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 getSize() const { return ignoreSize_; }
  92         ///
  93         bool getUnderline() const { return ignoreUnderline_; }
  94         ///
  95         bool getMarkUp() const { return ignoreMarkUp_; }
  96         ///
  97         bool getStrikeOut() const { return ignoreStrikeOut_; }
  98         ///
  99         bool getSectioning() const { return ignoreSectioning_; }
 100         ///
 101         bool getFrontMatter() const { return ignoreFrontMatter_; }
 102         ///
 103         bool getColor() const { return ignoreColor_; }
 104         ///
 105         bool getLanguage() const { return ignoreLanguage_; }
 106         ///
 107         bool getDeleted() const { return ignoreDeleted_; }
 108         ///
 109         void setIgnoreDeleted(bool value);
 110         ///
 111         bool getNonContent() const { return searchNonContent_; }
 112         ///
 113         void setIgnoreFormat(string const & type, bool value, bool fromUser = true);
 114
 115 private:
 116         ///
 117         bool ignoreFamily_ = false;
 118         ///
 119         bool ignoreSeries_ = false;
 120         ///
 121         bool ignoreShape_ = false;
 122         ///
 123         bool ignoreSize_ = true;
 124         ///
 125         bool ignoreUnderline_ = false;
 126         ///
 127         bool ignoreMarkUp_ = false;
 128         ///
 129         bool ignoreStrikeOut_ = false;
 130         ///
 131         bool ignoreSectioning_ = false;
 132         ///
 133         bool ignoreFrontMatter_ = false;
 134         ///
 135         bool ignoreColor_ = false;
 136         ///
 137         bool ignoreLanguage_ = false;
 138         bool userSelectedIgnoreLanguage_ = false;
 139         ///
 140         bool ignoreDeleted_ = true;
 141         ///
 142         bool searchNonContent_ = true;
 143 };
 144
 145 void IgnoreFormats::setIgnoreFormat(string const & type, bool value, bool fromUser)
 146 {
 147         if (type == "color") {
 148                 ignoreColor_ = value;
 149         }
 150         else if (type == "language") {
 151                 if (fromUser) {
 152                         userSelectedIgnoreLanguage_ = value;
 153                         ignoreLanguage_ = value;
 154                 }
 155                 else
 156                         ignoreLanguage_ = (value || userSelectedIgnoreLanguage_);
 157         }
 158         else if (type == "sectioning") {
 159                 ignoreSectioning_ = value;
 160                 ignoreFrontMatter_ = value;
 161         }
 162         else if (type == "font") {
 163                 ignoreSeries_ = value;
 164                 ignoreShape_ = value;
 165                 ignoreFamily_ = value;
 166         }
 167         else if (type == "series") {
 168                 ignoreSeries_ = value;
 169         }
 170         else if (type == "shape") {
 171                 ignoreShape_ = value;
 172         }
 173         else if (type == "size") {
 174                 ignoreSize_ = value;
 175         }
 176         else if (type == "family") {
 177                 ignoreFamily_ = value;
 178         }
 179         else if (type == "markup") {
 180                 ignoreMarkUp_ = value;
 181         }
 182         else if (type == "underline") {
 183                 ignoreUnderline_ = value;
 184         }
 185         else if (type == "strike") {
 186                 ignoreStrikeOut_ = value;
 187         }
 188         else if (type == "deleted") {
 189                 ignoreDeleted_ = value;
 190         }
 191         else if (type == "non-output-content") {
 192                 searchNonContent_ = !value;
 193         }
 194 }
 195
 196 // The global variable that can be changed from outside
 197 IgnoreFormats ignoreFormats;
 198
 199
 200 void setIgnoreFormat(string const & type, bool value, bool fromUser)
 201 {
 202         ignoreFormats.setIgnoreFormat(type, value, fromUser);
 203 }
 204
 205
 206 namespace {
 207
 208 bool parse_bool(docstring & howto, bool const defvalue = false)
 209 {
 210         if (howto.empty())
 211                 return defvalue;
 212         docstring var;
 213         howto = split(howto, var, ' ');
 214         return var == "1";
 215 }
 216
 217
 218 class MatchString
 219 {
 220 public:
 221         MatchString(docstring const & s, bool cs, bool mw)
 222                 : str(s), case_sens(cs), whole_words(mw)
 223         {}
 224
 225         // returns true if the specified string is at the specified position
 226         // del specifies whether deleted strings in ct mode will be considered
 227         int operator()(Paragraph const & par, pos_type pos, bool del = true) const
 228         {
 229                 return par.find(str, case_sens, whole_words, pos, del);
 230         }
 231
 232 private:
 233         // search string
 234         docstring str;
 235         // case sensitive
 236         bool case_sens;
 237         // match whole words only
 238         bool whole_words;
 239 };
 240
 241
 242 int findForward(DocIterator & cur, DocIterator const endcur,
 243                 MatchString const & match,
 244                 bool find_del = true, bool onlysel = false)
 245 {
 246         for (; cur; cur.forwardChar()) {
 247                 if (onlysel && endcur.pit() == cur.pit()
 248                     && endcur.idx() == cur.idx() && endcur.pos() < cur.pos())
 249                         break;
 250                 if (cur.inTexted()) {
 251                         int len = match(cur.paragraph(), cur.pos(), find_del);
 252                         if (len > 0)
 253                                 return len;
 254                 }
 255         }
 256         return 0;
 257 }
 258
 259
 260 int findBackwards(DocIterator & cur, DocIterator const endcur,
 261                   MatchString const & match,
 262                   bool find_del = true, bool onlysel = false)
 263 {
 264         while (cur) {
 265                 cur.backwardChar();
 266                 if (onlysel && endcur.pit() == cur.pit()
 267                     && endcur.idx() == cur.idx() && endcur.pos() > cur.pos())
 268                         break;
 269                 if (cur.inTexted()) {
 270                         int len = match(cur.paragraph(), cur.pos(), find_del);
 271                         if (len > 0)
 272                                 return len;
 273                 }
 274         }
 275         return 0;
 276 }
 277
 278
 279 bool searchAllowed(docstring const & str)
 280 {
 281         if (str.empty()) {
 282                 frontend::Alert::error(_("Search error"), _("Search string is empty"));
 283                 return false;
 284         }
 285         return true;
 286 }
 287
 288 } // namespace
 289
 290
 291 bool findOne(BufferView * bv, docstring const & searchstr,
 292              bool case_sens, bool whole, bool forward,
 293              bool find_del, bool check_wrap, bool const auto_wrap,
 294              bool instant, bool onlysel)
 295 {
 296         // Clean up previous selections with empty searchstr on instant
 297         if (searchstr.empty() && instant) {
 298                 if (bv->cursor().selection()) {
 299                         bv->setCursor(bv->cursor().selectionBegin());
 300                         bv->clearSelection();
 301                 }
 302                 return true;
 303         }
 304
 305         if (!searchAllowed(searchstr))
 306                 return false;
 307
 308         DocIterator const endcur = forward ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin();
 309
 310         if (onlysel && bv->cursor().selection()) {
 311                 docstring const matchstring = bv->cursor().selectionAsString(false);
 312                 docstring const lcmatchsting = support::lowercase(matchstring);
 313                 if (matchstring == searchstr || (!case_sens && lcmatchsting == lowercase(searchstr))) {
 314                         docstring q = _("The search string matches the selection, and search is limited to selection.\n"
 315                                         "Continue search outside?");
 316                         int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
 317                                 q, 0, 1, _("&Yes"), _("&No"));
 318                         if (search_answer == 0) {
 319                                 bv->clearSelection();
 320                                 if (findOne(bv, searchstr, case_sens, whole, forward,
 321                                             find_del, check_wrap, auto_wrap, false, false))
 322                                         return true;
 323                         }
 324                         return false;
 325                 }
 326         }
 327
 328         DocIterator cur = forward
 329                 ? ((instant || onlysel) ? bv->cursor().selectionBegin() : bv->cursor().selectionEnd())
 330                 : ((instant || onlysel) ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin());
 331
 332         MatchString const match(searchstr, case_sens, whole);
 333
 334         int match_len = forward
 335                 ? findForward(cur, endcur, match, find_del, onlysel)
 336                 : findBackwards(cur, endcur, match, find_del, onlysel);
 337
 338         if (match_len > 0)
 339                 bv->putSelectionAt(cur, match_len, !forward);
 340         else if (onlysel) {
 341                 docstring q = _("The search string was not found within the selection.\n"
 342                                 "Continue search outside?");
 343                 int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
 344                         q, 0, 1, _("&Yes"), _("&No"));
 345                 if (search_answer == 0) {
 346                         bv->clearSelection();
 347                         if (findOne(bv, searchstr, case_sens, whole, forward,
 348                                     find_del, check_wrap, auto_wrap, false, false))
 349                                 return true;
 350                 }
 351                 return false;
 352         }
 353         else if (check_wrap) {
 354                 DocIterator cur_orig(bv->cursor());
 355                 bool wrap = auto_wrap;
 356                 if (!auto_wrap) {
 357                         docstring q;
 358                         if (forward)
 359                                 q = _("End of file reached while searching forward.\n"
 360                                   "Continue searching from the beginning?");
 361                         else
 362                                 q = _("Beginning of file reached while searching backward.\n"
 363                                   "Continue searching from the end?");
 364                         int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 365                                 q, 0, 1, _("&Yes"), _("&No"));
 366                         wrap = wrap_answer == 0;
 367                 }
 368                 if (wrap) {
 369                         if (forward) {
 370                                 bv->cursor().clear();
 371                                 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
 372                         } else {
 373                                 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
 374                                 bv->cursor().backwardPos();
 375                         }
 376                         if (auto_wrap) {
 377                                 docstring const msg = forward
 378                                   ? _("Search reached end of document, continuing from beginning.")
 379                                   : _("Search reached beginning of document, continuing from end.");
 380                                 bv->message(msg);
 381                         }
 382                         bv->clearSelection();
 383                         if (findOne(bv, searchstr, case_sens, whole, forward,
 384                                     find_del, false, false, false, false))
 385                                 return true;
 386                 }
 387                 bv->cursor().setCursor(cur_orig);
 388                 return false;
 389         }
 390
 391         return match_len > 0;
 392 }
 393
 394
 395 namespace {
 396
 397 int replaceAll(BufferView * bv,
 398                docstring const & searchstr, docstring const & replacestr,
 399                bool case_sens, bool whole, bool onlysel)
 400 {
 401         Buffer & buf = bv->buffer();
 402
 403         if (!searchAllowed(searchstr) || buf.isReadonly())
 404                 return 0;
 405
 406         DocIterator startcur = bv->cursor().selectionBegin();
 407         DocIterator endcur = bv->cursor().selectionEnd();
 408         bool const had_selection = bv->cursor().selection();
 409
 410         MatchString const match(searchstr, case_sens, whole);
 411         int num = 0;
 412
 413         int const rsize = replacestr.size();
 414         int const ssize = searchstr.size();
 415
 416         Cursor cur(*bv);
 417         cur.setCursor(doc_iterator_begin(&buf));
 418         int match_len = findForward(cur, endcur, match, false, onlysel);
 419         while (match_len > 0) {
 420                 // Backup current cursor position and font.
 421                 pos_type const pos = cur.pos();
 422                 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
 423                 cur.recordUndo();
 424                 int ct_deleted_text = ssize -
 425                         cur.paragraph().eraseChars(pos, pos + match_len,
 426                                                    buf.params().track_changes);
 427                 cur.paragraph().insert(pos, replacestr, font,
 428                                        Change(buf.params().track_changes
 429                                               ? Change::INSERTED
 430                                               : Change::UNCHANGED));
 431                 for (int i = 0; i < rsize + ct_deleted_text
 432                      && cur.pos() < cur.lastpos(); ++i)
 433                         cur.forwardPos();
 434                 if (onlysel && cur.pit() == endcur.pit() && cur.idx() == endcur.idx()) {
 435                         // Adjust end of selection for replace-all in selection
 436                         if (rsize > ssize) {
 437                                 int const offset = rsize - ssize;
 438                                 for (int i = 0; i < offset + ct_deleted_text
 439                                      && endcur.pos() < endcur.lastpos(); ++i)
 440                                         endcur.forwardPos();
 441                         } else {
 442                                 int const offset = ssize - rsize;
 443                                 for (int i = 0; i < offset && endcur.pos() > 0; ++i)
 444                                         endcur.backwardPos();
 445                                 for (int i = 0; i < ct_deleted_text
 446                                      && endcur.pos() < endcur.lastpos(); ++i)
 447                                         endcur.forwardPos();
 448                         }
 449                 }
 450                 ++num;
 451                 match_len = findForward(cur, endcur, match, false, onlysel);
 452         }
 453
 454         bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
 455
 456         startcur.fixIfBroken();
 457         bv->setCursor(startcur);
 458
 459         // Reset selection, accounting for changes in selection
 460         if (had_selection) {
 461                 endcur.fixIfBroken();
 462                 bv->cursor().resetAnchor();
 463                 bv->setCursorSelectionTo(endcur);
 464         }
 465
 466         return num;
 467 }
 468
 469
 470 // the idea here is that we are going to replace the string that
 471 // is selected IF it is the search string.
 472 // if there is a selection, but it is not the search string, then
 473 // we basically ignore it. (FIXME We ought to replace only within
 474 // the selection.)
 475 // if there is no selection, then:
 476 //  (i) if some search string has been provided, then we find it.
 477 //      (think of how the dialog works when you hit "replace" the
 478 //      first time.)
 479 // (ii) if no search string has been provided, then we treat the
 480 //      word the cursor is in as the search string. (why? i have no
 481 //      idea.) but this only works in text?
 482 //
 483 // returns the number of replacements made (one, if any) and
 484 // whether anything at all was done.
 485 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
 486                            docstring const & replacestr, bool case_sens,
 487                            bool whole, bool forward, bool findnext, bool wrap,
 488                            bool onlysel)
 489 {
 490         Cursor & cur = bv->cursor();
 491         if (!cur.selection() || onlysel) {
 492                 // no selection, non-empty search string: find it
 493                 if (!searchstr.empty()) {
 494                         bool const found = findOne(bv, searchstr, case_sens, whole,
 495                                                    forward, true, findnext, wrap, false, onlysel);
 496                         return make_pair(found, 0);
 497                 }
 498                 // empty search string
 499                 if (!cur.inTexted())
 500                         // bail in math
 501                         return make_pair(false, 0);
 502                 // select current word and treat it as the search string.
 503                 // This causes a minor bug as undo will restore this selection,
 504                 // which the user did not create (#8986).
 505                 cur.innerText()->selectWord(cur, WHOLE_WORD);
 506                 searchstr = cur.selectionAsString(false, true);
 507         }
 508
 509         // if we still don't have a search string, report the error
 510         // and abort.
 511         if (!searchAllowed(searchstr))
 512                 return make_pair(false, 0);
 513
 514         bool have_selection = cur.selection();
 515         docstring const selected = cur.selectionAsString(false, true);
 516         bool match =
 517                 case_sens
 518                 ? searchstr == selected
 519                 : compare_no_case(searchstr, selected) == 0;
 520
 521         // no selection or current selection is not search word:
 522         // just find the search word
 523         if (!have_selection || !match) {
 524                 bool const found = findOne(bv, searchstr, case_sens, whole, forward,
 525                                            true, findnext, wrap, false, onlysel);
 526                 return make_pair(found, 0);
 527         }
 528
 529         // we're now actually ready to replace. if the buffer is
 530         // read-only, we can't, though.
 531         if (bv->buffer().isReadonly())
 532                 return make_pair(false, 0);
 533
 534         cap::replaceSelectionWithString(cur, replacestr);
 535         if (forward) {
 536                 cur.pos() += replacestr.length();
 537                 LASSERT(cur.pos() <= cur.lastpos(),
 538                         cur.pos() = cur.lastpos());
 539         }
 540         if (findnext)
 541                 findOne(bv, searchstr, case_sens, whole,
 542                         forward, false, findnext, wrap, false, onlysel);
 543
 544         return make_pair(true, 1);
 545 }
 546
 547 } // namespace
 548
 549
 550 docstring const find2string(docstring const & search,
 551                             bool casesensitive, bool matchword,
 552                             bool forward, bool wrap, bool instant,
 553                             bool onlysel)
 554 {
 555         odocstringstream ss;
 556         ss << search << '\n'
 557            << int(casesensitive) << ' '
 558            << int(matchword) << ' '
 559            << int(forward) << ' '
 560            << int(wrap) << ' '
 561            << int(instant) << ' '
 562            << int(onlysel);
 563         return ss.str();
 564 }
 565
 566
 567 docstring const replace2string(docstring const & replace,
 568                                docstring const & search,
 569                                bool casesensitive, bool matchword,
 570                                bool all, bool forward, bool findnext,
 571                                bool wrap, bool onlysel)
 572 {
 573         odocstringstream ss;
 574         ss << replace << '\n'
 575            << search << '\n'
 576            << int(casesensitive) << ' '
 577            << int(matchword) << ' '
 578            << int(all) << ' '
 579            << int(forward) << ' '
 580            << int(findnext) << ' '
 581            << int(wrap) << ' '
 582            << int(onlysel);
 583         return ss.str();
 584 }
 585
 586
 587 docstring const string2find(docstring const & argument,
 588                               bool &casesensitive,
 589                               bool &matchword,
 590                               bool &forward,
 591                               bool &wrap,
 592                               bool &instant,
 593                               bool &onlysel)
 594 {
 595         // data is of the form
 596         // "<search>
 597         //  <casesensitive> <matchword> <forward> <wrap> <onlysel>"
 598         docstring search;
 599         docstring howto = split(argument, search, '\n');
 600
 601         casesensitive = parse_bool(howto);
 602         matchword     = parse_bool(howto);
 603         forward       = parse_bool(howto, true);
 604         wrap          = parse_bool(howto);
 605         instant       = parse_bool(howto);
 606         onlysel       = parse_bool(howto);
 607
 608         return search;
 609 }
 610
 611
 612 bool lyxfind(BufferView * bv, FuncRequest const & ev)
 613 {
 614         if (!bv || ev.action() != LFUN_WORD_FIND)
 615                 return false;
 616
 617         //lyxerr << "find called, cmd: " << ev << endl;
 618         bool casesensitive;
 619         bool matchword;
 620         bool forward;
 621         bool wrap;
 622         bool instant;
 623         bool onlysel;
 624
 625         docstring search = string2find(ev.argument(), casesensitive,
 626                                        matchword, forward, wrap, instant, onlysel);
 627
 628         return findOne(bv, search, casesensitive, matchword, forward,
 629                        false, true, wrap, instant, onlysel);
 630 }
 631
 632
 633 bool lyxreplace(BufferView * bv, FuncRequest const & ev)
 634 {
 635         if (!bv || ev.action() != LFUN_WORD_REPLACE)
 636                 return false;
 637
 638         // data is of the form
 639         // "<search>
 640         //  <replace>
 641         //  <casesensitive> <matchword> <all> <forward> <findnext> <wrap> <onlysel>"
 642         docstring search;
 643         docstring rplc;
 644         docstring howto = split(ev.argument(), rplc, '\n');
 645         howto = split(howto, search, '\n');
 646
 647         bool casesensitive = parse_bool(howto);
 648         bool matchword     = parse_bool(howto);
 649         bool all           = parse_bool(howto);
 650         bool forward       = parse_bool(howto, true);
 651         bool findnext      = parse_bool(howto, true);
 652         bool wrap          = parse_bool(howto);
 653         bool onlysel       = parse_bool(howto);
 654
 655         if (!bv->cursor().selection())
 656                 // only selection only makes sense with selection
 657                 onlysel = false;
 658
 659         bool update = false;
 660
 661         int replace_count = 0;
 662         if (all) {
 663                 replace_count = replaceAll(bv, search, rplc, casesensitive,
 664                                            matchword, onlysel);
 665                 update = replace_count > 0;
 666         } else {
 667                 pair<bool, int> rv =
 668                         replaceOne(bv, search, rplc, casesensitive, matchword,
 669                                    forward, findnext, wrap, onlysel);
 670                 update = rv.first;
 671                 replace_count = rv.second;
 672         }
 673
 674         Buffer const & buf = bv->buffer();
 675         if (!update) {
 676                 // emit message signal.
