {
3104                                 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
3105                                 interval_.addIntervall(posdown-1, posdown+1);
3106                         }
3107                         else {
3108                                 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
3109                                 interval_.addIntervall(posdown, posdown+1);
3110                         }
3111                         int blk = interval_.nextNotIgnored(actual._dataEnd+1);
3112                         if (blk > posdown) {
3113                                 // Discard at most 1 space after empty item
3114                                 int count;
3115                                 for (count = 0; count < 1; count++) {
3116                                         if (interval_.par[blk+count] != ' ')
3117                                                 break;
3118                                 }
3119                                 if (count > 0)
3120                                         interval_.addIntervall(blk, blk+count);
3121                         }
3122                 }
3123                 break;
3124         }
3125         case KeyInfo::isSectioning: {
3126                 // Discard spaces before _tokenstart
3127                 int count;
3128                 int val = actual._tokenstart;
3129                 for (count = 0; count < actual._tokenstart;) {
3130                         val = interval_.previousNotIgnored(val-1);
3131                         if (val < 0 || interval_.par[val] != ' ')
3132                                 break;
3133                         else {
3134                                 count = actual._tokenstart - val;
3135                         }
3136                 }
3137                 if (actual.disabled) {
3138                         removeHead(actual, count);
3139                         nextKeyIdx = getNextKey();
3140                 } else {
3141                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3142                         nextKeyIdx = process(os, actual);
3143                 }
3144                 break;
3145         }
3146         case KeyInfo::isMath: {
3147                 // Same as regex, use the content unchanged
3148                 nextKeyIdx = getNextKey();
3149                 break;
3150         }
3151         case KeyInfo::isRegex: {
3152                 // DO NOT SPLIT ON REGEX
3153                 // Do not disable
3154                 nextKeyIdx = getNextKey();
3155                 break;
3156         }
3157         case KeyInfo::isIgnored: {
3158                 // Treat like a character for now
3159                 nextKeyIdx = getNextKey();
3160                 break;
3161         }
3162         case KeyInfo::isMain: {
3163                 if (interval_.par.substr(actual._dataStart, 2) == "% ")
3164                         interval_.addIntervall(actual._dataStart, actual._dataStart+2);
3165                 if (actual._tokenstart > 0) {
3166                         int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
3167                         if ((prev >= 0) && interval_.par[prev] == '%')
3168                                 interval_.addIntervall(prev, prev+1);
3169                 }
3170                 if (actual.disabled) {
3171                         removeHead(actual);
3172                         interval_.langcount++;
3173                         if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
3174                                         (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
3175                                 // Discard also the space before math-equation
3176                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3177                         }
3178                         nextKeyIdx = getNextKey();
3179                         // interval.resetOpenedP(actual._dataStart-1);
3180                 }
3181                 else {
3182                         if (actual._tokenstart < 26) {
3183                                 // for the first (and maybe dummy) language
3184                                 interval_.setForDefaultLang(actual);
3185                         }
3186                         interval_.resetOpenedP(actual._dataStart-1);
3187                 }
3188                 break;
3189         }
3190         case KeyInfo::invalid:
3191         case KeyInfo::headRemove:
3192                 // These two cases cannot happen, already handled
3193                 // fall through
3194         default: {
3195                 // LYXERR(Debug::INFO, "Unhandled keytype");
3196                 nextKeyIdx = getNextKey();
3197                 break;
3198         }
3199         }
3200         return nextKeyIdx;
3201 }
3202
3203 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
3204 {
3205         int end = interval_.nextNotIgnored(actual._dataEnd);
3206         int oldStart = actual._dataStart;
3207         int nextKeyIdx = getNextKey();
3208         while (true) {
3209                 if ((nextKeyIdx < 0) ||
3210                                 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
3211                                 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
3212                         if (oldStart <= end) {
3213                                 processRegion(oldStart, end);
3214                                 oldStart = end+1;
3215                         }
3216                         break;
3217                 }
3218                 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
3219
3220                 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
3221                         (void) dispatch(os, actual._dataStart, nextKey);
3222                         end = nextKey._tokenstart;
3223                         break;
3224                 }
3225                 processRegion(oldStart, nextKey._tokenstart);
3226                 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
3227
3228                 oldStart = nextKey._dataEnd+1;
3229         }
3230         // now nextKey is either invalid or is outside of actual._dataEnd
3231         // output the remaining and discard myself
3232         if (oldStart <= end) {
3233                 processRegion(oldStart, end);
3234         }
3235         if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
3236                 end += 1;
3237                 // This is the normal case.
3238                 // But if using the firstlanguage, the closing may be missing
3239         }
3240         // get minimum of 'end' and  'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
3241         int output_end;
3242         if (actual._dataEnd < end)
3243                 output_end = interval_.nextNotIgnored(actual._dataEnd);
3244         else if (interval_.par.size() > (size_t) end)
3245                 output_end = interval_.nextNotIgnored(end);
3246         else
3247                 output_end = interval_.par.size();
3248         if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
3249                 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
3250         }
3251         // Remove possible empty data
3252         int dstart = interval_.nextNotIgnored(actual._dataStart);
3253         while (interval_.isOpeningPar(dstart) == 1) {
3254                 interval_.addIntervall(dstart, dstart+1);
3255                 int dend = interval_.findclosing(dstart+1, output_end);
3256                 interval_.addIntervall(dend, dend+1);
3257                 dstart = interval_.nextNotIgnored(dstart+1);
3258         }
3259         if (dstart < output_end)
3260                 interval_.output(os, output_end);
3261         if (nextKeyIdx < 0)
3262                 interval_.addIntervall(0, end);
3263         else
3264                 interval_.addIntervall(actual._tokenstart, end);
3265         return nextKeyIdx;
3266 }
3267
3268 string splitOnKnownMacros(string par, bool isPatternString)
3269 {
3270         ostringstream os;
3271         LatexInfo li(par, isPatternString);
3272         // LYXERR(Debug::INFO, "Berfore split: " << par);
3273         KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
3274         DummyKey.head = "";
3275         DummyKey._tokensize = 0;
3276         DummyKey._dataStart = 0;
3277         DummyKey._dataEnd = par.length();
3278         DummyKey.disabled = true;
3279         int firstkeyIdx = li.getFirstKey();
3280         string s;
3281         if (firstkeyIdx >= 0) {
3282                 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
3283                 DummyKey._tokenstart = firstKey._tokenstart;
3284                 int nextkeyIdx;
3285                 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
3286                         // Use dummy firstKey
3287                         firstKey = DummyKey;
3288                         (void) li.setNextKey(firstkeyIdx);
3289                 }
3290                 else {
3291                         if (par.substr(firstKey._dataStart, 2) == "% ")
3292                                 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
3293                 }
3294                 nextkeyIdx = li.process(os, firstKey);
3295                 while (nextkeyIdx >= 0) {
3296                         // Check for a possible gap between the last
3297                         // entry and this one
3298                         int datastart = li.nextNotIgnored(firstKey._dataStart);
3299                         KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
3300                         if ((nextKey._tokenstart > datastart)) {
3301                                 // Handle the gap
3302                                 firstKey._dataStart = datastart;
3303                                 firstKey._dataEnd = par.length();
3304                                 (void) li.setNextKey(nextkeyIdx);
3305                                 // Fake the last opened parenthesis
3306                                 li.setForDefaultLang(firstKey);
3307                                 nextkeyIdx = li.process(os, firstKey);
3308                         }
3309                         else {
3310                                 if (nextKey.keytype != KeyInfo::isMain) {
3311                                         firstKey._dataStart = datastart;
3312                                         firstKey._dataEnd = nextKey._dataEnd+1;
3313                                         (void) li.setNextKey(nextkeyIdx);
3314                                         li.setForDefaultLang(firstKey);
3315                                         nextkeyIdx = li.process(os, firstKey);
3316                                 }
3317                                 else {
3318                                         nextkeyIdx = li.process(os, nextKey);
3319                                 }
3320                         }
3321                 }
3322                 // Handle the remaining
3323                 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
3324                 firstKey._dataEnd = par.length();
3325                 // Check if ! empty
3326                 if ((firstKey._dataStart < firstKey._dataEnd) &&
3327                                 (par[firstKey._dataStart] != '}')) {
3328                         li.setForDefaultLang(firstKey);
3329                         (void) li.process(os, firstKey);
3330                 }
3331                 s = os.str();
3332                 // return string definitelly impossible to match, but should be known
3333         }
3334         else
3335                 s = par;                            /* no known macros found */
3336         // LYXERR(Debug::INFO, "After split: " << s);
3337         return s;
3338 }
3339
3340 /*
3341  * Try to unify the language specs in the latexified text.
3342  * Resulting modified string is set to "", if
3343  * the searched tex does not contain all the features in the search pattern
3344  */
3345 static string correctlanguagesetting(string par, bool isPatternString, bool withformat, lyx::Buffer *pbuf = nullptr)
3346 {
3347         static Features regex_f;
3348         static int missed = 0;
3349         static bool regex_with_format = false;
3350
3351         int parlen = par.length();
3352
3353         while ((parlen > 0) && (par[parlen-1] == '\n')) {
3354                 parlen--;
3355         }
3356         if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
3357                 // Happens to be there in case of description or labeling environment
3358                 parlen--;
3359         }
3360         string result;
3361         if (withformat) {
3362                 // Split the latex input into pieces which
3363                 // can be digested by our search engine
3364                 LYXERR(Debug::FIND, "input: \"" << par << "\"");
3365                 if (isPatternString && (pbuf != nullptr)) { // Check if we should disable/enable test for language
3366                         // We check for polyglossia, because in runparams.flavor we use Flavor::XeTeX
3367                         string doclang = pbuf->params().language->polyglossia();
3368                         static regex langre("\\\\(foreignlanguage)\\{([^\\}]+)\\}");
3369                         smatch sub;
3370                         bool toIgnoreLang = true;
3371                         for (sregex_iterator it(par.begin(), par.end(), langre), end; it != end; ++it) {
3372                                 sub = *it;
3373                                 if (sub.str(2) != doclang) {
3374                                         toIgnoreLang = false;
3375                                         break;
3376                                 }
3377                         }
3378                         setIgnoreFormat("language", toIgnoreLang, false);
3379
3380                 }
3381                 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
3382                 LYXERR(Debug::FIND, "After splitOnKnownMacros:\n\"" << result << "\"");
3383         }
3384         else
3385                 result = par.substr(0, parlen);
3386         if (isPatternString) {
3387                 missed = 0;
3388                 if (withformat) {
3389                         regex_f = identifyFeatures(result);
3390                         string features = "";
3391                         for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3392                                 string a = it->first;
3393                                 regex_with_format = true;
3394                                 features += " " + a;
3395                                 // LYXERR(Debug::INFO, "Identified regex format:" << a);
3396                         }
3397                         LYXERR(Debug::FIND, "Identified Features" << features);
3398
3399                 }
3400         } else if (regex_with_format) {
3401                 Features info = identifyFeatures(result);
3402                 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3403                         string a = it->first;
3404                         bool b = it->second;
3405                         if (b && ! info[a]) {
3406                                 missed++;
3407                                 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
3408                                 return "";
3409                         }
3410                 }
3411
3412         }
3413         else {
3414                 // LYXERR(Debug::INFO, "No regex formats");
3415         }
3416         return result;
3417 }
3418
3419
3420 // Remove trailing closure of math, macros and environments, so to catch parts of them.
3421 static int identifyClosing(string & t)
3422 {
3423         int open_braces = 0;
3424         do {
3425                 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
3426                 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
3427                         continue;
3428                 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
3429                         continue;
3430                 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
3431                         continue;
3432                 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
3433                         ++open_braces;
3434                         continue;
3435                 }
3436                 break;
3437         } while (true);
3438         return open_braces;
3439 }
3440
3441 static int num_replaced = 0;
3442 static bool previous_single_replace = true;
3443
3444 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
3445 {
3446 #if QTSEARCH
3447         if (regexp_str.empty() || regexp2_str.empty()) {
3448                 regexIsValid = false;
3449                 regexError = "Invalid empty regex";
3450                 return;
3451         }
3452         // Handle \w properly
3453         QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3454         if (! opt.casesensitive) {
3455                 popts |= QRegularExpression::CaseInsensitiveOption;
3456         }
3457         regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3458         regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3459         regexError = "";
3460         if (regexp.isValid() && regexp2.isValid()) {
3461                 regexIsValid = true;
3462                 // Check '{', '}' pairs inside the regex
3463                 int balanced = 0;
3464                 int skip = 1;
3465                 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3466                         char c = par_as_string[i];
3467                         if (c == '\\') {
3468                                 skip = 2;
3469                                 continue;
3470                         }
3471                         if (c == '{')
3472                                 balanced++;
3473                         else if (c == '}') {
3474                                 balanced--;
3475                                 if (balanced < 0)
3476                                         break;
3477                         }
3478                         skip = 1;
3479                 }
3480                 if (balanced != 0) {
3481                         regexIsValid = false;
3482                         regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3483                 }
3484         }
3485         else {
3486                 regexIsValid = false;
3487                 if (!regexp.isValid())
3488                         regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3489                 else
3490                         regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3491         }
3492 #else
3493         (void)par_as_string;
3494         if (opt.casesensitive) {
3495                 regexp = regex(regexp_str);
3496                 regexp2 = regex(regexp2_str);
3497         }
3498         else {
3499                 regexp = regex(regexp_str, std::regex_constants::icase);
3500                 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3501         }
3502 #endif
3503 }
3504
3505 static void modifyRegexForMatchWord(string &t)
3506 {
3507         string s("");
3508         regex wordre("(\\\\)*((\\.|\\\\b))");
3509         size_t lastpos = 0;
3510         smatch sub;
3511         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3512                 sub = *it;
3513                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3514                         continue;
3515                 }
3516                 else if (sub.str(2) == "\\\\b")
3517                         return;
3518                 if (lastpos < (size_t) sub.position(2))
3519                         s += t.substr(lastpos, sub.position(2) - lastpos);
3520                 s += "\\S";
3521                 lastpos = sub.position(2) + sub.length(2);
3522         }
3523         if (lastpos == 0) {
3524                 s = "\\b" + t + "\\b";
3525                 t = s;
3526                 return;
3527         }
3528         else if (lastpos < t.length())
3529                 s += t.substr(lastpos, t.length() - lastpos);
3530         t = "\\b" + s + "\\b";
3531 }
3532
3533 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3534         : p_buf(&buf), p_first_buf(&buf), opt(opt)
3535 {
3536         Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3537         docstring const & ds = stringifySearchBuffer(find_buf, opt);
3538         if (ds.empty() ) {
3539                 CreateRegexp(opt, "", "", "");
3540                 return;
3541         }
3542         use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3543         if (opt.replace_all && previous_single_replace) {
3544                 previous_single_replace = false;
3545                 num_replaced = 0;
3546         }
3547         else if (!opt.replace_all) {
3548                 num_replaced = 0;       // count number of replaced strings
3549                 previous_single_replace = true;
3550         }
3551         // When using regexp, braces are hacked already by escape_for_regex()
3552         par_as_string = normalize(ds, opt.ignoreformat);
3553         open_braces = 0;
3554         close_wildcards = 0;
3555
3556         size_t lead_size = 0;
3557         // correct the language settings
3558         par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat, &buf);
3559         if (par_as_string.empty()) {
3560                 CreateRegexp(opt, "", "", "");