 677                 if (onlysel)
 678                         buf.message(_("String not found in selection."));
 679                 else
 680                         buf.message(_("String not found."));
 681         } else {
 682                 if (replace_count == 0) {
 683                         buf.message(_("String found."));
 684                 } else if (replace_count == 1) {
 685                         buf.message(_("String has been replaced."));
 686                 } else {
 687                         docstring const str = onlysel
 688                                         ? bformat(_("%1$d strings have been replaced in the selection."), replace_count)
 689                                         : bformat(_("%1$d strings have been replaced."), replace_count);
 690                         buf.message(str);
 691                 }
 692         }
 693         return update;
 694 }
 695
 696
 697 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
 698 {
 699         for (; cur; cur.forwardPos())
 700                 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
 701                         return true;
 702
 703         if (check_wrap) {
 704                 DocIterator cur_orig(bv->cursor());
 705                 docstring q = _("End of file reached while searching forward.\n"
 706                           "Continue searching from the beginning?");
 707                 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 708                         q, 0, 1, _("&Yes"), _("&No"));
 709                 if (wrap_answer == 0) {
 710                         bv->cursor().clear();
 711                         bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
 712                         bv->clearSelection();
 713                         cur.setCursor(bv->cursor().selectionBegin());
 714                         if (findNextChange(bv, cur, false))
 715                                 return true;
 716                 }
 717                 bv->cursor().setCursor(cur_orig);
 718         }
 719
 720         return false;
 721 }
 722
 723
 724 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
 725 {
 726         for (cur.backwardPos(); cur; cur.backwardPos()) {
 727                 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
 728                         return true;
 729         }
 730
 731         if (check_wrap) {
 732                 DocIterator cur_orig(bv->cursor());
 733                 docstring q = _("Beginning of file reached while searching backward.\n"
 734                           "Continue searching from the end?");
 735                 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 736                         q, 0, 1, _("&Yes"), _("&No"));
 737                 if (wrap_answer == 0) {
 738                         bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
 739                         bv->cursor().backwardPos();
 740                         bv->clearSelection();
 741                         cur.setCursor(bv->cursor().selectionBegin());
 742                         if (findPreviousChange(bv, cur, false))
 743                                 return true;
 744                 }
 745                 bv->cursor().setCursor(cur_orig);
 746         }
 747
 748         return false;
 749 }
 750
 751
 752 bool selectChange(Cursor & cur, bool forward)
 753 {
 754         if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
 755                 return false;
 756         Change ch = cur.paragraph().lookupChange(cur.pos());
 757
 758         CursorSlice tip1 = cur.top();
 759         for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
 760                 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
 761                 if (!ch2.isSimilarTo(ch))
 762                         break;
 763         }
 764         CursorSlice tip2 = cur.top();
 765         for (; tip2.pit() > 0 || tip2.pos() > 0;) {
 766                 tip2.backwardPos();
 767                 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
 768                 if (!ch2.isSimilarTo(ch)) {
 769                         // take a step forward to correctly set the selection
 770                         tip2.forwardPos();
 771                         break;
 772                 }
 773         }
 774         if (forward)
 775                 swap(tip1, tip2);
 776         cur.top() = tip1;
 777         cur.bv().mouseSetCursor(cur, false);
 778         cur.top() = tip2;
 779         cur.bv().mouseSetCursor(cur, true);
 780         return true;
 781 }
 782
 783
 784 namespace {
 785
 786
 787 bool findChange(BufferView * bv, bool forward)
 788 {
 789         Cursor cur(*bv);
 790         cur.setCursor(forward ? bv->cursor().selectionEnd()
 791                       : bv->cursor().selectionBegin());
 792         forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
 793         return selectChange(cur, forward);
 794 }
 795
 796 } // namespace
 797
 798 bool findNextChange(BufferView * bv)
 799 {
 800         return findChange(bv, true);
 801 }
 802
 803
 804 bool findPreviousChange(BufferView * bv)
 805 {
 806         return findChange(bv, false);
 807 }
 808
 809
 810
 811 namespace {
 812
 813 typedef vector<pair<string, string> > Escapes;
 814
 815 string string2regex(string in)
 816 {
 817         static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\$\\])" };
 818         string tempx = std::regex_replace(in, specialChars,  R"(\$&)" );
 819         // Special handling for ' '
 820         string temp("");
 821         int blanks = 0;
 822         for (unsigned i = 0; i < tempx.size(); i++) {
 823                 if (tempx[i] == ' ' || tempx[i] == '~' ) {
 824                         // normal blanks
 825                         blanks++;
 826                 }
 827                 else if ((tempx[i] == '\302' && tempx[i+1] == '\240')
 828                         || (tempx[i] == '\342' && tempx[i+1] == '\200')) {
 829                         // protected space
 830                         // thin space
 831                         blanks++;
 832                         i++;
 833                 }
 834                 else {
 835                         if (blanks > 0) {
 836                                 if (blanks > 1)
 837                                         temp += "\\s+";
 838                                 else
 839                                         temp += "\\s";
 840                         }
 841                         temp += tempx[i];
 842                         blanks = 0;
 843                 }
 844         }
 845         if (blanks > 0) {
 846                 if (blanks > 1)
 847                         temp += "\\s+";
 848                 else
 849                         temp += "\\s";
 850         }
 851
 852         string temp2("");
 853         size_t lastpos = 0;
 854         size_t fl_pos = 0;
 855         int offset = 1;
 856         while (fl_pos < temp.size()) {
 857                 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
 858                 if (fl_pos == string::npos)
 859                         break;
 860                 offset = 16;
 861                 temp2 += temp.substr(lastpos, fl_pos - lastpos);
 862                 temp2 += "\\n";
 863                 lastpos = fl_pos;
 864         }
 865         if (lastpos == 0)
 866                 return(temp);
 867         if (lastpos < temp.size()) {
 868                 temp2 += temp.substr(lastpos, temp.size() - lastpos);
 869         }
 870         return temp2;
 871 }
 872
 873 static void buildAccentsMap();
 874
 875 string correctRegex(string t, bool withformat)
 876 {
 877         /* Convert \backslash => \
 878          * and \{, \}, \[, \] => {, }, [, ]
 879          */
 880         string s("");
 881         static std::regex wordre("(\\\\)*(\\\\(( |[A-Za-z]+|[\\{\\}%])( |\\{\\})?|[\\[\\]\\{\\}]))");
 882         static std::regex protectedSpace { R"(~)" };
 883         size_t lastpos = 0;
 884         smatch sub;
 885         bool backslashed = false;
 886         if (accents.empty())
 887                 buildAccentsMap();
 888
 889         //LYXERR0("correctRegex input '" << t << "'");
 890         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
 891                 sub = *it;
 892                 string replace;
 893                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
 894                         continue;
 895                 }
 896                 else {
 897                         if (sub.str(4) == "backslash") {
 898                                 replace = "\\";
 899                                 {
 900                                         // transforms '\backslash \{' into '\{'
 901                                         string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
 902                                         if ((next == "\\{") || (next == "\\}") || (next == "\\ ")) {
 903                                                 replace = "";
 904                                                 backslashed = true;
 905                                         }
 906                                 }
 907                         }
 908                         else if (sub.str(4) == "mathcircumflex")
 909                                 replace = "^";
 910                         else if (backslashed) {
 911                                 backslashed = false;
 912                                 if (withformat) {
 913                                         if (sub.str(3) == "{")
 914                                                 replace = accents["braceleft"];
 915                                         else if (sub.str(3) == "}")
 916                                                 replace = accents["braceright"];
 917                                         else if (sub.str(3) == " ")
 918                                                 replace = "\\ ";
 919                                         else {
 920                                                 // else part should not exist
 921                                                 LASSERT(0, /**/);
 922                                         }
 923                                 }
 924                                 else {
 925                                         if (sub.str(3) == "{")
 926                                                 replace = "\\{";
 927                                         else if (sub.str(3) == "}")
 928                                                 replace = "\\}";
 929                                         else if (sub.str(3) == " ")
 930                                                 replace = "\\ ";
 931                                         else {
 932                                                 // else part should not exist
 933                                                 LASSERT(0, /**/);
 934                                         }
 935                                 }
 936                         }
 937                         else if (sub.str(4) == "{") // transforms '\{' into '{'
 938                                 replace = "{";
 939                         else if (sub.str(4) == "}")
 940                                 replace = "}";
 941                         else if (sub.str(4) == "%")
 942                                 replace = "%";
 943                         else if (sub.str(4) == " ")
 944                                 replace = " ";
 945                         else {
 946                                 AccentsIterator it_ac = accents.find(sub.str(4));
 947                                 if (it_ac == accents.end()) {
 948                                         replace = sub.str(2);
 949                                 }
 950                                 else {
 951                                         replace = it_ac->second;
 952                                 }
 953                         }
 954                 }
 955                 if (lastpos < (size_t) sub.position(2))
 956                         s += std::regex_replace(t.substr(lastpos, sub.position(2) - lastpos), protectedSpace, R"( )");
 957                 s += replace;
 958                 lastpos = sub.position(2) + sub.length(2);
 959         }
 960         if (lastpos == 0)
 961                 s = std::regex_replace(t, protectedSpace, R"( )");
 962         else if (lastpos < t.length())
 963                 s += std::regex_replace(t.substr(lastpos, t.length() - lastpos), protectedSpace, R"( )");
 964         // Handle quotes in regex
 965         // substitute all '„', '“', '»', '«' with '"'
 966         // and all '‚', '‘', '›', '‹' with "\'"
 967         static std::regex plainquotes { R"(„|“|»|«)" };
 968         static std::regex innerquotes { R"(‚|‘|›|‹)" };
 969         t = std::regex_replace(s, plainquotes, R"(")");
 970         s = std::regex_replace(t, innerquotes, R"(')");
 971         //LYXERR0("correctRegex output '" << s << "'");
 972         return s;
 973 }
 974
 975 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
 976 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
 977 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
 978 string escape_for_regex(string s, bool withformat)
 979 {
 980         size_t lastpos = 0;
 981         string result = "";
 982         while (lastpos < s.size()) {
 983                 size_t regex_pos = s.find("\\regexp{", lastpos);
 984                 if (regex_pos == string::npos) {
 985                         regex_pos = s.size();
 986                 }
 987                 if (regex_pos > lastpos) {
 988                         result += string2regex(s.substr(lastpos, regex_pos-lastpos));
 989                         lastpos = regex_pos;
 990                         if (lastpos == s.size())
 991                                 break;
 992                 }
 993                 size_t end_pos = s.find("\\endregexp", regex_pos + 8);
 994                 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
 995                 lastpos = end_pos + 13;
 996         }
 997         return result;
 998 }
 999
1000
1001 /// Wrapper for lyx::regex_replace with simpler interface
1002 bool regex_replace(string const & s, string & t, string const & searchstr,
1003                    string const & replacestr)
1004 {
1005         regex e(searchstr, regex_constants::ECMAScript);
1006         ostringstream oss;
1007         ostream_iterator<char, char> it(oss);
1008         regex_replace(it, s.begin(), s.end(), e, replacestr);
1009         // tolerate t and s be references to the same variable
1010         bool rv = (s != oss.str());
1011         t = oss.str();
1012         return rv;
1013 }
1014
1015 class MatchResult {
1016 public:
1017         enum range {
1018                 newIsTooFar,
1019                 newIsBetter,
1020                 newIsInvalid
1021         };
1022         int match_len;
1023         int match_prefix;
1024         int match2end;
1025         int pos;
1026         int leadsize;
1027         int pos_len;
1028         int searched_size;
1029         vector <string> result = vector <string>();
1030         MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {}
1031 };
1032
1033 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
1034 {
1035         if (newres.match2end < oldres.match2end)
1036                 return MatchResult::newIsTooFar;
1037         if (newres.match_len < oldres.match_len)
1038                 return MatchResult::newIsTooFar;
1039
1040         if (newres.match_len == oldres.match_len) {
1041                 if (newres.match2end == oldres.match2end)
1042                         return MatchResult::newIsBetter;
1043         }
1044         return MatchResult::newIsInvalid;
1045 }
1046
1047 /** The class performing a match between a position in the document and the FindAdvOptions.
1048  **/
1049
1050 class MatchStringAdv {
1051 public:
1052         MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
1053
1054         /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
1055          ** constructor as opt.search, under the opt.* options settings.
1056          **
1057          ** @param at_begin
1058          **     If set, then match is searched only against beginning of text starting at cur.
1059          **     If unset, then match is searched anywhere in text starting at cur.
1060          **
1061          ** @return
1062          ** The length of the matching text, or zero if no match was found.
1063          **/
1064         MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1065 #if QTSEARCH
1066         bool regexIsValid;
1067         string regexError;
1068 #endif
1069
1070 public:
1071         /// buffer
1072         lyx::Buffer * p_buf;
1073         /// first buffer on which search was started
1074         lyx::Buffer * const p_first_buf;
1075         /// options
1076         FindAndReplaceOptions const & opt;
1077
1078 private:
1079         /// Auxiliary find method (does not account for opt.matchword)
1080         MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1081         void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
1082
1083         /** Normalize a stringified or latexified LyX paragraph.
1084          **
1085          ** Normalize means:
1086          ** <ul>
1087          **   <li>if search is not casesensitive, then lowercase the string;
1088          **   <li>remove any newline at begin or end of the string;
1089          **   <li>replace any newline in the middle of the string with a simple space;
1090          **   <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
1091          ** </ul>
1092          **
1093          ** @todo Normalization should also expand macros, if the corresponding
1094          ** search option was checked.
1095          **/
1096         string normalize(docstring const & s, bool ignore_fomat) const;
1097         // normalized string to search
1098         string par_as_string;
1099         // regular expression to use for searching
1100         // regexp2 is same as regexp, but prefixed with a ".*?"
1101 #if QTSEARCH
1102         QRegularExpression regexp;
1103         QRegularExpression regexp2;
1104 #else
1105         regex regexp;
1106         regex regexp2;
1107 #endif
1108         // leading format material as string
1109         string lead_as_string;
1110         // par_as_string after removal of lead_as_string
1111         string par_as_string_nolead;
1112         // unmatched open braces in the search string/regexp
1113         int open_braces;
1114         // number of (.*?) subexpressions added at end of search regexp for closing
1115         // environments, math mode, styles, etc...
1116         int close_wildcards;
1117 public:
1118         // Are we searching with regular expressions ?