3561                 return;
3562         }
3563         opt.matchAtStart = false;
3564         if (!use_regexp) {
3565                 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
3566                 if (opt.ignoreformat) {
3567                         lead_size = 0;
3568                 }
3569                 else {
3570                         lead_size = identifyLeading(par_as_string);
3571                 }
3572                 lead_as_string = par_as_string.substr(0, lead_size);
3573                 string lead_as_regex_string = string2regex(lead_as_string);
3574                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3575                 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3576                 /* Handle whole words too in this case
3577                 */
3578                 if (opt.matchword) {
3579                         par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3580                         opt.matchword = false;
3581                 }
3582                 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3583                 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3584                 CreateRegexp(opt, regexp_str, regexp2_str);
3585                 use_regexp = true;
3586                 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3587                 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3588                 return;
3589         }
3590
3591         if (!opt.ignoreformat) {
3592                 lead_size = identifyLeading(par_as_string);
3593                 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
3594                 lead_as_string = par_as_string.substr(0, lead_size);
3595                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3596         }
3597
3598         // Here we are using regexp
3599         LASSERT(use_regexp, /**/);
3600         {
3601                 string lead_as_regexp;
3602                 if (lead_size > 0) {
3603                         lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3604                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "\\$$", "");
3605                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3606                         par_as_string = par_as_string_nolead;
3607                         LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
3608                         LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3609                 }
3610                 // LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3611                 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3612                 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3613                 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3614                 ++close_wildcards;
3615                 size_t lng = par_as_string.size();
3616                 if (!opt.ignoreformat) {
3617                         // Remove extra '\}' at end if not part of \{\.\}
3618                         while(lng > 2) {
3619                                 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3620                                         if (lng >= 6) {
3621                                                 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3622                                                         break;
3623                                         }
3624                                         lng -= 2;
3625                                         open_braces++;
3626                                 }
3627                                 else
3628                                         break;
3629                         }
3630                         if (lng < par_as_string.size())
3631                                 par_as_string = par_as_string.substr(0,lng);
3632                 }
3633                 LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
3634                 if ((lng > 0) && (par_as_string[0] == '^')) {
3635                         par_as_string = par_as_string.substr(1);
3636                         --lng;
3637                         opt.matchAtStart = true;
3638                 }
3639                 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3640                 // LYXERR(Debug::FIND, "Open braces: " << open_braces);
3641                 // LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3642
3643                 // If entered regexp must match at begin of searched string buffer
3644                 // Kornel: Added parentheses to use $1 for size of the leading string
3645                 string regexp_str;
3646                 string regexp2_str;
3647                 {
3648                         // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3649                         // Unfortunately is '\1', '\2', etc not working for strings with extra format
3650                         // so the convert has no effect in that case
3651                         for (int i = 7; i > 0; --i) {
3652                                 string orig = "\\\\" + std::to_string(i);
3653                                 string dest = "\\" + std::to_string(i+2);
3654                                 while (regex_replace(par_as_string, par_as_string, orig, dest));
3655                         }
3656                         if (opt.matchword) {
3657                                 modifyRegexForMatchWord(par_as_string);
3658                                 opt.matchword = false;
3659                         }
3660                         regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3661                         regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3662                 }
3663                 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3664                 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3665                 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3666         }
3667 }
3668
3669 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3670 {
3671         MatchResult mres;
3672
3673         mres.searched_size = len;
3674         if (at_begin &&
3675                         (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3676                 return mres;
3677
3678         docstring docstr = stringifyFromForSearch(opt, cur, len);
3679         string str;
3680         str = normalize(docstr, opt.ignoreformat);
3681         if (!opt.ignoreformat) {
3682                 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3683                 // remove closing '}' and '\n' to allow for use of '$' in regex
3684                 size_t lng = str.size();
3685                 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3686                         lng--;
3687                 if (lng != str.size()) {
3688                         str = str.substr(0, lng);
3689                 }
3690         }
3691         if (str.empty()) {
3692                 mres.match_len = -1;
3693                 return mres;
3694         }
3695         LYXERR(Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
3696
3697         LASSERT(use_regexp, /**/);
3698         {
3699                 // use_regexp always true
3700                 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3701 #if QTSEARCH
3702                 QString qstr = QString::fromStdString(str);
3703                 QRegularExpression const *p_regexp;
3704                 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3705                 if (at_begin) {
3706                         p_regexp = &regexp;
3707                 } else {
3708                         p_regexp = &regexp2;
3709                 }
3710                 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3711                 if (!match.hasMatch())
3712                         return mres;
3713 #else
3714                 regex const *p_regexp;
3715                 regex_constants::match_flag_type flags;
3716                 if (at_begin) {
3717                         flags = regex_constants::match_continuous;
3718                         p_regexp = &regexp;
3719                 } else {
3720                         flags = regex_constants::match_default;
3721                         p_regexp = &regexp2;
3722                 }
3723                 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3724                 if (re_it == sregex_iterator())
3725                         return mres;
3726                 match_results<string::const_iterator> const & m = *re_it;
3727 #endif
3728                 // Whole found string, including the leading
3729                 // std: m[0].second - m[0].first
3730                 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3731                 //
3732                 // Size of the leading string
3733                 // std: m[1].second - m[1].first
3734                 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3735                 int leadingsize = 0;
3736 #if QTSEARCH
3737                 if (match.lastCapturedIndex() > 0) {
3738                         leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3739                 }
3740
3741 #else
3742                 if (m.size() > 2) {
3743                         leadingsize = m[1].second - m[1].first;
3744                 }
3745 #endif
3746 #if QTSEARCH
3747                 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3748                 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3749                 // because of different number of closing at end of string
3750                 // we have to 'unify' the length of the post-match.
3751                 // Done by ignoring closing parenthesis and linefeeds at string end
3752                 int matchend = match.capturedEnd(0);
3753                 size_t strsize = qstr.size();
3754                 if (!opt.ignoreformat) {
3755                         while (mres.match_len > 0) {
3756                                 QChar c = qstr.at(matchend - 1);
3757                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3758                                         mres.match_len--;
3759                                         matchend--;
3760                                 }
3761                                 else
3762                                         break;
3763                         }
3764                         while (strsize > (size_t) match.capturedEnd(0)) {
3765                                 QChar c = qstr.at(strsize-1);
3766                                 if ((c == '\n') || (c == '}')) {
3767                                         --strsize;
3768                                 }
3769                                 else
3770                                         break;
3771                         }
3772                 }
3773                 // LYXERR0(qstr.toStdString());
3774                 mres.match2end = strsize - matchend;
3775                 mres.pos = match.capturedStart(2);
3776 #else
3777                 mres.match_prefix = m[2].second - m[2].first;
3778                 mres.match_len = m[0].second - m[2].second;
3779                 // ignore closing parenthesis and linefeeds at string end
3780                 size_t strend = m[0].second - m[0].first;
3781                 int matchend = strend;
3782                 size_t strsize = str.size();
3783                 if (!opt.ignoreformat) {
3784                         while (mres.match_len > 0) {
3785                                 char c = str.at(matchend - 1);
3786                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3787                                         mres.match_len--;
3788                                         matchend--;
3789                                 }
3790                                 else
3791                                         break;
3792                         }
3793                         while (strsize > strend) {
3794                                 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3795                                         --strsize;
3796                                 }
3797                                 else
3798                                         break;
3799                         }
3800                 }
3801                 // LYXERR0(str);
3802                 mres.match2end = strsize - matchend;
3803                 mres.pos = m[2].first - m[0].first;;
3804 #endif
3805                 if (mres.match2end < 0)
3806                         mres.match_len = 0;
3807                 mres.leadsize = leadingsize;
3808 #if QTSEARCH
3809                 if (mres.match_len > 0) {
3810                         string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3811                         mres.result.push_back(a0);
3812                         for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3813                                 mres.result.push_back(match.captured(i).toStdString());
3814                         }
3815                 }
3816 #else
3817                 if (mres.match_len > 0) {
3818                         string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3819                         mres.result.push_back(a0);
3820                         for (size_t i = 3; i < m.size(); i++) {
3821                                 mres.result.push_back(m[i]);
3822                         }
3823                 }
3824 #endif
3825                 return mres;
3826         }
3827 }
3828
3829
3830 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3831 {
3832         MatchResult mres = findAux(cur, len, at_begin);
3833         int res = mres.match_len;
3834         LYXERR(Debug::FIND,
3835                "res=" << res << ", at_begin=" << at_begin
3836                << ", matchAtStart=" << opt.matchAtStart
3837                << ", inTexted=" << cur.inTexted());
3838         if (opt.matchAtStart) {
3839                 if (cur.pos() != 0)
3840                         mres.match_len = 0;
3841                 else if (mres.match_prefix > 0)
3842                         mres.match_len = 0;
3843                 return mres;
3844         }
3845         else
3846                 return mres;
3847 }
3848
3849 #if 0
3850 static bool simple_replace(string &t, string from, string to)
3851 {
3852         regex repl("(\\\\)*(" + from + ")");
3853         string s("");
3854         size_t lastpos = 0;
3855         smatch sub;
3856         for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3857                 sub = *it;
3858                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3859                         continue;
3860                 if (lastpos < (size_t) sub.position(2))
3861                         s += t.substr(lastpos, sub.position(2) - lastpos);
3862                 s += to;
3863                 lastpos = sub.position(2) + sub.length(2);
3864         }
3865         if (lastpos == 0)
3866                 return false;
3867         else if (lastpos < t.length())
3868                 s += t.substr(lastpos, t.length() - lastpos);
3869         t = s;
3870         return true;
3871 }
3872 #endif
3873
3874 string MatchStringAdv::normalize(docstring const & s, bool ignore_format) const
3875 {
3876         string t;
3877         t = lyx::to_utf8(s);
3878         // Remove \n at begin
3879         while (!t.empty() && t[0] == '\n')
3880                 t = t.substr(1);
3881         // Remove [%]*\n at end
3882         while (!t.empty() && t[t.size() - 1] == '\n') {
3883                 size_t count = 1;
3884                 if (!ignore_format) {
3885                         while ((t.size() > 1 + count) && (t[t.size() - 1 - count] == '%'))
3886                                 count++;
3887                 }
3888                 t = t.substr(0, t.size() - count);
3889         }
3890         size_t pos;
3891         // Handle all other '\n'
3892         while ((pos = t.find("\n")) != string::npos) {
3893                 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3894                         // Handle '\\\n'
3895                         if (isAlnumASCII(t[pos+1])) {
3896                                 t.replace(pos-2, 3, " ");
3897                         }
3898                         else {
3899                                 t.replace(pos-2, 3, "");
3900                         }
3901                 }
3902                 else {
3903                         if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3904                                 // '\n' adjacent to non-alpha-numerics, discard
3905                                 t.replace(pos, 1, "");
3906                         }
3907                         else {
3908                                 // Replace all other \n with spaces
3909                                 t.replace(pos, 1, " ");
3910                         }
3911                         if (!ignore_format) {
3912                                 size_t count = 0;
3913                                 while ((pos > count + 1) && (t[pos - 1 -count] == '%')) {
3914                                         count++;
3915                                 }
3916                                 if (count > 0) {
3917                                         t.replace(pos - count, count, "");
3918                                 }
3919                         }
3920                 }
3921         }
3922         // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3923         // Kornel: Added textsl, textsf, textit, texttt and noun
3924         // + allow to seach for colored text too
3925         LYXERR(Debug::FIND, "Removing stale empty macros from: " << t);
3926         while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3927                 LYXERR(Debug::FIND, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3928         while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3929                 LYXERR(Debug::FIND, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3930         while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3931
3932         return t;
3933 }
3934
3935
3936 docstring stringifyFromCursor(DocIterator const & cur, int len)
3937 {
3938         LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3939         if (cur.inTexted()) {
3940                 Paragraph const & par = cur.paragraph();
3941                 // TODO what about searching beyond/across paragraph breaks ?