1119         bool use_regexp = false;
1120         static int valid_matches;
1121         static vector <string> matches;
1122         void FillResults(MatchResult &found_mr);
1123 };
1124
1125 int MatchStringAdv::valid_matches = 0;
1126 vector <string> MatchStringAdv::matches = vector <string>(10);
1127
1128 void MatchStringAdv::FillResults(MatchResult &found_mr)
1129 {
1130         if (found_mr.match_len > 0) {
1131                 valid_matches = found_mr.result.size();
1132                 for (size_t i = 0; i < found_mr.result.size(); i++)
1133                         matches[i] = found_mr.result[i];
1134         } else
1135                 valid_matches = 0;
1136 }
1137
1138 static void setFindParams(OutputParams &runparams)
1139 {
1140         runparams.flavor = Flavor::XeTeX;
1141         //runparams.use_polyglossia = true;
1142         runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1143         // No side effect of file copying and image conversion
1144         runparams.dryrun = true;
1145 }
1146
1147 static docstring buffer_to_latex(Buffer & buffer)
1148 {
1149         //OutputParams runparams(&buffer.params().encoding());
1150         OutputParams runparams(encodings.fromLyXName("utf8"));
1151         odocstringstream ods;
1152         otexstream os(ods);
1153         runparams.nice = true;
1154         setFindParams(runparams);
1155         if (ignoreFormats.getDeleted())
1156                 runparams.for_search = OutputParams::SearchWithoutDeleted;
1157         else
1158                 runparams.for_search = OutputParams::SearchWithDeleted;
1159         if (ignoreFormats.getNonContent()) {
1160                 runparams.for_search |= OutputParams::SearchNonOutput;
1161         }
1162         pit_type const endpit = buffer.paragraphs().size();
1163         for (pit_type pit = 0; pit != endpit; ++pit) {
1164                 TeXOnePar(buffer, buffer.text(), pit, os, runparams, string(), -1, -1, true);
1165                 LYXERR(Debug::FINDVERBOSE, "searchString up to here: " << ods.str());
1166         }
1167         return ods.str();
1168 }
1169
1170 static string latexNamesToUtf8(docstring strIn, bool withformat)
1171 {
1172         string addtmp = to_utf8(strIn);
1173         static regex const rmAcc("(\\\\)*("
1174                                          "\\\\([A-Za-z]+\\{.\\})"       // e.g. "ddot{A}" == sub.str(3)
1175                                         "|\\\\([A-Za-z]+)( |\\{\\})?"   // e.g. "LyX", "LyX{}", "LyX " == sub.str(4)
1176                                         ")"
1177                                 );
1178         size_t lastpos = 0;
1179         smatch sub;
1180         string replace;
1181         string add("");
1182         if (accents.empty())
1183                 buildAccentsMap();
1184         for (sregex_iterator it_add(addtmp.begin(), addtmp.end(), rmAcc), end; it_add != end; ++it_add) {
1185                 sub = *it_add;
1186                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
1187                         continue;
1188                 }
1189                 else {
1190                         string key;
1191                         if (sub.length(3) > 0)
1192                                 key = sub.str(3);
1193                         else
1194                                 key = sub.str(4);
1195                         AccentsIterator it_ac = accents.find(key);
1196                         if (it_ac == accents.end()) {
1197                                 replace = sub.str(2);
1198                         }
1199                         else {
1200                                 replace = it_ac->second;
1201                         }
1202                 }
1203                 if (lastpos < (size_t) sub.position(2))
1204                         add += addtmp.substr(lastpos, sub.position(2) - lastpos);
1205                 add += replace;
1206                 lastpos = sub.position(2) + sub.length(2);
1207         }
1208         if (lastpos == 0)
1209                 add = addtmp;
1210         else if (addtmp.length() > lastpos)
1211                 add += addtmp.substr(lastpos, addtmp.length() - lastpos);
1212         if (!withformat) {
1213                 static std::regex repltilde { R"(~)" };
1214                 add = std::regex_replace(add, repltilde, accents["lyxtilde"]);
1215         }
1216         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1217                         << add << "'");
1218         return add;
1219 }
1220
1221 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1222 {
1223         docstring str;
1224         if (!opt.ignoreformat) {
1225                 str = buffer_to_latex(buffer);
1226         } else {
1227                 // OutputParams runparams(&buffer.params().encoding());
1228                 OutputParams runparams(encodings.fromLyXName("utf8"));
1229                 runparams.nice = true;
1230                 setFindParams(runparams);
1231                 int option = AS_STR_INSETS |AS_STR_PLAINTEXT;
1232                 if (ignoreFormats.getDeleted()) {
1233                         option |= AS_STR_SKIPDELETE;
1234                         runparams.for_search = OutputParams::SearchWithoutDeleted;
1235                 }
1236                 else {
1237                         runparams.for_search = OutputParams::SearchWithDeleted;
1238                 }
1239                 if (ignoreFormats.getNonContent()) {
1240                         runparams.for_search |= OutputParams::SearchNonOutput;
1241                 }
1242                 string t("");
1243                 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1244                         Paragraph const & par = buffer.paragraphs().at(pit);
1245                         string add = latexNamesToUtf8(par.asString(pos_type(0), par.size(),
1246                                                                 option,
1247                                                                 &runparams), !opt.ignoreformat);
1248                         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1249                                 << add << "'");
1250                         t += add;
1251                 }
1252                 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1253                 while (regex_replace(t, t, "\\\\(text|lyxmathsym|ensuremath)\\{([^\\}]*)\\}", "$2"));
1254                 str = from_utf8(t);
1255         }
1256         return str;
1257 }
1258
1259
1260 /// Return separation pos between the leading material and the rest
1261 static size_t identifyLeading(string const & s)
1262 {
1263         string t = s;
1264         // @TODO Support \item[text]
1265         // Kornel: Added textsl, textsf, textit, texttt and noun
1266         // + allow to search for colored text too
1267         while (regex_replace(t, t, "^\\\\(("
1268                              "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1269                              "lyxaddress|lyxrightaddress|"
1270                              "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1271                              "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1272                              "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1273                              "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1274                || regex_replace(t, t, "^\\$", "")
1275                || regex_replace(t, t, "^\\\\\\[", "")
1276                || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1277                || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1278                ;
1279         LYXERR(Debug::FINDVERBOSE, "  after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1280         return s.size() - t.size();
1281 }
1282
1283 /*
1284  * Given a latexified string, retrieve some handled features
1285  * The features of the regex will later be compared with the features
1286  * of the searched text. If the regex features are not a
1287  * subset of the analized, then, in not format ignoring search
1288  * we can early stop the search in the relevant inset.
1289  */
1290 typedef map<string, bool> Features;
1291
1292 static Features identifyFeatures(string const & s)
1293 {
1294         static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1295         static regex const valid("^("
1296                 "("
1297                         "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1298                                 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1299                                 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1300                         "(u|uu)line|(s|x)out|uwave|"
1301                         "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1302                 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1303         smatch sub;
1304         bool displ = true;
1305         Features info;
1306
1307         for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1308                 sub = *it;
1309                 if (displ) {
1310                         if (sub.str(1).compare("regexp") == 0) {
1311                                 displ = false;
1312                                 continue;
1313                         }
1314                         string token = sub.str(1);
1315                         smatch sub2;
1316                         if (regex_match(token, sub2, valid)) {
1317                                 info[token] = true;
1318                         }
1319                         else {
1320                                 // ignore
1321                         }
1322                 }
1323                 else {
1324                         if (sub.str(1).compare("endregexp") == 0) {
1325                                 displ = true;
1326                                 continue;
1327                         }
1328                 }
1329         }
1330         return info;
1331 }
1332
1333 /*
1334  * defines values features of a key "\\[a-z]+{"
1335  */
1336 class KeyInfo {
1337 public:
1338         enum KeyType {
1339                 /* Char type with content discarded
1340                  * like \hspace{1cm} */
1341                 noContent,
1342                 /* Char, like \backslash */
1343                 isChar,
1344                 /* replace starting backslash with '#' */
1345                 isText,
1346                 /* \part, \section*, ... */
1347                 isSectioning,
1348                 /* title, author etc */
1349                 isTitle,
1350                 /* \foreignlanguage{ngerman}, ... */
1351                 isMain,
1352                 /* inside \code{}
1353                  * to discard language in content */
1354                 noMain,
1355                 isRegex,
1356                 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1357                 isMath,
1358                 /* fonts, colors, markups, ... */
1359                 isStandard,
1360                 /* footnotesize, ... large, ...
1361                  * Ignore all of them */
1362                 isSize,
1363                 invalid,
1364                 /* inputencoding, ...
1365                  * Discard also content, because they do not help in search */
1366                 doRemove,
1367                 /* twocolumns, ...
1368                  * like remove, but also all arguments */
1369                 removeWithArg,
1370                 /* item, listitem */
1371                 isList,
1372                 /* tex, latex, ... like isChar */
1373                 isIgnored,
1374                 /* like \lettrine[lines=5]{}{} */
1375                 cleanToStart,
1376                 // like isStandard, but always remove head
1377                 headRemove,
1378                 /* End of arguments marker for lettrine,
1379                  * so that they can be ignored */
1380                 endArguments
1381         };
1382         KeyInfo() = default;
1383         KeyInfo(KeyType type, int parcount, bool disable)
1384                 : keytype(type),
1385                   parenthesiscount(parcount),
1386                   disabled(disable) {}
1387         KeyType keytype = invalid;
1388         string head;
1389         int _tokensize = -1;
1390         int _tokenstart = -1;
1391         int _dataStart = -1;
1392         int _dataEnd = -1;
1393         int parenthesiscount = 1;
1394         bool disabled = false;
1395         bool used = false; /* by pattern */
1396 };
1397
1398 class Border {
1399 public:
1400         Border(int l=0, int u=0) : low(l), upper(u) {}
1401         int low;
1402         int upper;
1403 };
1404
1405 #define MAXOPENED 30
1406 class Intervall {
1407         bool isPatternString_;
1408 public:
1409         explicit Intervall(bool isPattern, string const & p)
1410                 : isPatternString_(isPattern), par(p), ignoreidx(-1),
1411                   actualdeptindex(0), hasTitle(false), langcount(0)
1412         {
1413                 depts[0] = 0;
1414                 closes[0] = 0;
1415         }
1416
1417         string par;
1418         int ignoreidx;
1419         static vector<Border> borders;
1420         int depts[MAXOPENED];
1421         int closes[MAXOPENED];
1422         int actualdeptindex;
1423         int previousNotIgnored(int) const;
1424         int nextNotIgnored(int) const;
1425         void handleOpenP(int i);
1426         void handleCloseP(int i, bool closingAllowed);
1427         void resetOpenedP(int openPos);
1428         void addIntervall(int upper);
1429         void addIntervall(int low, int upper); /* if explicit */
1430         void removeAccents();
1431         void setForDefaultLang(KeyInfo const & defLang) const;
1432         int findclosing(int start, int end, char up, char down, int repeat);
1433         void handleParentheses(int lastpos, bool closingAllowed);
1434         bool hasTitle;
1435         // Number of disabled language specs up
1436         // to current position in actual interval
1437         int langcount;
1438         int isOpeningPar(int pos) const;
1439         string titleValue;
1440         void output(ostringstream &os, int lastpos);
1441         // string show(int lastpos);
1442 };
1443
1444 vector<Border> Intervall::borders = vector<Border>(30);
1445
1446 int Intervall::isOpeningPar(int pos) const
1447 {
1448         if ((pos < 0) || (size_t(pos) >= par.size()))
1449                 return 0;
1450         if (par[pos] != '{')
1451                 return 0;
1452         if (size_t(pos) + 2 >= par.size())
1453                 return 1;
1454         if (par[pos+2] != '}')
1455                 return 1;
1456         if (par[pos+1] == '[' || par[pos+1] == ']')
1457                 return 3;
1458         return 1;
1459 }
1460
1461 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1462 {
1463         // Enable the use of first token again
1464         if (ignoreidx >= 0) {
1465                 int value = defLang._tokenstart + defLang._tokensize;
1466                 int borderidx = 0;
1467                 if (hasTitle)
1468                         borderidx = 1;
1469                 if (value > 0) {
1470                         if (borders[borderidx].low < value)
1471                                 borders[borderidx].low = value;
1472                         if (borders[borderidx].upper < value)
1473                                 borders[borderidx].upper = value;
1474                 }
1475         }
1476 }
1477
1478 static void checkDepthIndex(int val)
1479 {
1480         static int maxdepthidx = MAXOPENED-2;
1481         static int lastmaxdepth = 0;
1482         if (val > lastmaxdepth) {
1483                 LYXERR(Debug::INFO, "Depth reached " << val);
1484                 lastmaxdepth = val;
1485         }
1486         if (val > maxdepthidx) {
1487                 maxdepthidx = val;
1488                 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1489         }
1490 }
1491
1492 #if 0
1493 // Not needed, because borders are now dynamically expanded
1494 static void checkIgnoreIdx(int val)
1495 {
1496         static int lastmaxignore = -1;
1497         if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1498                 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1499                 lastmaxignore = val;
1500         }
1501 }
1502 #endif
1503
1504 /*
1505  * Expand the region of ignored parts of the input latex string
1506  * The region is only relevant in output()
1507  */
1508 void Intervall::addIntervall(int low, int upper)
1509 {
1510         int idx;
1511         if (low == upper) return;
1512         for (idx = ignoreidx+1; idx > 0; --idx) {
1513                 if (low > borders[idx-1].upper) {
1514                         break;
1515                 }
1516         }
1517         Border br(low, upper);
1518         if (idx > ignoreidx) {
1519                 if (borders.size() <= size_t(idx)) {
1520                         borders.push_back(br);
1521                 }
1522                 else {
1523                         borders[idx] = br;
1524                 }
1525                 ignoreidx = idx;
1526                 // checkIgnoreIdx(ignoreidx);
1527                 return;
1528         }
1529         else {
1530                 // Expand only if one of the new bound is inside the interwall
1531                 // We know here that br.low > borders[idx-1].upper
1532                 if (br.upper < borders[idx].low) {
1533                         // We have to insert at this pos
1534                         if (size_t(ignoreidx+1) >= borders.size()) {
1535                                 borders.push_back(borders[ignoreidx]);
1536                         }
1537                         else {
1538                                 borders[ignoreidx+1] = borders[ignoreidx];
1539                         }
1540                         for (int i = ignoreidx; i > idx; --i) {
1541                                 borders[i] = borders[i-1];
1542                         }
1543                         borders[idx] = br;
1544                         ignoreidx += 1;
1545                         // checkIgnoreIdx(ignoreidx);
1546                         return;
1547                 }
1548                 // Here we know, that we are overlapping
1549                 if (br.low > borders[idx].low)
1550                         br.low = borders[idx].low;
1551                 // check what has to be concatenated
1552                 int count = 0;
1553                 for (int i = idx; i <= ignoreidx; i++) {
1554                         if (br.upper >= borders[i].low) {
1555                                 count++;
1556                                 if (br.upper < borders[i].upper)
1557                                         br.upper = borders[i].upper;
1558                         }
1559                         else {
1560                                 break;
1561                         }
1562                 }
1563                 // count should be >= 1 here
1564                 borders[idx] = br;
1565                 if (count > 1) {
1566                         for (int i = idx + count; i <= ignoreidx; i++) {
1567                                 borders[i-count+1] = borders[i];
1568                         }
1569                         ignoreidx -= count - 1;
1570                         return;
1571                 }
1572         }
1573 }
1574
1575 static void buildaccent(string n, string param, string values)
1576 {
1577         stringstream s(n);
1578         string name;
1579         const char delim = '|';
1580         while (getline(s, name, delim)) {
1581                 size_t start = 0;
1582                 for (char c : param) {
1583                         string key = name + "{" + c + "}";
1584                         // get the corresponding utf8-value
1585                         if ((values[start] & 0xc0) != 0xc0) {
1586                                 // should not happen, utf8 encoding starts at least with 11xxxxxx
1587                                 // but value for '\dot{i}' is 'i', which is ascii
1588                                 if ((values[start] & 0x80) == 0) {
1589                                         // is ascii
1590                                         accents[key] = values.substr(start, 1);
1591                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1592                                 }
1593                                 start++;
1594                                 continue;
1595                         }
1596                         for (int j = 1; ;j++) {
1597                                 if (start + j >= values.size()) {
1598                                         accents[key] = values.substr(start, j);
1599                                         start = values.size() - 1;
1600                                         break;
1601                                 }
1602                                 else if ((values[start+j] & 0xc0) != 0x80) {
1603                                         // This is the first byte of following utf8 char
1604                                         accents[key] = values.substr(start, j);
1605                                         start += j;
1606                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1607                                         break;
1608                                 }
1609                         }
1610                 }
1611         }
1612 }
1613
1614 // Helper function
1615 static string getutf8(unsigned uchar)
1616 {
1617 #define maxc 5
1618         string ret = string();
1619         char c[maxc] = {0};
1620         if (uchar <= 0x7f) {
1621                 c[maxc-1] = uchar & 0x7f;
1622         }
1623         else {
1624                 unsigned char rest = 0x40;
1625                 unsigned char first = 0x80;
1626                 int start = maxc-1;
1627                 for (int i = start; i >=0; --i) {
1628                         if (uchar < rest) {
1629                                 c[i] = first + uchar;
1630                                 break;
1631                         }
1632                         c[i] = 0x80 | (uchar &  0x3f);
1633                         uchar >>= 6;
1634                         rest >>= 1;
1635                         first >>= 1;
1636                         first |= 0x80;
1637                 }
1638         }
1639         for (int i = 0; i < maxc; i++) {
1640                 if (c[i] == 0) continue;
1641                 ret += c[i];
1642         }
1643         return(ret);
1644 }
1645
1646 static void addAccents(string latex_in, string unicode_out)
1647 {
1648         latex_in = latex_in.substr(1);
1649         AccentsIterator it_ac = accents.find(latex_in);
1650         if (it_ac == accents.end()) {
1651                 accents[latex_in] = unicode_out;
1652         }
1653         else {
1654                 LYXERR0("Key " << latex_in  << " already set");
1655         }
1656 }
1657
1658 void static fillMissingUnicodesymbols()
1659 {
1660         addAccents("\\pounds", getutf8(0x00a3));
1661         addAccents("\\textsterling", getutf8(0x00a3));
1662         addAccents("\\textyen", getutf8(0x00a5));
1663         addAccents("\\yen", getutf8(0x00a5));
1664         addAccents("\\textsection", getutf8(0x00a7));
1665         addAccents("\\mathsection", getutf8(0x00a7));
1666         addAccents("\\textcopyright", getutf8(0x00a9));
1667         addAccents("\\copyright", getutf8(0x00a9));
1668         addAccents("\\textlnot", getutf8(0x00ac));
1669         addAccents("\\neg", getutf8(0x00ac));
1670         addAccents("\\textregistered", getutf8(0x00ae));
1671         addAccents("\\circledR", getutf8(0x00ae));
1672         addAccents("\\textpm", getutf8(0x00b1));
1673         addAccents("\\pm", getutf8(0x00b1));
1674         addAccents("\\textparagraph", getutf8(0x00b6));
1675         addAccents("\\mathparagraph", getutf8(0x00b6));
1676         addAccents("\\textperiodcentered", getutf8(0x00b7));
1677         addAccents("\\texttimes", getutf8(0x00d7));
1678         addAccents("\\times", getutf8(0x00d7));
1679         addAccents("\\O", getutf8(0x00d8));
1680         addAccents("\\dh", getutf8(0x00f0));
1681         addAccents("\\eth", getutf8(0x00f0));
1682         addAccents("\\textdiv", getutf8(0x00f7));
1683         addAccents("\\div", getutf8(0x00f7));
1684         addAccents("\\o", getutf8(0x00f8));
1685         addAccents("\\textcrlambda", getutf8(0x019b));
1686         addAccents("\\j", getutf8(0x0237));
1687         addAccents("\\textrevepsilon", getutf8(0x025c));
1688         addAccents("\\textbaru", getutf8(0x0289));
1689         addAccents("\\textquoteleft", getutf8(0x02bb));
1690         addAccents("\\textGamma", getutf8(0x0393));
1691         addAccents("\\Gamma", getutf8(0x0393));
1692         addAccents("\\textDelta", getutf8(0x0394));
1693         addAccents("\\Delta", getutf8(0x0394));
1694         addAccents("\\textTheta", getutf8(0x0398));
1695         addAccents("\\Theta", getutf8(0x0398));
1696         addAccents("\\textLambda", getutf8(0x039b));
1697         addAccents("\\Lambda", getutf8(0x039b));
1698         addAccents("\\textXi", getutf8(0x039e));
1699         addAccents("\\Xi", getutf8(0x039e));
1700         addAccents("\\textPi", getutf8(0x03a0));
1701         addAccents("\\Pi", getutf8(0x03a0));
1702         addAccents("\\textSigma", getutf8(0x03a3));
1703         addAccents("\\Sigma", getutf8(0x03a3));
1704         addAccents("\\textUpsilon", getutf8(0x03a5));