3942                 // TODO Try adding a AS_STR_INSERTS as last arg
3943                 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3944                                         int(par.size()) : cur.pos() + len;
3945                 // OutputParams runparams(&cur.buffer()->params().encoding());
3946                 OutputParams runparams(encodings.fromLyXName("utf8"));
3947                 runparams.nice = true;
3948                 runparams.flavor = Flavor::XeTeX;
3949                 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3950                 // No side effect of file copying and image conversion
3951                 runparams.dryrun = true;
3952                 int option = AS_STR_INSETS | AS_STR_PLAINTEXT;
3953                 if (ignoreFormats.getDeleted()) {
3954                         option |= AS_STR_SKIPDELETE;
3955                         runparams.for_search = OutputParams::SearchWithoutDeleted;
3956                 }
3957                 else {
3958                         runparams.for_search = OutputParams::SearchWithDeleted;
3959                 }
3960                 if (ignoreFormats.getNonContent()) {
3961                         runparams.for_search |= OutputParams::SearchNonOutput;
3962                 }
3963                 LYXERR(Debug::FIND, "Stringifying with cur: "
3964                        << cur << ", from pos: " << cur.pos() << ", end: " << end);
3965                 return from_utf8(latexNamesToUtf8(par.asString(cur.pos(), end,
3966                                                                option,
3967                                                                &runparams)));
3968         } else if (cur.inMathed()) {
3969                 CursorSlice cs = cur.top();
3970                 MathData md = cs.cell();
3971                 MathData::const_iterator it_end =
3972                                 (( len == -1 || cs.pos() + len > int(md.size()))
3973                                  ? md.end()
3974                                  : md.begin() + cs.pos() + len );
3975                 MathData md2;
3976                 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
3977                         md2.push_back(*it);
3978                 docstring s = from_utf8(latexNamesToUtf8(asString(md2)));
3979                 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3980                 return s;
3981         }
3982         LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3983         return docstring();
3984 }
3985
3986 /** Computes the LaTeX export of buf starting from cur and ending len positions
3987  * after cur, if len is positive, or at the paragraph or innermost inset end
3988  * if len is -1.
3989  */
3990 docstring latexifyFromCursor(DocIterator const & cur, int len)
3991 {
3992         /*
3993         LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3994         LYXERR(Debug::FIND, "  with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3995                << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3996         */
3997         Buffer const & buf = *cur.buffer();
3998
3999         odocstringstream ods;
4000         otexstream os(ods);
4001         //OutputParams runparams(&buf.params().encoding());
4002         OutputParams runparams(encodings.fromLyXName("utf8"));
4003         runparams.nice = false;
4004         runparams.flavor = Flavor::XeTeX;
4005         runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4006         // No side effect of file copying and image conversion
4007         runparams.dryrun = true;
4008         if (ignoreFormats.getDeleted()) {
4009                 runparams.for_search = OutputParams::SearchWithoutDeleted;
4010         }
4011         else {
4012                 runparams.for_search = OutputParams::SearchWithDeleted;
4013         }
4014         if (ignoreFormats.getNonContent()) {
4015                 runparams.for_search |= OutputParams::SearchNonOutput;
4016         }
4017
4018         if (cur.inTexted()) {
4019                 // @TODO what about searching beyond/across paragraph breaks ?
4020                 pos_type endpos = cur.paragraph().size();
4021                 if (len != -1 && endpos > cur.pos() + len)
4022                         endpos = cur.pos() + len;
4023                 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
4024                           string(), cur.pos(), endpos);
4025                 string s = lyx::to_utf8(ods.str());
4026                 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
4027                 return(lyx::from_utf8(s));
4028         } else if (cur.inMathed()) {
4029                 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
4030                 for (int s = cur.depth() - 1; s >= 0; --s) {
4031                         CursorSlice const & cs = cur[s];
4032                         if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
4033                                 TeXMathStream ws(os);
4034                                 cs.asInsetMath()->asHullInset()->header_write(ws);
4035                                 break;
4036                         }
4037                 }
4038
4039                 CursorSlice const & cs = cur.top();
4040                 MathData md = cs.cell();
4041                 MathData::const_iterator it_end =
4042                                 ((len == -1 || cs.pos() + len > int(md.size()))
4043                                  ? md.end()
4044                                  : md.begin() + cs.pos() + len);
4045                 MathData md2;
4046                 for (MathData::const_iterator it = md.begin() + cs.pos();
4047                      it != it_end; ++it)
4048                         md2.push_back(*it);
4049
4050                 ods << asString(md2);
4051                 // Retrieve the math environment type, and add '$' or '$]'
4052                 // or others (\end{equation}) accordingly
4053                 for (int s = cur.depth() - 1; s >= 0; --s) {
4054                         CursorSlice const & cs2 = cur[s];
4055                         InsetMath * inset = cs2.asInsetMath();
4056                         if (inset && inset->asHullInset()) {
4057                                 TeXMathStream ws(os);
4058                                 inset->asHullInset()->footer_write(ws);
4059                                 break;
4060                         }
4061                 }
4062                 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
4063         } else {
4064                 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
4065         }
4066         return ods.str();
4067 }
4068
4069 #if defined(ResultsDebug)
4070 // Debugging output
4071 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
4072 {
4073         LYXERR0( "from:\t\t\t" << from);
4074         string status;
4075         if (mres.pos_len > 0) {
4076                 // Set in finalize
4077                 status = "FINALSEARCH";
4078         }
4079         else {
4080                 if (mres.match_len > 0) {
4081                         if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
4082                                 status = "Good Match";
4083                         else
4084                                 status = "Matched in";
4085                 }
4086                 else
4087                         status = "MissedSearch";
4088         }
4089
4090         LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
4091         if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
4092                 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
4093         if ((mres.pos > 0) || (mres.match_prefix > 0))
4094                 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
4095         for (size_t i = 0; i < mres.result.size(); i++)
4096                 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
4097 }
4098 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
4099 #else
4100 #define displayMres(s, txt, cur)
4101 #endif
4102
4103 /** Finalize an advanced find operation, advancing the cursor to the innermost
4104  ** position that matches, plus computing the length of the matching text to
4105  ** be selected
4106  ** Return the cur.pos() difference between start and end of found match
4107  **/
4108 MatchResult findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
4109 {
4110         // Search the foremost position that matches (avoids find of entire math
4111         // inset when match at start of it)
4112         DocIterator old_cur(cur.buffer());
4113         MatchResult mres;
4114         static MatchResult fail = MatchResult();
4115         MatchResult max_match;
4116         // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
4117         // Happens with e.g. hyperlinks
4118         // either one sees "http://www.bla.bla" or nothing
4119         // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
4120         // and although we search for only 3 chars, we find the whole hyperlink inset
4121         bool at_begin = (expected.match_prefix == 0);
4122         if (!match.opt.forward && match.opt.ignoreformat) {
4123                 if (expected.pos > 0)
4124                         return fail;
4125         }
4126         LASSERT(at_begin, /**/);
4127         if (expected.match_len > 0 && at_begin) {
4128                 // Search for deepest match
4129                 old_cur = cur;
4130                 max_match = expected;
4131                 do {
4132                         size_t d = cur.depth();
4133                         cur.forwardPos();
4134                         if (!cur)
4135                                 break;
4136                         if (cur.depth() < d)
4137                                 break;
4138                         if (cur.depth() == d)
4139                                 break;
4140                         size_t lastd = d;
4141                         while (cur && cur.depth() > lastd) {
4142                                 lastd = cur.depth();
4143                                 mres = match(cur, -1, at_begin);
4144                                 displayMres(mres, "Checking innermost", cur);
4145                                 if (mres.match_len > 0)
4146                                         break;
4147                                 // maybe deeper?
4148                                 cur.forwardPos();
4149                         }
4150                         if (mres.match_len < expected.match_len)
4151                                 break;
4152                         max_match = mres;
4153                         old_cur = cur;;
4154                 } while(1);
4155                 cur = old_cur;
4156         }
4157         else {
4158                 // (expected.match_len <= 0)
4159                 mres = match(cur);      /* match valid only if not searching whole words */
4160                 displayMres(mres, "Start with negative match", cur);
4161                 max_match = mres;
4162         }
4163         if (max_match.match_len <= 0) return fail;
4164         LYXERR(Debug::FIND, "Ok");
4165
4166         // Compute the match length
4167         int len = 1;
4168         if (cur.pos() + len > cur.lastpos())
4169                 return fail;
4170
4171         LASSERT(match.use_regexp, /**/);
4172         {
4173                 int minl = 1;
4174                 int maxl = cur.lastpos() - cur.pos();
4175                 // Greedy behaviour while matching regexps
4176                 while (maxl > minl) {
4177                         MatchResult mres2;
4178                         mres2 = match(cur, len, at_begin);
4179                         displayMres(mres2, "Finalize loop", cur);
4180                         int actual_match_len = mres2.match_len;
4181                         if (actual_match_len >= max_match.match_len) {
4182                                 // actual_match_len > max_match _can_ happen,
4183                                 // if the search area splits
4184                                 // some following word so that the regex
4185                                 // (e.g. 'r.*r\b' matches 'r' from the middle of the
4186                                 // splitted word)
4187                                 // This means, the len value is too big
4188                                 actual_match_len = max_match.match_len;
4189                                 max_match = mres2;
4190                                 max_match.match_len = actual_match_len;
4191                                 maxl = len;
4192                                 if (maxl - minl < 4)
4193                                         len = (maxl + minl)/2;
4194                                 else
4195                                         len = minl + (maxl - minl + 3)/4;
4196                         }
4197                         else {
4198                                 // (actual_match_len < max_match.match_len)
4199                                 minl = len + 1;
4200                                 len = (maxl + minl)/2;
4201                         }
4202                 }
4203                 len = minl;
4204                 old_cur = cur;
4205                 // Search for real start of matched characters
4206                 while (len > 1) {
4207                         MatchResult actual_match;
4208                         do {
4209                                 cur.forwardPos();
4210                         } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
4211                         if (cur.depth() < old_cur.depth()) {
4212                                 // Outer inset?
4213                                 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
4214                                 break;
4215                         }
4216                         if (cur.pos() != old_cur.pos()) {
4217                                 // OK, forwarded 1 pos in actual inset
4218                                 actual_match = match(cur, len-1, at_begin);
4219                                 if (actual_match.match_len == max_match.match_len) {
4220                                         // Ha, got it! The shorter selection has the same match length
4221                                         len--;
4222                                         old_cur = cur;
4223                                         max_match = actual_match;
4224                                 }
4225                                 else {
4226                                         // OK, the shorter selection matches less chars, revert to previous value
4227                                         cur = old_cur;
4228                                         break;
4229                                 }
4230                         }
4231                         else {
4232                                 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
4233                                 actual_match = match(cur, len, at_begin);
4234                                 if (actual_match.match_len == max_match.match_len) {
4235                                         old_cur = cur;
4236                                         max_match = actual_match;
4237                                 }
4238                         }
4239                 }
4240                 if (len == 0)
4241                         return fail;
4242                 else {
4243                         max_match.pos_len = len;
4244                         displayMres(max_match, "SEARCH RESULT", cur)
4245                                         return max_match;
4246                 }
4247         }
4248 }
4249
4250 /// Finds forward
4251 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
4252 {
4253         if (!cur)
4254                 return 0;
4255         bool repeat = false;
4256         DocIterator orig_cur;   // to be used if repeat not successful
4257         MatchResult orig_mres;
4258         while (!theApp()->longOperationCancelled() && cur) {
4259                 //(void) findAdvForwardInnermost(cur);
4260                 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
4261                 MatchResult mres = match(cur, -1, false);
4262                 string msg = "Starting";
4263                 if (repeat)
4264                         msg = "Repeated";
4265                 displayMres(mres, msg + " findForwardAdv", cur)
4266                                 int match_len = mres.match_len;
4267                 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
4268                         LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
4269                         match_len = 0;
4270                 }
4271                 if (match_len <= 0) {
4272                         // This should exit nested insets, if any, or otherwise undefine the currsor.
4273                         cur.pos() = cur.lastpos();
4274                         LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
4275                         cur.forwardPos();
4276                 }
4277                 else {  // match_len > 0
4278                         // Try to find the begin of searched string
4279                         int increment;
4280                         int firstInvalid = cur.lastpos() - cur.pos();
4281                         {
4282                                 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
4283                                 int incrcur = (firstInvalid + 1 )*3/4;
4284                                 if (incrcur < incrmatch)
4285                                         increment = incrcur;
4286                                 else
4287                                         increment = incrmatch;
4288                                 if (increment < 1)
4289                                         increment = 1;
4290                         }
4291                         LYXERR(Debug::FIND, "Set increment to " << increment);
4292                         while (increment > 0) {
4293                                 DocIterator old_cur = cur;
4294                                 if (cur.pos() + increment >= cur.lastpos()) {
4295                                         increment /= 2;
4296                                         continue;
4297                                 }
4298                                 cur.pos() = cur.pos() + increment;
4299                                 MatchResult mres2 = match(cur, -1, false);
4300                                 displayMres(mres2, "findForwardAdv loop", cur)
4301                                                 switch (interpretMatch(mres, mres2)) {
4302                                         case MatchResult::newIsTooFar:
4303                                                 // behind the expected match
4304                                                 firstInvalid = increment;
4305                                                 cur = old_cur;
4306                                                 increment /= 2;
4307                                                 break;
4308                                         case MatchResult::newIsBetter:
4309                                                 // not reached yet, but cur.pos()+increment is bettert
4310                                                 mres = mres2;
4311                                                 firstInvalid -= increment;
4312                                                 if (increment > firstInvalid*3/4)
4313                                                         increment = firstInvalid*3/4;
4314                                                 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
4315                                                         if (increment >= mres2.match_prefix)
4316                                                                 increment = (mres2.match_prefix+1)*3/4;
4317                                                 }
4318                                                 break;
4319                                         default:
4320                                                 // Todo@
4321                                                 // Handle not like MatchResult::newIsTooFar
4322                                                 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
4323                                                 firstInvalid--;
4324                                                 increment = increment*3/4;
4325                                                 cur = old_cur;
4326                                                 break;
4327                                 }
4328                         }
4329                         if (mres.match_len > 0) {
4330                                 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
4331                                         // The match seems to indicate some deeper level
4332                                         repeat = true;
4333                                         orig_cur = cur;
4334                                         orig_mres = mres;
4335                                         cur.forwardPos();
4336                                         continue;
4337                                 }
4338                         }
4339                         else if (repeat) {
4340                                 // should never be reached.