1705         addAccents("\\Upsilon", getutf8(0x03a5));
1706         addAccents("\\textPhi", getutf8(0x03a6));
1707         addAccents("\\Phi", getutf8(0x03a6));
1708         addAccents("\\textPsi", getutf8(0x03a8));
1709         addAccents("\\Psi", getutf8(0x03a8));
1710         addAccents("\\textOmega", getutf8(0x03a9));
1711         addAccents("\\Omega", getutf8(0x03a9));
1712         addAccents("\\textalpha", getutf8(0x03b1));
1713         addAccents("\\alpha", getutf8(0x03b1));
1714         addAccents("\\textbeta", getutf8(0x03b2));
1715         addAccents("\\beta", getutf8(0x03b2));
1716         addAccents("\\textgamma", getutf8(0x03b3));
1717         addAccents("\\gamma", getutf8(0x03b3));
1718         addAccents("\\textdelta", getutf8(0x03b4));
1719         addAccents("\\delta", getutf8(0x03b4));
1720         addAccents("\\textepsilon", getutf8(0x03b5));
1721         addAccents("\\varepsilon", getutf8(0x03b5));
1722         addAccents("\\textzeta", getutf8(0x03b6));
1723         addAccents("\\zeta", getutf8(0x03b6));
1724         addAccents("\\texteta", getutf8(0x03b7));
1725         addAccents("\\eta", getutf8(0x03b7));
1726         addAccents("\\texttheta", getutf8(0x03b8));
1727         addAccents("\\theta", getutf8(0x03b8));
1728         addAccents("\\textiota", getutf8(0x03b9));
1729         addAccents("\\iota", getutf8(0x03b9));
1730         addAccents("\\textkappa", getutf8(0x03ba));
1731         addAccents("\\kappa", getutf8(0x03ba));
1732         addAccents("\\textlambda", getutf8(0x03bb));
1733         addAccents("\\lambda", getutf8(0x03bb));
1734         addAccents("\\textmu", getutf8(0x03bc));
1735         addAccents("\\mu", getutf8(0x03bc));
1736         addAccents("\\textnu", getutf8(0x03bd));
1737         addAccents("\\nu", getutf8(0x03bd));
1738         addAccents("\\textxi", getutf8(0x03be));
1739         addAccents("\\xi", getutf8(0x03be));
1740         addAccents("\\textpi", getutf8(0x03c0));
1741         addAccents("\\pi", getutf8(0x03c0));
1742         addAccents("\\textrho", getutf8(0x03c1));
1743         addAccents("\\rho", getutf8(0x03c1));
1744         addAccents("\\textfinalsigma", getutf8(0x03c2));
1745         addAccents("\\varsigma", getutf8(0x03c2));
1746         addAccents("\\textsigma", getutf8(0x03c3));
1747         addAccents("\\sigma", getutf8(0x03c3));
1748         addAccents("\\texttau", getutf8(0x03c4));
1749         addAccents("\\tau", getutf8(0x03c4));
1750         addAccents("\\textupsilon", getutf8(0x03c5));
1751         addAccents("\\upsilon", getutf8(0x03c5));
1752         addAccents("\\textphi", getutf8(0x03c6));
1753         addAccents("\\varphi", getutf8(0x03c6));
1754         addAccents("\\textchi", getutf8(0x03c7));
1755         addAccents("\\chi", getutf8(0x03c7));
1756         addAccents("\\textpsi", getutf8(0x03c8));
1757         addAccents("\\psi", getutf8(0x03c8));
1758         addAccents("\\textomega", getutf8(0x03c9));
1759         addAccents("\\omega", getutf8(0x03c9));
1760         addAccents("\\textdigamma", getutf8(0x03dd));
1761         addAccents("\\digamma", getutf8(0x03dd));
1762         addAccents("\\hebalef", getutf8(0x05d0));
1763         addAccents("\\aleph", getutf8(0x05d0));
1764         addAccents("\\hebbet", getutf8(0x05d1));
1765         addAccents("\\beth", getutf8(0x05d1));
1766         addAccents("\\hebgimel", getutf8(0x05d2));
1767         addAccents("\\gimel", getutf8(0x05d2));
1768         addAccents("\\hebdalet", getutf8(0x05d3));
1769         addAccents("\\daleth", getutf8(0x05d3));
1770         addAccents("\\hebhe", getutf8(0x05d4));
1771         addAccents("\\hebvav", getutf8(0x05d5));
1772         addAccents("\\hebzayin", getutf8(0x05d6));
1773         addAccents("\\hebhet", getutf8(0x05d7));
1774         addAccents("\\hebtet", getutf8(0x05d8));
1775         addAccents("\\hebyod", getutf8(0x05d9));
1776         addAccents("\\hebfinalkaf", getutf8(0x05da));
1777         addAccents("\\hebkaf", getutf8(0x05db));
1778         addAccents("\\heblamed", getutf8(0x05dc));
1779         addAccents("\\hebfinalmem", getutf8(0x05dd));
1780         addAccents("\\hebmem", getutf8(0x05de));
1781         addAccents("\\hebfinalnun", getutf8(0x05df));
1782         addAccents("\\hebnun", getutf8(0x05e0));
1783         addAccents("\\hebsamekh", getutf8(0x05e1));
1784         addAccents("\\hebayin", getutf8(0x05e2));
1785         addAccents("\\hebfinalpe", getutf8(0x05e3));
1786         addAccents("\\hebpe", getutf8(0x05e4));
1787         addAccents("\\hebfinaltsadi", getutf8(0x05e5));
1788         addAccents("\\hebtsadi", getutf8(0x05e6));
1789         addAccents("\\hebqof", getutf8(0x05e7));
1790         addAccents("\\hebresh", getutf8(0x05e8));
1791         addAccents("\\hebshin", getutf8(0x05e9));
1792         addAccents("\\hebtav", getutf8(0x05ea));
1793
1794         // Thai characters
1795         addAccents("\\thaiKoKai", getutf8(0x0e01));
1796         addAccents("\\thaiKhoKhai", getutf8(0x0e02));
1797         addAccents("\\thaiKhoKhuat", getutf8(0x0e03));
1798         addAccents("\\thaiKhoKhwai", getutf8(0x0e04));
1799         addAccents("\\thaiKhoKhon", getutf8(0x0e05));
1800         addAccents("\\thaiKhoRakhang", getutf8(0x0e06));
1801         addAccents("\\thaiNgoNgu", getutf8(0x0e07));
1802         addAccents("\\thaiChoChan", getutf8(0x0e08));
1803         addAccents("\\thaiChoChing", getutf8(0x0e09));
1804         addAccents("\\thaiChoChang", getutf8(0x0e0a));
1805         addAccents("\\thaiSoSo", getutf8(0x0e0b));
1806         addAccents("\\thaiChoChoe", getutf8(0x0e0c));
1807         addAccents("\\thaiYoYing", getutf8(0x0e0d));
1808         addAccents("\\thaiDoChada", getutf8(0x0e0e));
1809         addAccents("\\thaiToPatak", getutf8(0x0e0f));
1810         addAccents("\\thaiThoThan", getutf8(0x0e10));
1811         addAccents("\\thaiThoNangmontho", getutf8(0x0e11));
1812         addAccents("\\thaiThoPhuthao", getutf8(0x0e12));
1813         addAccents("\\thaiNoNen", getutf8(0x0e13));
1814         addAccents("\\thaiDoDek", getutf8(0x0e14));
1815         addAccents("\\thaiToTao", getutf8(0x0e15));
1816         addAccents("\\thaiThoThung", getutf8(0x0e16));
1817         addAccents("\\thaiThoThahan", getutf8(0x0e17));
1818         addAccents("\\thaiThoThong", getutf8(0x0e18));
1819         addAccents("\\thaiNoNu", getutf8(0x0e19));
1820         addAccents("\\thaiBoBaimai", getutf8(0x0e1a));
1821         addAccents("\\thaiPoPla", getutf8(0x0e1b));
1822         addAccents("\\thaiPhoPhung", getutf8(0x0e1c));
1823         addAccents("\\thaiFoFa", getutf8(0x0e1d));
1824         addAccents("\\thaiPhoPhan", getutf8(0x0e1e));
1825         addAccents("\\thaiFoFan", getutf8(0x0e1f));
1826         addAccents("\\thaiPhoSamphao", getutf8(0x0e20));
1827         addAccents("\\thaiMoMa", getutf8(0x0e21));
1828         addAccents("\\thaiYoYak", getutf8(0x0e22));
1829         addAccents("\\thaiRoRua", getutf8(0x0e23));
1830         addAccents("\\thaiRu", getutf8(0x0e24));
1831         addAccents("\\thaiLoLing", getutf8(0x0e25));
1832         addAccents("\\thaiLu", getutf8(0x0e26));
1833         addAccents("\\thaiWoWaen", getutf8(0x0e27));
1834         addAccents("\\thaiSoSala", getutf8(0x0e28));
1835         addAccents("\\thaiSoRusi", getutf8(0x0e29));
1836         addAccents("\\thaiSoSua", getutf8(0x0e2a));
1837         addAccents("\\thaiHoHip", getutf8(0x0e2b));
1838         addAccents("\\thaiLoChula", getutf8(0x0e2c));
1839         addAccents("\\thaiOAng", getutf8(0x0e2d));
1840         addAccents("\\thaiHoNokhuk", getutf8(0x0e2e));
1841         addAccents("\\thaiPaiyannoi", getutf8(0x0e2f));
1842         addAccents("\\thaiSaraA", getutf8(0x0e30));
1843         addAccents("\\thaiMaiHanakat", getutf8(0x0e31));
1844         addAccents("\\thaiSaraAa", getutf8(0x0e32));
1845         addAccents("\\thaiSaraAm", getutf8(0x0e33));
1846         addAccents("\\thaiSaraI", getutf8(0x0e34));
1847         addAccents("\\thaiSaraIi", getutf8(0x0e35));
1848         addAccents("\\thaiSaraUe", getutf8(0x0e36));
1849         addAccents("\\thaiSaraUee", getutf8(0x0e37));
1850         addAccents("\\thaiSaraU", getutf8(0x0e38));
1851         addAccents("\\thaiSaraUu", getutf8(0x0e39));
1852         addAccents("\\thaiPhinthu", getutf8(0x0e3a));
1853         addAccents("\\thaiSaraE", getutf8(0x0e40));
1854         addAccents("\\thaiSaraAe", getutf8(0x0e41));
1855         addAccents("\\thaiSaraO", getutf8(0x0e42));
1856         addAccents("\\thaiSaraAiMaimuan", getutf8(0x0e43));
1857         addAccents("\\thaiSaraAiMaimalai", getutf8(0x0e44));
1858         addAccents("\\thaiLakkhangyao", getutf8(0x0e45));
1859         addAccents("\\thaiMaiyamok", getutf8(0x0e46));
1860         addAccents("\\thaiMaitaikhu", getutf8(0x0e47));
1861         addAccents("\\thaiMaiEk", getutf8(0x0e48));
1862         addAccents("\\thaiMaiTho", getutf8(0x0e49));
1863         addAccents("\\thaiMaiTri", getutf8(0x0e4a));
1864         addAccents("\\thaiMaiChattawa", getutf8(0x0e4b));
1865         addAccents("\\thaiThanthakhat", getutf8(0x0e4c));
1866         addAccents("\\thaiNikhahit", getutf8(0x0e4d));
1867         addAccents("\\thaiYamakkan", getutf8(0x0e4e));
1868         addAccents("\\thaiFongman", getutf8(0x0e4f));
1869         addAccents("\\thaizero", getutf8(0x0e50));
1870         addAccents("\\thaione", getutf8(0x0e51));
1871         addAccents("\\thaitwo", getutf8(0x0e52));
1872         addAccents("\\thaithree", getutf8(0x0e53));
1873         addAccents("\\thaifour", getutf8(0x0e54));
1874         addAccents("\\thaifive", getutf8(0x0e55));
1875         addAccents("\\thaisix", getutf8(0x0e56));
1876         addAccents("\\thaiseven", getutf8(0x0e57));
1877         addAccents("\\thaieight", getutf8(0x0e58));
1878         addAccents("\\thainine", getutf8(0x0e59));
1879         addAccents("\\thaiAngkhankhu", getutf8(0x0e5a));
1880         addAccents("\\thaiKhomut", getutf8(0x0e5b));
1881         addAccents("\\dag", getutf8(0x2020));
1882         addAccents("\\dagger", getutf8(0x2020));
1883         addAccents("\\textdagger", getutf8(0x2020));
1884         addAccents("\\ddag", getutf8(0x2021));
1885         addAccents("\\ddagger", getutf8(0x2021));
1886         addAccents("\\textdaggerdbl", getutf8(0x2021));
1887         addAccents("\\textbullet", getutf8(0x2022));
1888         addAccents("\\bullet", getutf8(0x2022));
1889         addAccents("\\dots", getutf8(0x2026));
1890         addAccents("\\ldots", getutf8(0x2026));
1891         addAccents("\\textellipsis", getutf8(0x2026));
1892         addAccents("\\textasciiacute", getutf8(0x2032));
1893         addAccents("\\prime", getutf8(0x2032));
1894         addAccents("\\textacutedbl", getutf8(0x2033));
1895         addAccents("\\dprime", getutf8(0x2033));
1896         addAccents("\\textasciigrave", getutf8(0x2035));
1897         addAccents("\\backprime", getutf8(0x2035));
1898         addAccents("\\textsubcircum{ }", getutf8(0x2038));
1899         addAccents("\\caretinsert", getutf8(0x2038));
1900         addAccents("\\textasteriskcentered", getutf8(0x204e));
1901         addAccents("\\ast", getutf8(0x204e));
1902         addAccents("\\textmho", getutf8(0x2127));
1903         addAccents("\\mho", getutf8(0x2127));
1904         addAccents("\\textleftarrow", getutf8(0x2190));
1905         addAccents("\\leftarrow", getutf8(0x2190));
1906         addAccents("\\textuparrow", getutf8(0x2191));
1907         addAccents("\\uparrow", getutf8(0x2191));
1908         addAccents("\\textrightarrow", getutf8(0x2192));
1909         addAccents("\\rightarrow", getutf8(0x2192));
1910         addAccents("\\textdownarrow", getutf8(0x2193));
1911         addAccents("\\downarrow", getutf8(0x2193));
1912         addAccents("\\textglobrise", getutf8(0x2197));
1913         addAccents("\\nearrow", getutf8(0x2197));
1914         addAccents("\\textglobfall", getutf8(0x2198));
1915         addAccents("\\searrow", getutf8(0x2198));
1916         addAccents("\\textsurd", getutf8(0x221a));
1917         addAccents("\\surd", getutf8(0x221a));
1918         addAccents("\\textbigcircle", getutf8(0x25ef));
1919         addAccents("\\bigcirc", getutf8(0x25ef));
1920         addAccents("\\FiveStar", getutf8(0x2605));
1921         addAccents("\\bigstar", getutf8(0x2605));
1922         addAccents("\\FiveStarOpen", getutf8(0x2606));
1923         addAccents("\\bigwhitestar", getutf8(0x2606));
1924         addAccents("\\Checkmark", getutf8(0x2713));
1925         addAccents("\\checkmark", getutf8(0x2713));
1926         addAccents("\\CrossMaltese", getutf8(0x2720));
1927         addAccents("\\maltese", getutf8(0x2720));
1928         addAccents("\\textlangle", getutf8(0x27e8));
1929         addAccents("\\langle", getutf8(0x27e8));
1930         addAccents("\\textrangle", getutf8(0x27e9));
1931         addAccents("\\rangle", getutf8(0x27e9));
1932 }
1933
1934 static void buildAccentsMap()
1935 {
1936         accents["imath"] = "ı";
1937         accents["i"] = "ı";
1938         accents["jmath"] = "ȷ";
1939         accents["cdot"] = "·";
1940         accents["textasciicircum"] = "^";
1941         accents["mathcircumflex"] = "^";
1942         accents["guillemotright"] = "»";
1943         accents["guillemotleft"] = "«";
1944         accents["hairspace"]     = getutf8(0xf0000);    // select from free unicode plane 15
1945         accents["thinspace"]     = getutf8(0xf0002);    // and used _only_ by findadv
1946         accents["negthinspace"]  = getutf8(0xf0003);    // to omit backslashed latex macros
1947         accents["medspace"]      = getutf8(0xf0004);    // See https://en.wikipedia.org/wiki/Private_Use_Areas
1948         accents["negmedspace"]   = getutf8(0xf0005);
1949         accents["thickspace"]    = getutf8(0xf0006);
1950         accents["negthickspace"] = getutf8(0xf0007);
1951         accents["lyx"]           = getutf8(0xf0010);    // Used logos
1952         accents["LyX"]           = getutf8(0xf0010);
1953         accents["tex"]           = getutf8(0xf0011);
1954         accents["TeX"]           = getutf8(0xf0011);
1955         accents["latex"]         = getutf8(0xf0012);
1956         accents["LaTeX"]         = getutf8(0xf0012);
1957         accents["latexe"]        = getutf8(0xf0013);
1958         accents["LaTeXe"]        = getutf8(0xf0013);
1959         accents["lyxarrow"]      = getutf8(0xf0020);
1960         accents["braceleft"]     = getutf8(0xf0030);
1961         accents["braceright"]    = getutf8(0xf0031);
1962         accents["lyxtilde"]      = getutf8(0xf0032);
1963         accents["sim"]           = getutf8(0xf0032);
1964         accents["backslash lyx"]           = getutf8(0xf0010);  // Used logos inserted with starting \backslash
1965         accents["backslash LyX"]           = getutf8(0xf0010);
1966         accents["backslash tex"]           = getutf8(0xf0011);
1967         accents["backslash TeX"]           = getutf8(0xf0011);
1968         accents["backslash latex"]         = getutf8(0xf0012);
1969         accents["backslash LaTeX"]         = getutf8(0xf0012);
1970         accents["backslash latexe"]        = getutf8(0xf0013);
1971         accents["backslash LaTeXe"]        = getutf8(0xf0013);
1972         accents["backslash lyxarrow"]      = getutf8(0xf0020);
1973         accents["ddot{\\imath}"] = "ï";
1974         buildaccent("ddot", "aAeEhHiIoOtuUwWxXyY",
1975                     "äÄëËḧḦïÏöÖẗüÜẅẄẍẌÿŸ");   // umlaut
1976         buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1977                     "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ");      // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1978         accents["acute{\\imath}"] = "í";
1979         buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1980                     "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1981         buildaccent("dacute|H|h", "oOuU", "őŐűŰ");  // double acute
1982         buildaccent("mathring|r", "aAuUwy",
1983                     "åÅůŮẘẙ");  // ring
1984         accents["check{\\imath}"] = "ǐ";
1985         accents["check{\\jmath}"] = "ǰ";
1986         buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1987                     "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ");        // caron
1988         accents["hat{\\imath}"] = "î";
1989         accents["hat{\\jmath}"] = "ĵ";
1990         buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1991                     "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ");  // circ
1992         accents["bar{\\imath}"] = "ī";
1993         buildaccent("bar|=", "aAeEiIoOuUyY",
1994                     "āĀēĒīĪōŌūŪȳȲ");        // macron
1995         accents["tilde{\\imath}"] = "ĩ";
1996         buildaccent("tilde", "aAeEiInNoOuUvVyY",
1997                     "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ");  // tilde
1998         accents["breve{\\imath}"] = "ĭ";
1999         buildaccent("breve|u", "aAeEgGiIoOuU",
2000                     "ăĂĕĔğĞĭĬŏŎŭŬ");        // breve
2001         accents["grave{\\imath}"] = "ì";
2002         buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
2003                     "àÀèÈìÌòÒùÙǹǸẁẀỳỲ");    // grave
2004         buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
2005                     "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ");      // dot below
2006         buildaccent("ogonek|k", "AaEeIiUuOo",
2007                     "ĄąĘęĮįŲųǪǫ");    // ogonek
2008         buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
2009                     "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ");        // cedilla
2010         buildaccent("subring|textsubring", "Aa",
2011                     "Ḁḁ");  // subring
2012         buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
2013                     "ḒḓḘḙḼḽṊṋṰṱṶṷ");    // subcircum
2014         buildaccent("subtilde|textsubtilde", "EeIiUu",
2015                     "ḚḛḬḭṴṵ");      // subtilde
2016         accents["dgrave{\\imath}"] = "ȉ";
2017         accents["textdoublegrave{\\i}"] = "ȉ";
2018         buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
2019                     "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
2020         accents["rcap{\\imath}"] = "ȋ";
2021         accents["textroundcap{\\i}"] = "ȋ";
2022         buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
2023                     "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
2024         buildaccent("slashed", "oO",
2025                     "øØ"); // slashed
2026         fillMissingUnicodesymbols(); // Add some still not handled entries contained in 'unicodesynbols'
2027         // LYXERR0("Number of accents " << accents.size());
2028 }
2029
2030 /*
2031  * Created accents in math or regexp environment
2032  * are macros, but we need the utf8 equivalent
2033  */
2034 void Intervall::removeAccents()
2035 {
2036         if (accents.empty())
2037                 buildAccentsMap();
2038         static regex const accre("\\\\("
2039                                  "([\\S]|[A-Za-z]+)\\{[^\\{\\}]+\\}"
2040                                  "|("
2041                                  "(backslash ([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))"
2042                                  "|[A-Za-z]+"
2043                                  ")"
2044                                  "(?![a-zA-Z]))");
2045         smatch sub;
2046         for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
2047                 sub = *itacc;
2048                 string key = sub.str(1);
2049                 AccentsIterator it_ac = accents.find(key);
2050                 if (it_ac != accents.end()) {
2051                         string val = it_ac->second;
2052                         size_t pos = sub.position(size_t(0));
2053                         for (size_t i = 0; i < val.size(); i++) {
2054                                 par[pos+i] = val[i];
2055                         }
2056                         // Remove possibly following space too
2057                         if (par[pos+sub.str(0).size()] == ' ')
2058                                 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
2059                         else
2060                                 addIntervall(pos+val.size(), pos + sub.str(0).size());
2061                         for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
2062                                 // remove traces of any remaining chars
2063                                 par[i] = ' ';
2064                         }
2065                 }
2066                 else {
2067                         LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
2068                 }
2069         }
2070 }
2071
2072 void Intervall::handleOpenP(int i)
2073 {
2074         actualdeptindex++;
2075         depts[actualdeptindex] = i+1;
2076         closes[actualdeptindex] = -1;
2077         checkDepthIndex(actualdeptindex);
2078 }
2079
2080 void Intervall::handleCloseP(int i, bool closingAllowed)
2081 {
2082         if (actualdeptindex <= 0) {
2083                 if (! closingAllowed)
2084                         LYXERR(Debug::FINDVERBOSE, "Bad closing parenthesis in latex");  /* should not happen, but the latex input may be wrong */
2085                 // if we are at the very end
2086                 addIntervall(i, i+1);
2087         }
2088         else {
2089                 closes[actualdeptindex] = i+1;
2090                 actualdeptindex--;
2091         }
2092 }
2093
2094 void Intervall::resetOpenedP(int openPos)
2095 {
2096         // Used as initializer for foreignlanguage entry
2097         actualdeptindex = 1;
2098         depts[1] = openPos+1;
2099         closes[1] = -1;
2100 }
2101
2102 int Intervall::previousNotIgnored(int start) const
2103 {
2104         int idx = 0;                          /* int intervalls */
2105         for (idx = ignoreidx; idx >= 0; --idx) {
2106                 if (start > borders[idx].upper)
2107                         return start;
2108                 if (start >= borders[idx].low)
2109                         start = borders[idx].low-1;
2110         }
2111         return start;
2112 }
2113
2114 int Intervall::nextNotIgnored(int start) const
2115 {
2116         int idx = 0;                          /* int intervalls */
2117         for (idx = 0; idx <= ignoreidx; idx++) {
2118                 if (start < borders[idx].low)
2119                         return start;
2120                 if (start < borders[idx].upper)
2121                         start = borders[idx].upper;
2122         }
2123         return start;
2124 }
2125
2126 typedef unordered_map<string, KeyInfo> KeysMap;
2127 typedef unordered_map<string, KeyInfo>::const_iterator KeysIterator;
2128 typedef vector< KeyInfo> Entries;
2129 static KeysMap keys = unordered_map<string, KeyInfo>();
2130
2131 class LatexInfo {
2132 private:
2133         int entidx_;
2134         Entries entries_;
2135         Intervall interval_;
2136         void buildKeys(bool);
2137         void buildEntries(bool);
2138         void makeKey(const string &, KeyInfo, bool isPatternString);
2139         void processRegion(int start, int region_end); /*  remove {} parts */
2140         void removeHead(KeyInfo const &, int count=0);
2141
2142 public:
2143         LatexInfo(string const & par, bool isPatternString)
2144                 : entidx_(-1), interval_(isPatternString, par)
2145         {
2146                 buildKeys(isPatternString);
2147                 entries_ = vector<KeyInfo>();
2148                 buildEntries(isPatternString);
2149         }
2150         int getFirstKey() {
2151                 entidx_ = 0;
2152                 if (entries_.empty()) {
2153                         return -1;
2154                 }
2155                 if (entries_[0].keytype == KeyInfo::isTitle) {
2156                         interval_.hasTitle = true;
2157                         if (! entries_[0].disabled) {
2158                                 interval_.titleValue = entries_[0].head;
2159                         }
2160                         else {
2161                                 interval_.titleValue = "";
2162                         }
2163                         removeHead(entries_[0]);
2164                         if (entries_.size() > 1)
2165                                 return 1;
2166                         else
2167                                 return -1;
2168                 }
2169                 return 0;
2170         }
2171         int getNextKey() {
2172                 entidx_++;
2173                 if (int(entries_.size()) > entidx_) {
2174                         return entidx_;
2175                 }
2176                 else {
2177                         return -1;
2178                 }
2179         }
2180         bool setNextKey(int idx) {
2181                 if ((idx == entidx_) && (entidx_ >= 0)) {
2182                         entidx_--;
2183                         return true;
2184                 }
2185                 else
2186                         return false;
2187         }
2188         int find(int start, KeyInfo::KeyType keytype) const {
2189                 if (start < 0)
2190                         return -1;
2191                 int tmpIdx = start;
2192                 while (tmpIdx < int(entries_.size())) {
2193                         if (entries_[tmpIdx].keytype == keytype)
2194                                 return tmpIdx;
2195                         tmpIdx++;
2196                 }
2197                 return -1;
2198         }
2199         int process(ostringstream & os, KeyInfo const & actual);