4341                                 cur = orig_cur;
4342                                 mres = orig_mres;
4343                         }
4344                         // LYXERR0("Leaving first loop");
4345                         LYXERR(Debug::FIND, "Finalizing 1");
4346                         MatchResult found_match = findAdvFinalize(cur, match, mres);
4347                         if (found_match.match_len > 0) {
4348                                 LASSERT(found_match.pos_len > 0, /**/);
4349                                 match.FillResults(found_match);
4350                                 return found_match.pos_len;
4351                         }
4352                         else {
4353                                 // try next possible match
4354                                 cur.forwardPos();
4355                                 repeat = false;
4356                                 continue;
4357                         }
4358                 }
4359         }
4360         return 0;
4361 }
4362
4363
4364 /// Find the most backward consecutive match within same paragraph while searching backwards.
4365 MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, MatchResult &expected)
4366 {
4367         DocIterator cur_begin = cur;
4368         cur_begin.pos() = 0;
4369         DocIterator tmp_cur = cur;
4370         MatchResult mr = findAdvFinalize(tmp_cur, match, expected);
4371         Inset & inset = cur.inset();
4372         for (; cur != cur_begin; cur.backwardPos()) {
4373                 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
4374                 DocIterator new_cur = cur;
4375                 new_cur.backwardPos();
4376                 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
4377                         break;
4378                 MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
4379                 if (new_mr.match_len == mr.match_len)
4380                         break;
4381                 mr = new_mr;
4382         }
4383         LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
4384         return mr;
4385 }
4386
4387
4388 /// Finds backwards
4389 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
4390 {
4391         if (! cur)
4392                 return 0;
4393         // Backup of original position
4394         DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4395         if (cur == cur_begin)
4396                 return 0;
4397         cur.backwardPos();
4398         DocIterator cur_orig(cur);
4399         bool pit_changed = false;
4400         do {
4401                 cur.pos() = 0;
4402                 MatchResult found_match = match(cur, -1, false);
4403
4404                 if (found_match.match_len > 0) {
4405                         if (pit_changed)
4406                                 cur.pos() = cur.lastpos();
4407                         else
4408                                 cur.pos() = cur_orig.pos();
4409                         LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
4410                         DocIterator cur_prev_iter;
4411                         do {
4412                                 found_match = match(cur);
4413                                 LYXERR(Debug::FIND, "findBackAdv3: found_match="
4414                                        << (found_match.match_len > 0) << ", cur: " << cur);
4415                                 if (found_match.match_len > 0) {
4416                                         MatchResult found_mr = findMostBackwards(cur, match, found_match);
4417                                         if (found_mr.pos_len > 0) {
4418                                                 match.FillResults(found_mr);
4419                                                 return found_mr.pos_len;
4420                                         }
4421                                 }
4422
4423                                 // Stop if begin of document reached
4424                                 if (cur == cur_begin)
4425                                         break;
4426                                 cur_prev_iter = cur;
4427                                 cur.backwardPos();
4428                         } while (true);
4429                 }
4430                 if (cur == cur_begin)
4431                         break;
4432                 if (cur.pit() > 0)
4433                         --cur.pit();
4434                 else
4435                         cur.backwardPos();
4436                 pit_changed = true;
4437         } while (!theApp()->longOperationCancelled());
4438         return 0;
4439 }
4440
4441
4442 } // namespace
4443
4444
4445 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4446                                  DocIterator const & cur, int len)
4447 {
4448         if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4449                 return docstring();
4450         if (!opt.ignoreformat)
4451                 return latexifyFromCursor(cur, len);
4452         else
4453                 return stringifyFromCursor(cur, len);
4454 }
4455
4456
4457 FindAndReplaceOptions::FindAndReplaceOptions(
4458                 docstring const & _find_buf_name, bool _casesensitive,
4459                 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4460                 docstring const & _repl_buf_name, bool _keep_case,
4461                 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4462         : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4463           forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4464           repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4465 {
4466 }
4467
4468
4469 namespace {
4470
4471
4472 /** Check if 'len' letters following cursor are all non-lowercase */
4473 static bool allNonLowercase(Cursor const & cur, int len)
4474 {
4475         pos_type beg_pos = cur.selectionBegin().pos();
4476         pos_type end_pos = cur.selectionBegin().pos() + len;
4477         if (len > cur.lastpos() + 1 - beg_pos) {
4478                 LYXERR(Debug::FIND, "This should not happen, more debug needed");
4479                 len = cur.lastpos() + 1 - beg_pos;
4480                 end_pos = beg_pos + len;
4481         }
4482         for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4483                 if (isLowerCase(cur.paragraph().getChar(pos)))
4484                         return false;
4485         return true;
4486 }
4487
4488
4489 /** Check if first letter is upper case and second one is lower case */
4490 static bool firstUppercase(Cursor const & cur)
4491 {
4492         char_type ch1, ch2;
4493         pos_type pos = cur.selectionBegin().pos();
4494         if (pos >= cur.lastpos() - 1) {
4495                 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
4496                 return false;
4497         }
4498         ch1 = cur.paragraph().getChar(pos);
4499         ch2 = cur.paragraph().getChar(pos + 1);
4500         bool result = isUpperCase(ch1) && isLowerCase(ch2);
4501         LYXERR(Debug::FIND, "firstUppercase(): "
4502                << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4503                << ch2 << "(" << char(ch2) << ")"
4504                << ", result=" << result << ", cur=" << cur);
4505         return result;
4506 }
4507
4508
4509 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4510  **
4511  ** \fixme What to do with possible further paragraphs in replace buffer ?