2200         int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
2201         // string show(int lastpos) { return interval.show(lastpos);}
2202         int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);}
2203         KeyInfo &getKeyInfo(int keyinfo) {
2204                 static KeyInfo invalidInfo = KeyInfo();
2205                 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
2206                         return invalidInfo;
2207                 else
2208                         return entries_[keyinfo];
2209         }
2210         void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);}
2211         void addIntervall(int low, int up) { interval_.addIntervall(low, up); }
2212 };
2213
2214
2215 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
2216 {
2217         int skip = 0;
2218         int depth = 0;
2219         for (int i = start; i < end; i += 1 + skip) {
2220                 char c;
2221                 c = par[i];
2222                 skip = 0;
2223                 if (c == '\\') skip = 1;
2224                 else if (c == up) {
2225                         depth++;
2226                 }
2227                 else if (c == down) {
2228                         if (depth == 0) {
2229                                 repeat--;
2230                                 if ((repeat <= 0) || (par[i+1] != up))
2231                                         return i;
2232                         }
2233                         --depth;
2234                 }
2235         }
2236         return end;
2237 }
2238
2239 class MathInfo {
2240         class MathEntry {
2241         public:
2242                 string wait;
2243                 size_t mathEnd;
2244                 size_t mathpostfixsize;
2245                 size_t mathStart;
2246                 size_t mathprefixsize;
2247                 size_t mathSize;
2248         };
2249         size_t actualIdx_;
2250         vector<MathEntry> entries_;
2251 public:
2252         MathInfo() {
2253                 actualIdx_ = 0;
2254         }
2255         void insert(string const & wait, size_t start, size_t prefixsize, size_t end, size_t postfixsize) {
2256                 MathEntry m = MathEntry();
2257                 m.wait = wait;
2258                 m.mathStart = start;
2259                 m.mathprefixsize = prefixsize;
2260                 m.mathEnd = end + postfixsize;
2261                 m.mathpostfixsize = postfixsize;
2262                 m.mathSize = m.mathEnd - m.mathStart;
2263                 entries_.push_back(m);
2264         }
2265         bool empty() const { return entries_.empty(); }
2266         size_t getEndPos() const {
2267                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2268                         return 0;
2269                 }
2270                 return entries_[actualIdx_].mathEnd;
2271         }
2272         size_t getStartPos() const {
2273                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2274                         return 100000;                    /*  definitely enough? */
2275                 }
2276                 return entries_[actualIdx_].mathStart;
2277         }
2278         size_t getPrefixSize() const {
2279                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2280                         return 0;
2281                 }
2282                 return entries_[actualIdx_].mathprefixsize;
2283         }
2284         size_t getPostfixSize() const {
2285                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2286                         return 0;
2287                 }
2288                 return entries_[actualIdx_].mathpostfixsize;
2289         }
2290         size_t getFirstPos() {
2291                 actualIdx_ = 0;
2292                 return getStartPos();
2293         }
2294         size_t getSize() const {
2295                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2296                         return size_t(0);
2297                 }
2298                 return entries_[actualIdx_].mathSize;
2299         }
2300         void incrEntry() { actualIdx_++; }
2301 };
2302
2303 void LatexInfo::buildEntries(bool isPatternString)
2304 {
2305         static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|x?x?alignat)\\*?\\})(\\{[0-9]+\\})?)");
2306         static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
2307         static bool disableLanguageOverride = false;
2308         smatch sub, submath;
2309         bool evaluatingRegexp = false;
2310         MathInfo mi;
2311         bool evaluatingMath = false;
2312         bool evaluatingCode = false;
2313         size_t codeEnd = 0;
2314         bool evaluatingOptional = false;
2315         size_t optionalEnd = 0;
2316         int codeStart = -1;
2317         KeyInfo found;
2318         bool math_end_waiting = false;
2319         size_t math_pos = 10000;
2320         size_t math_prefix_size = 1;
2321         string math_end;
2322         static vector<string> usedText = vector<string>();
2323         static bool removeMathHull = false;
2324
2325         interval_.removeAccents();
2326
2327         for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
2328                 submath = *itmath;
2329                 if ((submath.position(2) - submath.position(0)) %2 == 1) {
2330                         // prefixed by odd count of '\\'
2331                         continue;
2332                 }
2333                 if (math_end_waiting) {
2334                         size_t pos = submath.position(size_t(2));
2335                         if ((math_end == "$") &&
2336                                         (submath.str(2) == "$")) {
2337                                 mi.insert("$", math_pos, 1, pos, 1);
2338                                 math_end_waiting = false;
2339                         }
2340                         else if ((math_end == "\\]") &&
2341                                  (submath.str(2) == "\\]")) {
2342                                 mi.insert("\\]", math_pos, 2, pos, 2);
2343                                 math_end_waiting = false;
2344                         }
2345                         else if ((submath.str(3).compare("end") == 0) &&
2346                                  (submath.str(5).compare(math_end) == 0)) {
2347                                 mi.insert(math_end, math_pos, math_prefix_size, pos, submath.str(2).length());
2348                                 math_end_waiting = false;
2349                         }
2350                         else
2351                                 continue;
2352                 }
2353                 else {
2354                         if (submath.str(3).compare("begin") == 0) {
2355                                 math_end_waiting = true;
2356                                 math_end = submath.str(5);
2357                                 math_pos = submath.position(size_t(2));
2358                                 math_prefix_size = submath.str(2).length();
2359                         }
2360                         else if (submath.str(2).compare("\\[") == 0) {
2361                                 math_end_waiting = true;
2362                                 math_end = "\\]";
2363                                 math_pos = submath.position(size_t(2));
2364                         }
2365                         else if (submath.str(2) == "$") {
2366                                 size_t pos = submath.position(size_t(2));
2367                                 math_end_waiting = true;
2368                                 math_end = "$";
2369                                 math_pos = pos;
2370                         }
2371                 }
2372         }
2373         // Ignore language if there is math somewhere in pattern-string
2374         if (isPatternString) {
2375                 for (auto s: usedText) {
2376                         // Remove entries created in previous search runs
2377                         keys.erase(s);
2378                 }
2379                 usedText = vector<string>();
2380                 if (! mi.empty()) {
2381                         // Disable language
2382                         keys["foreignlanguage"].disabled = true;
2383                         disableLanguageOverride = true;
2384                         removeMathHull = false;
2385                 }
2386                 else {
2387                         removeMathHull = true;  // used later if not isPatternString
2388                         disableLanguageOverride = false;
2389                 }
2390         }
2391         else {
2392                 if (disableLanguageOverride) {
2393                         keys["foreignlanguage"].disabled = true;
2394                 }
2395         }
2396         math_pos = mi.getFirstPos();
2397         for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
2398                 sub = *it;
2399                 if ((sub.position(2) - sub.position(0)) %2 == 1) {
2400                         // prefixed by odd count of '\\'
2401                         continue;
2402                 }
2403                 string key = sub.str(5);
2404                 if (key == "") {
2405                         if (sub.str(2)[0] == '\\')
2406                                 key = sub.str(2)[1];
2407                         else {
2408                                 key = sub.str(2);
2409                         }
2410                 }
2411                 KeysIterator it_key = keys.find(key);
2412                 if (it_key != keys.end()) {
2413                         if (it_key->second.keytype == KeyInfo::headRemove) {
2414                                 KeyInfo found1 = it_key->second;
2415                                 found1.disabled = true;
2416                                 found1.head = "\\" + key + "{";
2417                                 found1._tokenstart = sub.position(size_t(2));
2418                                 found1._tokensize = found1.head.length();
2419                                 found1._dataStart = found1._tokenstart + found1.head.length();
2420                                 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
2421                                 found1._dataEnd = endpos;
2422                                 removeHead(found1);
2423                                 continue;
2424                         }
2425                 }
2426                 if (evaluatingRegexp) {
2427                         if (sub.str(3).compare("endregexp") == 0) {
2428                                 evaluatingRegexp = false;
2429                                 // found._tokenstart already set
2430                                 found._dataEnd = sub.position(size_t(2)) + 13;
2431                                 found._dataStart = found._dataEnd;
2432                                 found._tokensize = found._dataEnd - found._tokenstart;
2433                                 found.parenthesiscount = 0;
2434                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2435                         }
2436                         else {
2437                                 continue;
2438                         }
2439                 }
2440                 else {
2441                         if (evaluatingMath) {
2442                                 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
2443                                         continue;
2444                                 evaluatingMath = false;
2445                                 mi.incrEntry();
2446                                 math_pos = mi.getStartPos();
2447                         }
2448                         if (it_key == keys.end()) {
2449                                 found = KeyInfo(KeyInfo::isStandard, 0, true);
2450                                 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
2451                                 found = KeyInfo(KeyInfo::isText, 0, false);
2452                                 if (isPatternString) {
2453                                         found.keytype = KeyInfo::isChar;
2454                                         found.disabled = false;
2455                                         found.used = true;
2456                                 }
2457                                 keys[key] = found;
2458                                 usedText.push_back(key);
2459                         }
2460                         else
2461                                 found = keys[key];
2462                         if (key.compare("regexp") == 0) {
2463                                 evaluatingRegexp = true;
2464                                 found._tokenstart = sub.position(size_t(2));
2465                                 found._tokensize = 0;
2466                                 continue;
2467                         }
2468                 }
2469                 // Handle the other params of key
2470                 if (found.keytype == KeyInfo::isIgnored)
2471                         continue;
2472                 else if (found.keytype == KeyInfo::isMath) {
2473                         if (size_t(sub.position(size_t(2))) == math_pos) {
2474                                 found = keys[key];
2475                                 found._tokenstart = sub.position(size_t(2));
2476                                 found._tokensize = mi.getSize();
2477                                 found._dataEnd = found._tokenstart + found._tokensize;
2478                                 found._dataStart = found._dataEnd;
2479                                 found.parenthesiscount = 0;
2480                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2481                                 if (removeMathHull) {
2482                                         interval_.addIntervall(found._tokenstart, found._tokenstart + mi.getPrefixSize());
2483                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize(), found._dataEnd);
2484                                 }
2485                                 else {
2486                                         // Treate all math constructs as simple math
2487                                         interval_.par[found._tokenstart] = '$';
2488                                         interval_.par[found._dataEnd - mi.getPostfixSize()] = '$';
2489                                         interval_.addIntervall(found._tokenstart + 1, found._tokenstart + mi.getPrefixSize());
2490                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize() + 1, found._dataEnd);
2491                                 }
2492                                 evaluatingMath = true;
2493                         }
2494                         else {
2495                                 // begin|end of unknown env, discard
2496                                 // First handle tables
2497                                 // longtable|tabular
2498                                 bool discardComment;
2499                                 found = keys[key];
2500                                 found.keytype = KeyInfo::doRemove;
2501                                 if ((sub.str(7).compare("longtable") == 0) ||
2502                                                 (sub.str(7).compare("tabular") == 0)) {
2503                                         discardComment = true;        /* '%' */
2504                                 }
2505                                 else {
2506                                         discardComment = false;
2507                                         static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
2508                                         smatch sub2;
2509                                         string token = sub.str(7);
2510                                         if (regex_match(token, sub2, removeArgs)) {
2511                                                 found.keytype = KeyInfo::removeWithArg;
2512                                         }
2513                                 }
2514                                 // discard spaces before pos(2)
2515                                 int pos = sub.position(size_t(2));
2516                                 int count;
2517                                 for (count = 0; pos - count > 0; count++) {
2518                                         char c = interval_.par[pos-count-1];
2519                                         if (discardComment) {
2520                                                 if ((c != ' ') && (c != '%'))
2521                                                         break;
2522                                         }
2523                                         else if (c != ' ')
2524                                                 break;
2525                                 }
2526                                 found._tokenstart = pos - count;
2527                                 if (sub.str(3).compare(0, 5, "begin") == 0) {
2528                                         size_t pos1 = pos + sub.str(2).length();
2529                                         if (sub.str(7).compare("cjk") == 0) {
2530                                                 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2531                                                 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2532                                                         pos1 += 2;
2533                                                 found.keytype = KeyInfo::isMain;
2534                                                 found._dataStart = pos1;
2535                                                 found._dataEnd = interval_.par.length();
2536                                                 found.disabled = keys["foreignlanguage"].disabled;
2537                                                 found.used = keys["foreignlanguage"].used;
2538                                                 found._tokensize = pos1 - found._tokenstart;
2539                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2540                                         }
2541                                         else {
2542                                                 // Swallow possible optional params
2543                                                 while (interval_.par[pos1] == '[') {
2544                                                         pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2545                                                 }
2546                                                 // Swallow also the eventual parameter
2547                                                 if (interval_.par[pos1] == '{') {
2548                                                         found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2549                                                 }
2550                                                 else {
2551                                                         found._dataEnd = pos1;
2552                                                 }
2553                                                 found._dataStart = found._dataEnd;
2554                                                 found._tokensize = count + found._dataEnd - pos;
2555                                                 found.parenthesiscount = 0;
2556                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2557                                                 found.disabled = true;
2558                                         }
2559                                 }
2560                                 else {
2561                                         // Handle "\end{...}"
2562                                         found._dataStart = pos + sub.str(2).length();
2563                                         found._dataEnd = found._dataStart;
2564                                         found._tokensize = count + found._dataEnd - pos;
2565                                         found.parenthesiscount = 0;
2566                                         found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2567                                         found.disabled = true;
2568                                 }
2569                         }
2570                 }
2571                 else if (found.keytype != KeyInfo::isRegex) {
2572                         found._tokenstart = sub.position(size_t(2));
2573                         if (found.parenthesiscount == 0) {
2574                                 // Probably to be discarded
2575                                 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2576                                 char following = interval_.par[following_pos];
2577                                 if (following == ' ')
2578                                         found.head = "\\" + sub.str(5) + " ";
2579                                 else if (following == '=') {
2580                                         // like \uldepth=1000pt
2581                                         found.head = sub.str(2);
2582                                 }
2583                                 else
2584                                         found.head = "\\" + key;
2585                                 found._tokensize = found.head.length();
2586                                 found._dataEnd = found._tokenstart + found._tokensize;
2587                                 found._dataStart = found._dataEnd;
2588                         }
2589                         else {
2590                                 int params = found._tokenstart + key.length() + 1;
2591                                 if (evaluatingOptional) {
2592                                         if (size_t(found._tokenstart) > optionalEnd) {
2593                                                 evaluatingOptional = false;
2594                                         }
2595                                         else {
2596                                                 found.disabled = true;
2597                                         }
2598                                 }
2599                                 int optend = params;
2600                                 while (interval_.par[optend] == '[') {
2601                                         // discard optional parameters
2602                                         optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2603                                 }
2604                                 if (optend > params) {
2605                                         key += interval_.par.substr(params, optend-params);
2606                                         evaluatingOptional = true;
2607                                         optionalEnd = optend;
2608                                         if (found.keytype == KeyInfo::isSectioning) {
2609                                                 // Remove optional values (but still keep in header)
2610                                                 interval_.addIntervall(params, optend);
2611                                         }
2612                                 }
2613                                 string token = sub.str(7);
2614                                 int closings;
2615                                 if (interval_.par[optend] != '{') {
2616                                         closings = 0;
2617                                         found.parenthesiscount = 0;
2618                                         found.head = "\\" + key;
2619                                 }
2620                                 else
2621                                         closings = found.parenthesiscount;
2622                                 if (found.parenthesiscount == 1) {
2623                                         found.head = "\\" + key + "{";
2624                                 }
2625                                 else if (found.parenthesiscount > 1) {
2626                                         if (token != "") {
2627                                                 found.head = sub.str(2) + "{";
2628                                                 closings = found.parenthesiscount - 1;
2629                                         }
2630                                         else {
2631                                                 found.head = "\\" + key + "{";
2632                                         }
2633                                 }
2634                                 found._tokensize = found.head.length();
2635                                 found._dataStart = found._tokenstart + found.head.length();
2636                                 if (found.keytype == KeyInfo::doRemove) {
2637                                         if (closings > 0) {
2638                                                 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2639                                                 if (endpar >= interval_.par.length())
2640                                                         found._dataStart = interval_.par.length();
2641                                                 else
2642                                                         found._dataStart = endpar;
2643                                                 found._tokensize = found._dataStart - found._tokenstart;
2644                                         }
2645                                         else {
2646                                                 found._dataStart = found._tokenstart + found._tokensize;
2647                                         }