4512  **/
4513 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4514 {
4515         ParagraphList::iterator pit = buffer.paragraphs().begin();
4516         LASSERT(!pit->empty(), /**/);
4517         pos_type right = pos_type(1);
4518         pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4519         right = pit->size();
4520         pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4521 }
4522 } // namespace
4523
4524 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4525 {
4526         // Should replace the string "$" + std::to_string(matchnum) with replacement
4527         // if the char '$' is not prefixed with odd number of char '\\'
4528         static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4529         string s;
4530         size_t lastpos = 0;
4531         smatch sub;
4532         for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4533                 sub = *it;
4534                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4535                         continue;
4536                 int num = stoi(sub.str(3), nullptr, 10);
4537                 if (num >= maxmatchnum)
4538                         continue;
4539                 if (lastpos < (size_t) sub.position(2))
4540                         s += t.substr(lastpos, sub.position(2) - lastpos);
4541                 s += replacements[num];
4542                 lastpos = sub.position(2) + sub.length(2);
4543         }
4544         if (lastpos == 0)
4545                 return false;
4546         else if (lastpos < t.length())
4547                 s += t.substr(lastpos, t.length() - lastpos);
4548         t = s;
4549         return true;
4550 }
4551
4552 ///
4553 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4554 {
4555         Cursor & cur = bv->cursor();
4556         if (opt.repl_buf_name.empty()
4557                         || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4558                         || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4559                 return 0;
4560
4561         DocIterator sel_beg = cur.selectionBegin();
4562         DocIterator sel_end = cur.selectionEnd();
4563         if (&sel_beg.inset() != &sel_end.inset()
4564                         || sel_beg.pit() != sel_end.pit()
4565                         || sel_beg.idx() != sel_end.idx())
4566                 return 0;
4567         int sel_len = sel_end.pos() - sel_beg.pos();
4568         LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4569                << ", sel_len: " << sel_len << endl);
4570         if (sel_len == 0)
4571                 return 0;
4572         LASSERT(sel_len > 0, return 0);
4573
4574         if (!matchAdv(sel_beg, sel_len).match_len)
4575                 return 0;
4576
4577         // Build a copy of the replace buffer, adapted to the KeepCase option
4578         Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4579         ostringstream oss;
4580         repl_buffer_orig.write(oss);
4581         string lyx = oss.str();
4582         if (matchAdv.valid_matches > 0)
4583                 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4584         Buffer repl_buffer(string(), false);
4585         repl_buffer.setInternal(true);
4586         repl_buffer.setUnnamed(true);
4587         LASSERT(repl_buffer.readString(lyx), return 0);
4588         if (opt.keep_case && sel_len >= 2) {
4589                 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4590                 if (cur.inTexted()) {
4591                         if (firstUppercase(cur))
4592                                 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4593                         else if (allNonLowercase(cur, sel_len))
4594                                 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4595                 }
4596         }
4597         cap::cutSelection(cur, false);
4598         if (cur.inTexted()) {
4599                 repl_buffer.changeLanguage(
4600                                         repl_buffer.language(),
4601                                         cur.getFont().language());
4602                 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4603                 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4604                 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4605                                         repl_buffer.params().documentClassPtr(),
4606                                         repl_buffer.params().authors(),
4607                                         bv->buffer().errorList("Paste"));
4608                 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4609                 sel_len = repl_buffer.paragraphs().begin()->size();
4610         } else if (cur.inMathed()) {
4611                 odocstringstream ods;
4612                 otexstream os(ods);
4613                 // OutputParams runparams(&repl_buffer.params().encoding());
4614                 OutputParams runparams(encodings.fromLyXName("utf8"));
4615                 runparams.nice = false;
4616                 runparams.flavor = Flavor::XeTeX;
4617                 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4618                 runparams.dryrun = true;
4619                 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4620                 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4621                 docstring repl_latex = ods.str();
4622                 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4623                 string s;
4624                 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4625                 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4626                 repl_latex = from_utf8(s);
4627                 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4628                 MathData ar(cur.buffer());
4629                 asArray(repl_latex, ar, Parse::NORMAL);
4630                 cur.insert(ar);
4631                 sel_len = ar.size();
4632                 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4633         }
4634         if (cur.pos() >= sel_len)
4635                 cur.pos() -= sel_len;
4636         else
4637                 cur.pos() = 0;
4638         LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4639         bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4640         bv->processUpdateFlags(Update::Force);
4641         return 1;
4642 }
4643
4644
4645 /// Perform a FindAdv operation.
4646 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4647 {
4648         DocIterator cur;
4649         int pos_len = 0;
4650
4651         // e.g., when invoking word-findadv from mini-buffer wither with
4652         //       wrong options syntax or before ever opening advanced F&R pane
4653         if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4654                 return false;
4655
4656         try {
4657                 MatchStringAdv matchAdv(bv->buffer(), opt);
4658 #if QTSEARCH
4659                 if (!matchAdv.regexIsValid) {
4660                         bv->message(lyx::from_utf8(matchAdv.regexError));
4661                         return(false);
4662                 }
4663 #endif
4664                 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4665                 if (length > 0)
4666                         bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4667                 num_replaced += findAdvReplace(bv, opt, matchAdv);
4668                 cur = bv->cursor();
4669                 if (opt.forward)
4670                         pos_len = findForwardAdv(cur, matchAdv);
4671                 else
4672                         pos_len = findBackwardsAdv(cur, matchAdv);
4673         } catch (exception & ex) {
4674                 bv->message(from_utf8(ex.what()));
4675                 return false;
4676         }
4677
4678         if (pos_len == 0) {
4679                 if (num_replaced > 0) {
4680                         switch (num_replaced)
4681                         {
4682                         case 1:
4683                                 bv->message(_("One match has been replaced."));
4684                                 break;
4685                         case 2:
4686                                 bv->message(_("Two matches have been replaced."));
4687                                 break;
4688                         default:
4689                                 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4690                                 break;
4691                         }
4692                         num_replaced = 0;
4693                 }
4694                 else {
4695                         bv->message(_("Match not found."));
4696                 }
4697                 return false;
4698         }
4699
4700         if (num_replaced > 0)
4701                 bv->message(_("Match has been replaced."));
4702         else
4703                 bv->message(_("Match found."));
4704
4705         if (cur.pos() + pos_len > cur.lastpos()) {
4706                 // Prevent crash in bv->putSelectionAt()
4707                 // Should never happen, maybe LASSERT() here?
4708                 pos_len = cur.lastpos() - cur.pos();
4709         }
4710         LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4711         bv->putSelectionAt(cur, pos_len, !opt.forward);
4712
4713         return true;
4714 }
4715
4716
4717 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4718 {
4719         os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4720            << opt.casesensitive << ' '
4721            << opt.matchword << ' '
4722            << opt.forward << ' '
4723            << opt.expandmacros << ' '
4724            << opt.ignoreformat << ' '
4725            << opt.replace_all << ' '
4726            << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4727            << opt.keep_case << ' '
4728            << int(opt.scope) << ' '
4729            << int(opt.restr);
4730
4731         LYXERR(Debug::FIND, "built: " << os.str());
4732
4733         return os;
4734 }
4735
4736
4737 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4738 {
4739         // LYXERR(Debug::FIND, "parsing");
4740         string s;
4741         string line;
4742         getline(is, line);
4743         while (line != "EOSS") {
4744                 if (! s.empty())
4745                         s = s + "\n";
4746                 s = s + line;
4747                 if (is.eof())   // Tolerate malformed request
4748                         break;
4749                 getline(is, line);
4750         }
4751         // LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4752         opt.find_buf_name = from_utf8(s);
4753         is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4754         is.get();       // Waste space before replace string
4755         s = "";
4756         getline(is, line);
4757         while (line != "EOSS") {
4758                 if (! s.empty())
4759                         s = s + "\n";
4760                 s = s + line;
4761                 if (is.eof())   // Tolerate malformed request
4762                         break;
4763                 getline(is, line);
4764         }
4765         // LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4766         opt.repl_buf_name = from_utf8(s);
4767         is >> opt.keep_case;
4768         int i;
4769         is >> i;
4770         opt.scope = FindAndReplaceOptions::SearchScope(i);
4771         is >> i;
4772         opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4773
4774         /*
4775         LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4776                << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4777                << opt.scope << ' ' << opt.restr);
4778         */
4779         return is;
4780 }
4781
4782 } // namespace lyx