2648                                         closings = 0;
2649                                 }
2650                                 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2651                                         found._dataStart += 15;
2652                                 }
2653                                 size_t endpos;
2654                                 if (closings < 1)
2655                                         endpos = found._dataStart - 1;
2656                                 else
2657                                         endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2658                                 if (found.keytype == KeyInfo::isList) {
2659                                         // Check if it really is list env
2660                                         static regex const listre("^([a-z]+)$");
2661                                         smatch sub2;
2662                                         if (!regex_match(token, sub2, listre)) {
2663                                                 // Change the key of this entry. It is not in a list/item environment
2664                                                 found.keytype = KeyInfo::endArguments;
2665                                         }
2666                                 }
2667                                 if (found.keytype == KeyInfo::noMain) {
2668                                         evaluatingCode = true;
2669                                         codeEnd = endpos;
2670                                         codeStart = found._dataStart;
2671                                 }
2672                                 else if (evaluatingCode) {
2673                                         if (size_t(found._dataStart) > codeEnd)
2674                                                 evaluatingCode = false;
2675                                         else if (found.keytype == KeyInfo::isMain) {
2676                                                 // Disable this key, treate it as standard
2677                                                 found.keytype = KeyInfo::isStandard;
2678                                                 found.disabled = true;
2679                                                 if ((codeEnd +1 >= interval_.par.length()) &&
2680                                                                 (found._tokenstart == codeStart)) {
2681                                                         // trickery, because the code inset starts
2682                                                         // with \selectlanguage ...
2683                                                         codeEnd = endpos;
2684                                                         if (entries_.size() > 1) {
2685                                                                 entries_[entries_.size()-1]._dataEnd = codeEnd;
2686                                                         }
2687                                                 }
2688                                         }
2689                                 }
2690                                 if ((endpos == interval_.par.length()) &&
2691                                                 (found.keytype == KeyInfo::doRemove)) {
2692                                         // Missing closing => error in latex-input?
2693                                         // therefore do not delete remaining data
2694                                         found._dataStart -= 1;
2695                                         found._dataEnd = found._dataStart;
2696                                 }
2697                                 else
2698                                         found._dataEnd = endpos;
2699                         }
2700                         if (isPatternString) {
2701                                 keys[key].used = true;
2702                         }
2703                 }
2704                 entries_.push_back(found);
2705         }
2706 }
2707
2708 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2709 {
2710         stringstream s(keysstring);
2711         string key;
2712         const char delim = '|';
2713         while (getline(s, key, delim)) {
2714                 KeyInfo keyII(keyI);
2715                 if (isPatternString) {
2716                         keyII.used = false;
2717                 }
2718                 else if ( !keys[key].used)
2719                         keyII.disabled = true;
2720                 keys[key] = keyII;
2721         }
2722 }
2723
2724 void LatexInfo::buildKeys(bool isPatternString)
2725 {
2726
2727         static bool keysBuilt = false;
2728         if (keysBuilt && !isPatternString) return;
2729
2730         // Keys to ignore in any case
2731         makeKey("text|textcyrillic|lyxmathsym|ensuremath", KeyInfo(KeyInfo::headRemove, 1, true), true);
2732         // Known standard keys with 1 parameter.
2733         // Split is done, if not at start of region
2734         makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2735         makeKey("textbf",               KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2736         makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2737         makeKey("uuline|uline|uwave",   KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2738         makeKey("emph|noun",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2739         makeKey("sout|xout",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2740
2741         makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2742                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2743         makeKey("section*|subsection*|subsubsection*|paragraph*",
2744                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2745         makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2746         makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2747         // Regex
2748         makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2749
2750         // Split is done, if not at start of region
2751         makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2752         makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2753
2754         // Split is done always.
2755         makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2756
2757         // Known charaters
2758         // No split
2759         makeKey("backslash|textbackslash|slash",  KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2760         makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2761         makeKey("textasciiacute|texemdash",       KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2762         makeKey("dots|ldots",                     KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2763         // Spaces
2764         makeKey("quad|qquad|hfill|dotfill",               KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2765         makeKey("textvisiblespace|nobreakspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2766         makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2767         makeKey("thickspace|medspace|thinspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2768         // Skip
2769         // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2770         // Custom space/skip, remove the content (== length value)
2771         makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2772         // Found in fr/UserGuide.lyx
2773         makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2774         // quotes
2775         makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2776         makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2777         // Known macros to remove (including their parameter)
2778         // No split
2779         makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2780         makeKey("addtocounter|setlength",                 KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2781         // handle like standard keys with 1 parameter.
2782         makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2783
2784         if (ignoreFormats.getDeleted()) {
2785                 // Ignore deleted text
2786                 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2787         }
2788         else {
2789                 // but preserve added text
2790                 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2791         }
2792         makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2793
2794         // Macros to remove, but let the parameter survive
2795         // No split
2796         makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2797
2798         // Remove language spec from content of these insets
2799         makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2800
2801         // Same effect as previous, parameter will survive (because there is no one anyway)
2802         // No split
2803         makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2804         // Remove table decorations
2805         makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2806         // Discard shape-header.
2807         // For footnote or shortcut too, because of lang settings
2808         // and wrong handling if used 'KeyInfo::noMain'
2809         makeKey("circlepar|diamondpar|heartpar|nutpar",  KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2810         makeKey("trianglerightpar|hexagonpar|starpar",   KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2811         makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2812         makeKey("triangleleftpar|shapepar|dropuppar",    KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2813         makeKey("hphantom|vphantom|note|footnote|shortcut|include|includegraphics",     KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2814         makeKey("textgreek|textcyrillic", KeyInfo(KeyInfo::isStandard, 1, true), false);
2815         makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2816         // like ('tiny{}' or '\tiny ' ... )
2817         makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, ignoreFormats.getSize()), isPatternString);
2818
2819         // Survives, like known character
2820         // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2821         makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2822         makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2823
2824         makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2825         makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2826         makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2827
2828         makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2829         // Remove RTL/LTR marker
2830         makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2831         makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2832         makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2833         makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2834         makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2835         makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2836         makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2837         if (isPatternString) {
2838                 // Allow the first searched string to rebuild the keys too
2839                 keysBuilt = false;
2840         }
2841         else {
2842                 // no need to rebuild again
2843                 keysBuilt = true;
2844         }
2845 }
2846
2847 /*
2848  * Keep the list of actual opened parentheses actual
2849  * (e.g. depth == 4 means there are 4 '{' not processed yet)
2850  */
2851 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2852 {
2853         int skip = 0;
2854         for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2855                 char c;
2856                 c = par[i];
2857                 skip = 0;
2858                 if (c == '\\') skip = 1;
2859                 else if (c == '{') {
2860                         handleOpenP(i);
2861                 }
2862                 else if (c == '}') {
2863                         handleCloseP(i, closingAllowed);
2864                 }
2865         }
2866 }
2867
2868 #if (0)
2869 string Intervall::show(int lastpos)
2870 {
2871         int idx = 0;                          /* int intervalls */
2872         string s;
2873         int i = 0;
2874         if ((unsigned) lastpos > par.size())
2875                 lastpos = par.size();
2876         for (idx = 0; idx <= ignoreidx; idx++) {
2877                 while (i < lastpos) {
2878                         int printsize;
2879                         if (i <= borders[idx].low) {
2880                                 if (borders[idx].low > lastpos)
2881                                         printsize = lastpos - i;
2882                                 else
2883                                         printsize = borders[idx].low - i;
2884                                 s += par.substr(i, printsize);
2885                                 i += printsize;
2886                                 if (i >= borders[idx].low)
2887                                         i = borders[idx].upper;
2888                         }
2889                         else {
2890                                 i = borders[idx].upper;
2891                                 break;
2892                         }
2893                 }
2894         }
2895         if (lastpos > i) {
2896                 s += par.substr(i, lastpos-i);
2897         }
2898         return s;
2899 }
2900 #endif
2901
2902 void Intervall::output(ostringstream &os, int lastpos)
2903 {
2904         // get number of chars to output
2905         int idx = 0;                          /* int intervalls */
2906         int i = 0;
2907         int printed = 0;
2908         string startTitle = titleValue;
2909         for (idx = 0; idx <= ignoreidx; idx++) {
2910                 if (i < lastpos) {
2911                         if (i <= borders[idx].low) {
2912                                 int printsize;
2913                                 if (borders[idx].low > lastpos)
2914                                         printsize = lastpos - i;
2915                                 else
2916                                         printsize = borders[idx].low - i;
2917                                 if (printsize > 0) {
2918                                         os << startTitle << par.substr(i, printsize);
2919                                         i += printsize;
2920                                         printed += printsize;
2921                                         startTitle = "";
2922                                 }
2923                                 handleParentheses(i, false);
2924                                 if (i >= borders[idx].low)
2925                                         i = borders[idx].upper;
2926                         }
2927                         else {
2928                                 i = borders[idx].upper;
2929                         }
2930                 }
2931                 else
2932                         break;
2933         }
2934         if (lastpos > i) {
2935                 os << startTitle << par.substr(i, lastpos-i);
2936                 printed += lastpos-i;
2937         }
2938         handleParentheses(lastpos, false);
2939         int startindex;
2940         if (keys["foreignlanguage"].disabled)
2941                 startindex = actualdeptindex-langcount;
2942         else
2943                 startindex = actualdeptindex;
2944         for (int i = startindex; i > 0; --i) {
2945                 os << "}";
2946         }
2947         if (hasTitle && (printed > 0))
2948                 os << "}";
2949         if (! isPatternString_)
2950                 os << "\n";
2951         handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2952 }
2953
2954 void LatexInfo::processRegion(int start, int region_end)
2955 {
2956         while (start < region_end) {          /* Let {[} and {]} survive */
2957                 int cnt = interval_.isOpeningPar(start);
2958                 if (cnt == 1) {
2959                         // Closing is allowed past the region
2960                         int closing = interval_.findclosing(start+1, interval_.par.length());
2961                         interval_.addIntervall(start, start+1);
2962                         interval_.addIntervall(closing, closing+1);
2963                 }
2964                 else if (cnt == 3)
2965                         start += 2;
2966                 start = interval_.nextNotIgnored(start+1);
2967         }
2968 }
2969
2970 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2971 {
2972         if (actual.parenthesiscount == 0) {
2973                 // "{\tiny{} ...}" ==> "{{} ...}"
2974                 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2975         }
2976         else {
2977                 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2978                 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2979                 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2980         }
2981 }
2982
2983 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2984 {
2985         int nextKeyIdx = 0;
2986         switch (actual.keytype)
2987         {
2988         case KeyInfo::isTitle: {
2989                 removeHead(actual);
2990                 nextKeyIdx = getNextKey();
2991                 break;
2992         }
2993         case KeyInfo::cleanToStart: {
2994                 actual._dataEnd = actual._dataStart;
2995                 nextKeyIdx = getNextKey();
2996                 // Search for end of arguments
2997                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2998                 if (tmpIdx > 0) {
2999                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3000                                 entries_[i].disabled = true;
3001                         }
3002                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3003                 }
3004                 while (interval_.par[actual._dataEnd] == ' ')
3005                         actual._dataEnd++;
3006                 interval_.addIntervall(0, actual._dataEnd+1);
3007                 interval_.actualdeptindex = 0;
3008                 interval_.depts[0] = actual._dataEnd+1;
3009                 interval_.closes[0] = -1;
3010                 break;
3011         }
3012         case KeyInfo::isText:
3013                 interval_.par[actual._tokenstart] = '#';
3014                 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
3015                 nextKeyIdx = getNextKey();
3016                 break;
3017         case KeyInfo::noContent: {          /* char like "\hspace{2cm}" */
3018                 if (actual.disabled)
3019                         interval_.addIntervall(actual._tokenstart, actual._dataEnd);
3020                 else
3021                         interval_.addIntervall(actual._dataStart, actual._dataEnd);
3022         }
3023                 // fall through
3024         case KeyInfo::isChar: {
3025                 nextKeyIdx = getNextKey();
3026                 break;
3027         }
3028         case KeyInfo::isSize: {
3029                 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
3030                         if (actual.parenthesiscount == 0)
3031                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
3032                         else {
3033                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3034                         }
3035                         nextKeyIdx = getNextKey();
3036                 } else {
3037                         // Here _dataStart points to '{', so correct it
3038                         actual._dataStart += 1;
3039                         actual._tokensize += 1;
3040                         actual.parenthesiscount = 1;
3041                         if (interval_.par[actual._dataStart] == '}') {
3042                                 // Determine the end if used like '{\tiny{}...}'
3043                                 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
3044                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3045                         }
3046                         else {
3047                                 // Determine the end if used like '\tiny{...}'
3048                                 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
3049                         }
3050                         // Split on this key if not at start
3051                         int start = interval_.nextNotIgnored(previousStart);
3052                         if (start < actual._tokenstart) {
3053                                 interval_.output(os, actual._tokenstart);
3054                                 interval_.addIntervall(start, actual._tokenstart);
3055                         }
3056                         // discard entry if at end of actual
3057                         nextKeyIdx = process(os, actual);
3058                 }
3059                 break;
3060         }
3061         case KeyInfo::endArguments: {
3062                 // Remove trailing '{}' too
3063                 actual._dataStart += 1;
3064                 actual._dataEnd += 1;
3065                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3066                 nextKeyIdx = getNextKey();
3067                 break;
3068         }
3069         case KeyInfo::noMain:
3070                 // fall through
3071         case KeyInfo::isStandard: {
3072                 if (actual.disabled) {
3073                         removeHead(actual);
3074                         processRegion(actual._dataStart, actual._dataStart+1);
3075                         nextKeyIdx = getNextKey();
3076                 } else {
3077                         // Split on this key if not at datastart of calling entry
3078                         int start = interval_.nextNotIgnored(previousStart);
3079                         if (start < actual._tokenstart) {
3080                                 interval_.output(os, actual._tokenstart);
3081                                 interval_.addIntervall(start, actual._tokenstart);
3082                         }
3083                         // discard entry if at end of actual
3084                         nextKeyIdx = process(os, actual);
3085                 }
3086                 break;
3087         }
3088         case KeyInfo::removeWithArg: {
3089                 nextKeyIdx = getNextKey();
3090                 // Search for end of arguments
3091                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3092                 if (tmpIdx > 0) {
3093                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3094                                 entries_[i].disabled = true;
3095                         }
3096                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3097                 }
3098                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3099                 break;
3100         }
3101         case KeyInfo::doRemove: {
3102                 // Remove the key with all parameters and following spaces
3103                 size_t pos;
3104                 size_t start;
3105                 if (interval_.par[actual._dataEnd-1] == ' ' || interval_.par[actual._dataEnd-1] == '}')
3106                         start = actual._dataEnd;
3107                 else
3108                         start = actual._dataEnd+1;
3109                 for (pos = start; pos < interval_.par.length(); pos++) {
3110                         if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
3111                                 break;
3112                 }
3113                 // Remove also enclosing parentheses [] and {}
3114                 int numpars = 0;
3115                 int spaces = 0;
3116                 while (actual._tokenstart > numpars) {
3117                         if (pos+numpars >= interval_.par.size())
3118                                 break;
3119                         else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
3120                                 numpars++;
3121                         else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
3122                                 numpars++;
3123                         else
3124                                 break;
3125                 }
3126                 if (numpars > 0) {
3127                         if (interval_.par[pos+numpars] == ' ')
3128                                 spaces++;
3129                 }
3130
3131                 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
3132                 nextKeyIdx = getNextKey();
3133                 break;
3134         }
3135         case KeyInfo::isList: {
3136                 // Discard space before _tokenstart
3137                 int count;
3138                 for (count = 0; count < actual._tokenstart; count++) {
3139                         if (interval_.par[actual._tokenstart-count-1] != ' ')
3140                                 break;
3141                 }
3142                 nextKeyIdx = getNextKey();
3143                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3144                 if (tmpIdx > 0) {
3145                         // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
3146                         // with arguments
3147                         // How else can we catch this one?
3148                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3149                                 entries_[i].disabled = true;
3150                         }
3151                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3152                 }
3153                 else if (nextKeyIdx > 0) {
3154                         // Ignore any lang entries inside data region
3155                         for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
3156                                 if (entries_[i].keytype == KeyInfo::isMain)
3157                                         entries_[i].disabled = true;
3158                         }
3159                 }
3160                 if (actual.disabled) {
3161                         interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
3162                 }
3163                 else {
3164                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3165                 }
3166                 if (interval_.par[actual._dataEnd+1] == '[') {
3167                         int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
3168                         if ((interval_.par[actual._dataEnd+2] == '{') &&
3169                                         (interval_.par[posdown-1] == '}')) {
3170                                 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
3171                                 interval_.addIntervall(posdown-1, posdown+1);
3172                         }
3173                         else {
3174                                 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
3175                                 interval_.addIntervall(posdown, posdown+1);
3176                         }
3177                         int blk = interval_.nextNotIgnored(actual._dataEnd+1);
3178                         if (blk > posdown) {
3179                                 // Discard at most 1 space after empty item
3180                                 int count;
3181                                 for (count = 0; count < 1; count++) {
3182                                         if (interval_.par[blk+count] != ' ')
3183                                                 break;
3184                                 }
3185                                 if (count > 0)
3186                                         interval_.addIntervall(blk, blk+count);
3187                         }
3188                 }
3189                 break;
3190         }
3191         case KeyInfo::isSectioning: {
3192                 // Discard spaces before _tokenstart
3193                 int count;
3194                 int val = actual._tokenstart;
3195                 for (count = 0; count < actual._tokenstart;) {
3196                         val = interval_.previousNotIgnored(val-1);
3197                         if (val < 0 || interval_.par[val] != ' ')
3198                                 break;
3199                         else {
3200                                 count = actual._tokenstart - val;
3201                         }
3202                 }
3203                 if (actual.disabled) {
3204                         removeHead(actual, count);
3205                         nextKeyIdx = getNextKey();
3206                 } else {
3207                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3208                         nextKeyIdx = process(os, actual);
3209                 }
3210                 break;
3211         }
3212         case KeyInfo::isMath: {
3213                 // Same as regex, use the content unchanged
3214                 nextKeyIdx = getNextKey();
3215                 break;
3216         }
3217         case KeyInfo::isRegex: {
3218                 // DO NOT SPLIT ON REGEX
3219                 // Do not disable
3220                 nextKeyIdx = getNextKey();
3221                 break;
3222         }
3223         case KeyInfo::isIgnored: {
3224                 // Treat like a character for now
3225                 nextKeyIdx = getNextKey();
3226                 break;
3227         }
3228         case KeyInfo::isMain: {
3229                 if (interval_.par.substr(actual._dataStart, 2) == "% ")
3230                         interval_.addIntervall(actual._dataStart, actual._dataStart+2);
3231                 if (actual._tokenstart > 0) {
3232                         int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
3233                         if ((prev >= 0) && interval_.par[prev] == '%')
3234                                 interval_.addIntervall(prev, prev+1);
3235                 }
3236                 if (actual.disabled) {
3237                         removeHead(actual);
3238                         interval_.langcount++;
3239                         if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
3240                                         (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
3241                                 // Discard also the space before math-equation
3242                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3243                         }
3244                         nextKeyIdx = getNextKey();
3245                         // interval.resetOpenedP(actual._dataStart-1);
3246                 }
3247                 else {
3248                         if (actual._tokenstart < 26) {
3249                                 // for the first (and maybe dummy) language
3250                                 interval_.setForDefaultLang(actual);
3251                         }
3252                         interval_.resetOpenedP(actual._dataStart-1);
3253                 }
3254                 break;
3255         }
3256         case KeyInfo::invalid:
3257         case KeyInfo::headRemove:
3258                 // These two cases cannot happen, already handled
3259                 // fall through
3260         default: {
3261                 // LYXERR(Debug::INFO, "Unhandled keytype");
3262                 nextKeyIdx = getNextKey();
3263                 break;
3264         }
3265         }
3266         return nextKeyIdx;
3267 }
3268
3269 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
3270 {
3271         int end = interval_.nextNotIgnored(actual._dataEnd);
3272         int oldStart = actual._dataStart;
3273         int nextKeyIdx = getNextKey();
3274         while (true) {
3275                 if ((nextKeyIdx < 0) ||
3276                                 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
3277                                 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
3278                         if (oldStart <= end) {
3279                                 processRegion(oldStart, end);
3280                                 oldStart = end+1;
3281                         }
3282                         break;
3283                 }
3284                 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
3285
3286                 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
3287                         (void) dispatch(os, actual._dataStart, nextKey);
3288                         end = nextKey._tokenstart;
3289                         break;
3290                 }
3291                 processRegion(oldStart, nextKey._tokenstart);
3292                 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
3293
3294                 oldStart = nextKey._dataEnd+1;
3295         }
3296         // now nextKey is either invalid or is outside of actual._dataEnd
3297         // output the remaining and discard myself
3298         if (oldStart <= end) {
3299                 processRegion(oldStart, end);
3300         }
3301         if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
3302                 end += 1;
3303                 // This is the normal case.
3304                 // But if using the firstlanguage, the closing may be missing
3305         }
3306         // get minimum of 'end' and  'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
3307         int output_end;
3308         if (actual._dataEnd < end)
3309                 output_end = interval_.nextNotIgnored(actual._dataEnd);
3310         else if (interval_.par.size() > (size_t) end)
3311                 output_end = interval_.nextNotIgnored(end);
3312         else
3313                 output_end = interval_.par.size();
3314         if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
3315                 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
3316         }
3317         // Remove possible empty data
3318         int dstart = interval_.nextNotIgnored(actual._dataStart);
3319         while (interval_.isOpeningPar(dstart) == 1) {
3320                 interval_.addIntervall(dstart, dstart+1);
3321                 int dend = interval_.findclosing(dstart+1, output_end);
3322                 interval_.addIntervall(dend, dend+1);
3323                 dstart = interval_.nextNotIgnored(dstart+1);
3324         }
3325         if (dstart < output_end)
3326                 interval_.output(os, output_end);
3327         if (nextKeyIdx < 0)
3328                 interval_.addIntervall(0, end);
3329         else
3330                 interval_.addIntervall(actual._tokenstart, end);
3331         return nextKeyIdx;
3332 }
3333
3334 string splitOnKnownMacros(string par, bool isPatternString)
3335 {
3336         ostringstream os;
3337         LatexInfo li(par, isPatternString);
3338         // LYXERR(Debug::INFO, "Berfore split: " << par);
3339         KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
3340         DummyKey.head = "";
3341         DummyKey._tokensize = 0;
3342         DummyKey._dataStart = 0;
3343         DummyKey._dataEnd = par.length();
3344         DummyKey.disabled = true;
3345         int firstkeyIdx = li.getFirstKey();
3346         string s;
3347         if (firstkeyIdx >= 0) {
3348                 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
3349                 DummyKey._tokenstart = firstKey._tokenstart;
3350                 int nextkeyIdx;
3351                 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
3352                         // Use dummy firstKey
3353                         firstKey = DummyKey;
3354                         (void) li.setNextKey(firstkeyIdx);
3355                 }
3356                 else {
3357                         if (par.substr(firstKey._dataStart, 2) == "% ")
3358                                 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
3359                 }
3360                 nextkeyIdx = li.process(os, firstKey);
3361                 while (nextkeyIdx >= 0) {
3362                         // Check for a possible gap between the last
3363                         // entry and this one
3364                         int datastart = li.nextNotIgnored(firstKey._dataStart);
3365                         KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
3366                         if ((nextKey._tokenstart > datastart)) {
3367                                 // Handle the gap
3368                                 firstKey._dataStart = datastart;
3369                                 firstKey._dataEnd = par.length();
3370                                 (void) li.setNextKey(nextkeyIdx);
3371                                 // Fake the last opened parenthesis
3372                                 li.setForDefaultLang(firstKey);
3373                                 nextkeyIdx = li.process(os, firstKey);
3374                         }
3375                         else {
3376                                 if (nextKey.keytype != KeyInfo::isMain) {
3377                                         firstKey._dataStart = datastart;
3378                                         firstKey._dataEnd = nextKey._dataEnd+1;
3379                                         (void) li.setNextKey(nextkeyIdx);
3380                                         li.setForDefaultLang(firstKey);
3381                                         nextkeyIdx = li.process(os, firstKey);
3382                                 }
3383                                 else {
3384                                         nextkeyIdx = li.process(os, nextKey);
3385                                 }
3386                         }
3387                 }
3388                 // Handle the remaining
3389                 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
3390                 firstKey._dataEnd = par.length();
3391                 // Check if ! empty
3392                 if ((firstKey._dataStart < firstKey._dataEnd) &&
3393                                 (par[firstKey._dataStart] != '}')) {
3394                         li.setForDefaultLang(firstKey);
3395                         (void) li.process(os, firstKey);
3396                 }
3397                 s = os.str();
3398                 // return string definitelly impossible to match, but should be known
3399         }
3400         else
3401                 s = par;                            /* no known macros found */
3402         // LYXERR(Debug::INFO, "After split: " << s);
3403         return s;
3404 }
3405
3406 /*
3407  * Try to unify the language specs in the latexified text.
3408  * Resulting modified string is set to "", if
3409  * the searched tex does not contain all the features in the search pattern
3410  */
3411 static string correctlanguagesetting(string par, bool isPatternString, bool withformat, lyx::Buffer *pbuf = nullptr)
3412 {
3413         static Features regex_f;
3414         static int missed = 0;
3415         static bool regex_with_format = false;
3416
3417         int parlen = par.length();
3418
3419         while ((parlen > 0) && (par[parlen-1] == '\n')) {
3420                 parlen--;
3421         }
3422 #if 0
3423         if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
3424                 // Happens to be there in case of description or labeling environment
3425                 parlen--;
3426         }
3427 #endif
3428         string result;
3429         if (withformat) {
3430                 // Split the latex input into pieces which
3431                 // can be digested by our search engine
3432                 LYXERR(Debug::FINDVERBOSE, "input: \"" << par << "\"");
3433                 if (isPatternString && (pbuf != nullptr)) { // Check if we should disable/enable test for language
3434                         // We check for polyglossia, because in runparams.flavor we use Flavor::XeTeX
3435                         string doclang = pbuf->params().language->polyglossia();
3436                         static regex langre("\\\\(foreignlanguage)\\{([^\\}]+)\\}");
3437                         smatch sub;
3438                         bool toIgnoreLang = true;
3439                         for (sregex_iterator it(par.begin(), par.end(), langre), end; it != end; ++it) {
3440                                 sub = *it;
3441                                 if (sub.str(2) != doclang) {
3442                                         toIgnoreLang = false;
3443                                         break;
3444                                 }
3445                         }
3446                         setIgnoreFormat("language", toIgnoreLang, false);
3447
3448                 }
3449                 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
3450                 LYXERR(Debug::FINDVERBOSE, "After splitOnKnownMacros:\n\"" << result << "\"");
3451         }
3452         else
3453                 result = par.substr(0, parlen);
3454         if (isPatternString) {
3455                 missed = 0;
3456                 if (withformat) {
3457                         regex_f = identifyFeatures(result);
3458                         string features = "";
3459                         for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3460                                 string a = it->first;
3461                                 regex_with_format = true;
3462                                 features += " " + a;
3463                                 // LYXERR(Debug::INFO, "Identified regex format:" << a);
3464                         }
3465                         LYXERR(Debug::FINDVERBOSE, "Identified Features" << features);
3466
3467                 }
3468         } else if (regex_with_format) {
3469                 Features info = identifyFeatures(result);
3470                 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3471                         string a = it->first;
3472                         bool b = it->second;
3473                         if (b && ! info[a]) {
3474                                 missed++;
3475                                 LYXERR(Debug::FINDVERBOSE, "Missed(" << missed << " " << a <<", srclen = " << parlen );
3476                                 return "";
3477                         }
3478                 }
3479
3480         }
3481         else {
3482                 // LYXERR(Debug::INFO, "No regex formats");
3483         }
3484         return result;
3485 }
3486
3487
3488 // Remove trailing closure of math, macros and environments, so to catch parts of them.
3489 static void identifyClosing(string & t, bool ignoreformat)
3490 {
3491         do {
3492                 LYXERR(Debug::FINDVERBOSE, "identifyClosing(): t now is '" << t << "'");
3493                 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
3494                         continue;
3495                 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
3496                         continue;
3497                 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]+\\*?\\}$", "$1"))
3498                         continue;
3499                 if (! ignoreformat) {
3500                         if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1"))
3501                                 continue;
3502                 }
3503                 break;
3504         } while (true);
3505         return;
3506 }
3507
3508 static int num_replaced = 0;
3509 static bool previous_single_replace = true;
3510
3511 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
3512 {
3513 #if QTSEARCH
3514         if (regexp_str.empty() || regexp2_str.empty()) {
3515                 regexIsValid = false;
3516                 regexError = "Invalid empty regex";
3517                 return;
3518         }
3519         // Handle \w properly
3520         QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3521         if (! opt.casesensitive) {
3522                 popts |= QRegularExpression::CaseInsensitiveOption;
3523         }
3524         regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3525         regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3526         regexError = "";
3527         if (regexp.isValid() && regexp2.isValid()) {
3528                 regexIsValid = true;
3529                 // Check '{', '}' pairs inside the regex
3530                 int balanced = 0;
3531                 int skip = 1;
3532                 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3533                         char c = par_as_string[i];
3534                         if (c == '\\') {
3535                                 skip = 2;
3536                                 continue;
3537                         }
3538                         if (c == '{')
3539                                 balanced++;
3540                         else if (c == '}') {
3541                                 balanced--;
3542                                 if (balanced < 0)
3543                                         break;
3544                         }
3545                         skip = 1;
3546                 }
3547                 if (balanced != 0) {
3548                         regexIsValid = false;
3549                         regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3550                 }
3551         }
3552         else {
3553                 regexIsValid = false;
3554                 if (!regexp.isValid())
3555                         regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3556                 else
3557                         regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3558         }
3559 #else
3560         (void)par_as_string;
3561         if (opt.casesensitive) {
3562                 regexp = regex(regexp_str);
3563                 regexp2 = regex(regexp2_str);
3564         }
3565         else {
3566                 regexp = regex(regexp_str, std::regex_constants::icase);
3567                 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3568         }
3569 #endif
3570 }
3571
3572 static void modifyRegexForMatchWord(string &t)
3573 {
3574         string s("");
3575         regex wordre("(\\\\)*((\\.|\\\\b))");
3576         size_t lastpos = 0;
3577         smatch sub;
3578         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3579                 sub = *it;
3580                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3581                         continue;
3582                 }
3583                 else if (sub.str(2) == "\\\\b")
3584                         return;
3585                 if (lastpos < (size_t) sub.position(2))
3586                         s += t.substr(lastpos, sub.position(2) - lastpos);
3587                 s += "\\S";
3588                 lastpos = sub.position(2) + sub.length(2);
3589         }
3590         if (lastpos == 0) {
3591                 s = "\\b" + t + "\\b";
3592                 t = s;
3593                 return;
3594         }
3595         else if (lastpos < t.length())
3596                 s += t.substr(lastpos, t.length() - lastpos);
3597         t = "\\b" + s + "\\b";
3598 }
3599
3600 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3601         : p_buf(&buf), p_first_buf(&buf), opt(opt)
3602 {
3603         Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3604         docstring const & ds = stringifySearchBuffer(find_buf, opt);
3605         if (ds.empty() ) {
3606                 CreateRegexp(opt, "", "", "");
3607                 return;
3608         }
3609         use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3610         if (opt.replace_all && previous_single_replace) {
3611                 previous_single_replace = false;
3612                 num_replaced = 0;
3613         }
3614         else if (!opt.replace_all) {
3615                 num_replaced = 0;       // count number of replaced strings
3616                 previous_single_replace = true;
3617         }
3618         // When using regexp, braces are hacked already by escape_for_regex()
3619         par_as_string = normalize(ds, opt.ignoreformat);
3620         open_braces = 0;
3621         close_wildcards = 0;
3622
3623         size_t lead_size = 0;
3624         // correct the language settings
3625         par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat, &buf);
3626         if (par_as_string.empty()) {
3627                 CreateRegexp(opt, "", "", "");
3628                 return;
3629         }
3630         opt.matchAtStart = false;
3631         if (!use_regexp) {
3632                 identifyClosing(par_as_string, opt.ignoreformat); // Removes math closings ($, ], ...) at end of string
3633                 if (opt.ignoreformat) {
3634                         lead_size = 0;
3635                 }
3636                 else {
3637                         lead_size = identifyLeading(par_as_string);
3638                 }
3639                 lead_as_string = par_as_string.substr(0, lead_size);
3640                 string lead_as_regex_string = string2regex(lead_as_string);
3641                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3642                 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3643                 /* Handle whole words too in this case
3644                 */
3645                 if (opt.matchword) {
3646                         par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3647                         opt.matchword = false;
3648                 }
3649                 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3650                 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3651                 CreateRegexp(opt, regexp_str, regexp2_str);
3652                 use_regexp = true;
3653                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3654                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3655                 return;
3656         }
3657
3658         if (!opt.ignoreformat) {
3659                 lead_size = identifyLeading(par_as_string);
3660                 LYXERR(Debug::FINDVERBOSE, "Lead_size: " << lead_size);
3661                 lead_as_string = par_as_string.substr(0, lead_size);
3662                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3663         }
3664
3665         // Here we are using regexp
3666         LASSERT(use_regexp, /**/);
3667         {
3668                 string lead_as_regexp;
3669                 if (lead_size > 0) {
3670                         lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3671                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "\\$$", "");
3672                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3673                         par_as_string = par_as_string_nolead;
3674                         LYXERR(Debug::FINDVERBOSE, "lead_as_regexp is '" << lead_as_regexp << "'");
3675                         LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3676                 }
3677                 // LYXERR(Debug::FINDVERBOSE, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3678                 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3679                 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3680                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3681                 ++close_wildcards;
3682                 size_t lng = par_as_string.size();
3683                 if (!opt.ignoreformat) {
3684                         // Remove extra '\}' at end if not part of \{\.\}
3685                         while(lng > 2) {
3686                                 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3687                                         if (lng >= 6) {
3688                                                 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3689                                                         break;
3690                                         }
3691                                         lng -= 2;
3692                                         open_braces++;
3693                                 }
3694                                 else
3695                                         break;
3696                         }
3697                         if (lng < par_as_string.size())
3698                                 par_as_string = par_as_string.substr(0,lng);
3699                 }
3700                 LYXERR(Debug::FINDVERBOSE, "par_as_string after correctRegex is '" << par_as_string << "'");
3701                 if ((lng > 0) && (par_as_string[0] == '^')) {
3702                         par_as_string = par_as_string.substr(1);
3703                         --lng;
3704                         opt.matchAtStart = true;
3705                 }
3706                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3707                 // LYXERR(Debug::FINDVERBOSE, "Open braces: " << open_braces);
3708                 // LYXERR(Debug::FINDVERBOSE, "Replaced text (to be used as regex): " << par_as_string);
3709
3710                 // If entered regexp must match at begin of searched string buffer
3711                 // Kornel: Added parentheses to use $1 for size of the leading string
3712                 string regexp_str;
3713                 string regexp2_str;
3714                 {
3715                         // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3716                         // Unfortunately is '\1', '\2', etc not working for strings with extra format
3717                         // so the convert has no effect in that case
3718                         for (int i = 7; i > 0; --i) {
3719                                 string orig = "\\\\" + std::to_string(i);
3720                                 string dest = "\\" + std::to_string(i+2);
3721                                 while (regex_replace(par_as_string, par_as_string, orig, dest));
3722                         }
3723                         if (opt.matchword) {
3724                                 modifyRegexForMatchWord(par_as_string);
3725                                 opt.matchword = false;
3726                         }
3727                         regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3728                         regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3729                 }
3730                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3731                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3732                 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3733         }
3734 }
3735
3736 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3737 {
3738         MatchResult mres;
3739
3740         mres.searched_size = len;
3741         if (at_begin &&
3742                         (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3743                 return mres;
3744
3745         docstring docstr = stringifyFromForSearch(opt, cur, len);
3746         string str;
3747         str = normalize(docstr, opt.ignoreformat);
3748         if (!opt.ignoreformat) {
3749                 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3750                 // remove closing '}' and '\n' to allow for use of '$' in regex
3751                 size_t lng = str.size();
3752                 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3753                         lng--;
3754                 if (lng != str.size()) {
3755                         str = str.substr(0, lng);
3756                 }
3757                 // Replace occurences of '~' to ' '
3758                 static std::regex specialChars { R"(~)" };
3759                 str = std::regex_replace(str, specialChars,  R"( )" );
3760         }
3761         if (str.empty()) {
3762                 mres.match_len = -1;
3763                 return mres;
3764         }
3765         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
3766
3767         LASSERT(use_regexp, /**/);
3768         {
3769                 // use_regexp always true
3770                 LYXERR(Debug::FINDVERBOSE, "Searching in regexp mode: at_begin=" << at_begin);
3771 #if QTSEARCH
3772                 QString qstr = QString::fromStdString(str);
3773                 QRegularExpression const *p_regexp;
3774                 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3775                 if (at_begin) {
3776                         p_regexp = &regexp;
3777                 } else {
3778                         p_regexp = &regexp2;
3779                 }
3780                 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3781                 if (!match.hasMatch())
3782                         return mres;
3783 #else
3784                 regex const *p_regexp;
3785                 regex_constants::match_flag_type flags;
3786                 if (at_begin) {
3787                         flags = regex_constants::match_continuous;
3788                         p_regexp = &regexp;
3789                 } else {
3790                         flags = regex_constants::match_default;
3791                         p_regexp = &regexp2;
3792                 }
3793                 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3794                 if (re_it == sregex_iterator())
3795                         return mres;
3796                 match_results<string::const_iterator> const & m = *re_it;
3797 #endif
3798                 // Whole found string, including the leading
3799                 // std: m[0].second - m[0].first
3800                 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3801                 //
3802                 // Size of the leading string
3803                 // std: m[1].second - m[1].first
3804                 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3805                 int leadingsize = 0;
3806 #if QTSEARCH
3807                 if (match.lastCapturedIndex() > 0) {
3808                         leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3809                 }
3810
3811 #else
3812                 if (m.size() > 2) {
3813                         leadingsize = m[1].second - m[1].first;
3814                 }
3815 #endif
3816 #if QTSEARCH
3817                 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3818                 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3819                 // because of different number of closing at end of string
3820                 // we have to 'unify' the length of the post-match.
3821                 // Done by ignoring closing parenthesis and linefeeds at string end
3822                 int matchend = match.capturedEnd(0);
3823                 size_t strsize = qstr.size();
3824                 if (!opt.ignoreformat) {
3825                         while (mres.match_len > 1) {
3826                                 QChar c = qstr.at(matchend - 1);
3827                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3828                                         mres.match_len--;
3829                                         matchend--;
3830                                 }
3831                                 else
3832                                         break;
3833                         }
3834                         while (strsize > (size_t) match.capturedEnd(0)) {
3835                                 QChar c = qstr.at(strsize-1);
3836                                 if ((c == '\n') || (c == '}')) {
3837                                         --strsize;
3838                                 }
3839                                 else
3840                                         break;
3841                         }
3842                 }
3843                 // LYXERR0(qstr.toStdString());
3844                 mres.match2end = strsize - matchend;
3845                 mres.pos = match.capturedStart(2);
3846 #else
3847                 mres.match_prefix = m[2].second - m[2].first;
3848                 mres.match_len = m[0].second - m[2].second;
3849                 // ignore closing parenthesis and linefeeds at string end
3850                 size_t strend = m[0].second - m[0].first;
3851                 int matchend = strend;
3852                 size_t strsize = str.size();
3853                 if (!opt.ignoreformat) {
3854                         while (mres.match_len > 1) {
3855                                 char c = str.at(matchend - 1);
3856                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3857                                         mres.match_len--;
3858                                         matchend--;
3859                                 }
3860                                 else
3861                                         break;
3862                         }
3863                         while (strsize > strend) {
3864                                 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3865                                         --strsize;
3866                                 }
3867                                 else
3868                                         break;
3869                         }
3870                 }
3871                 // LYXERR0(str);
3872                 mres.match2end = strsize - matchend;
3873                 mres.pos = m[2].first - m[0].first;;
3874 #endif
3875                 if (mres.match2end < 0)
3876                         mres.match_len = 0;
3877                 mres.leadsize = leadingsize;
3878 #if QTSEARCH
3879                 if (mres.match_len > 0) {
3880                         string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3881                         mres.result.push_back(a0);
3882                         for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3883                                 mres.result.push_back(match.captured(i).toStdString());
3884                         }
3885                 }
3886 #else
3887                 if (mres.match_len > 0) {
3888                         string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3889                         mres.result.push_back(a0);
3890                         for (size_t i = 3; i < m.size(); i++) {
3891                                 mres.result.push_back(m[i]);
3892                         }
3893                 }
3894 #endif
3895                 return mres;
3896         }
3897 }
3898
3899
3900 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3901 {
3902         MatchResult mres = findAux(cur, len, at_begin);
3903         int res = mres.match_len;
3904         LYXERR(Debug::FINDVERBOSE,
3905                "res=" << res << ", at_begin=" << at_begin
3906                << ", matchAtStart=" << opt.matchAtStart
3907                << ", inTexted=" << cur.inTexted());
3908         if (opt.matchAtStart) {
3909                 if (cur.pos() != 0)
3910                         mres.match_len = 0;
3911                 else if (mres.match_prefix > 0)
3912                         mres.match_len = 0;
3913                 return mres;
3914         }
3915         else
3916                 return mres;
3917 }
3918
3919 #if 0
3920 static bool simple_replace(string &t, string from, string to)
3921 {
3922         regex repl("(\\\\)*(" + from + ")");
3923         string s("");
3924         size_t lastpos = 0;
3925         smatch sub;
3926         for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3927                 sub = *it;
3928                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3929                         continue;
3930                 if (lastpos < (size_t) sub.position(2))
3931                         s += t.substr(lastpos, sub.position(2) - lastpos);
3932                 s += to;
3933                 lastpos = sub.position(2) + sub.length(2);
3934         }
3935         if (lastpos == 0)
3936                 return false;
3937         else if (lastpos < t.length())
3938                 s += t.substr(lastpos, t.length() - lastpos);
3939         t = s;
3940         return true;
3941 }
3942 #endif
3943
3944 string MatchStringAdv::normalize(docstring const & s, bool ignore_format) const
3945 {
3946         string t;
3947         t = lyx::to_utf8(s);
3948         // Remove \n at begin
3949         while (!t.empty() && t[0] == '\n')
3950                 t = t.substr(1);
3951         // Remove \n* at end
3952         while (!t.empty() && t[t.size() - 1] == '\n') {
3953                 t = t.substr(0, t.size() - 1);
3954         }
3955         size_t pos;
3956         // Handle all other '\n'
3957         while ((pos = t.find("\n")) != string::npos) {
3958                 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3959                         // Handle '\\\n'
3960                         if (isAlnumASCII(t[pos+1])) {
3961                                 t.replace(pos-2, 3, " ");
3962                         }
3963                         else {
3964                                 t.replace(pos-2, 3, "");
3965                         }
3966                 }
3967                 else {
3968                         if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3969                                 // '\n' adjacent to non-alpha-numerics, discard
3970                                 t.replace(pos, 1, "");
3971                         }
3972                         else {
3973                                 // Replace all other \n with spaces
3974                                 t.replace(pos, 1, " ");
3975                         }
3976                         if (!ignore_format) {
3977                                 size_t count = 0;
3978                                 while ((pos > count + 1) && (t[pos - 1 -count] == '%')) {
3979                                         count++;
3980                                 }
3981                                 if (count > 0) {
3982                                         t.replace(pos - count, count, "");
3983                                 }
3984                         }
3985                 }
3986         }
3987         // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3988         // Kornel: Added textsl, textsf, textit, texttt and noun
3989         // + allow to seach for colored text too
3990         LYXERR(Debug::FINDVERBOSE, "Removing stale empty macros from: " << t);
3991         while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3992                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3993         while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3994                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\section{}, \\part{}, \\paragraph{} macros from: " << t);
3995         while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3996
3997         return t;
3998 }
3999
4000
4001 docstring stringifyFromCursor(DocIterator const & cur, int len)
4002 {
4003         LYXERR(Debug::FINDVERBOSE, "Stringifying with len=" << len << " from cursor at pos: " << cur);
4004         if (cur.inTexted()) {
4005                 Paragraph const & par = cur.paragraph();
4006                 // TODO what about searching beyond/across paragraph breaks ?
4007                 // TODO Try adding a AS_STR_INSERTS as last arg
4008                 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
4009                                         int(par.size()) : cur.pos() + len;
4010                 // OutputParams runparams(&cur.buffer()->params().encoding());
4011                 OutputParams runparams(encodings.fromLyXName("utf8"));
4012                 runparams.nice = true;
4013                 setFindParams(runparams);
4014                 int option = AS_STR_INSETS | AS_STR_PLAINTEXT;
4015                 if (ignoreFormats.getDeleted()) {
4016                         option |= AS_STR_SKIPDELETE;
4017                         runparams.for_search = OutputParams::SearchWithoutDeleted;
4018                 }
4019                 else {
4020                         runparams.for_search = OutputParams::SearchWithDeleted;
4021                 }
4022                 if (ignoreFormats.getNonContent()) {
4023                         runparams.for_search |= OutputParams::SearchNonOutput;
4024                 }
4025                 LYXERR(Debug::FINDVERBOSE, "Stringifying with cur: "
4026                        << cur << ", from pos: " << cur.pos() << ", end: " << end);
4027                 docstring res = from_utf8(latexNamesToUtf8(par.asString(cur.pos(), end,
4028                                                                         option,
4029                                                                         &runparams), false));
4030                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Stringified text from pos(" << cur.pos() << ") len(" << len << "): " << res);
4031                 return res;
4032         } else if (cur.inMathed()) {
4033                 CursorSlice cs = cur.top();
4034                 MathData md = cs.cell();
4035                 MathData::const_iterator it_end =
4036                                 (( len == -1 || cs.pos() + len > int(md.size()))
4037                                  ? md.end()
4038                                  : md.begin() + cs.pos() + len );
4039                 MathData md2;
4040                 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
4041                         md2.push_back(*it);
4042                 docstring res = from_utf8(latexNamesToUtf8(asString(md2), false));
4043                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Stringified math from pos(" << cur.pos() << ") len(" << len << "): " << res);
4044                 return res;
4045         }
4046         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
4047         return docstring();
4048 }
4049
4050 /** Computes the LaTeX export of buf starting from cur and ending len positions
4051  * after cur, if len is positive, or at the paragraph or innermost inset end
4052  * if len is -1.
4053  */
4054 docstring latexifyFromCursor(DocIterator const & cur, int len)
4055 {
4056         /*
4057         LYXERR(Debug::FINDVERBOSE, "Latexifying with len=" << len << " from cursor at pos: " << cur);
4058         LYXERR(Debug::FINDVERBOSE, "  with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
4059                << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
4060         */
4061         Buffer const & buf = *cur.buffer();
4062
4063         odocstringstream ods;
4064         otexstream os(ods);
4065         //OutputParams runparams(&buf.params().encoding());
4066         OutputParams runparams(encodings.fromLyXName("utf8"));
4067         runparams.nice = false;
4068         setFindParams(runparams);
4069         if (ignoreFormats.getDeleted()) {
4070                 runparams.for_search = OutputParams::SearchWithoutDeleted;
4071         }
4072         else {
4073                 runparams.for_search = OutputParams::SearchWithDeleted;
4074         }
4075         if (ignoreFormats.getNonContent()) {
4076                 runparams.for_search |= OutputParams::SearchNonOutput;
4077         }
4078
4079         if (cur.inTexted()) {
4080                 // @TODO what about searching beyond/across paragraph breaks ?
4081                 pos_type endpos = cur.paragraph().size();
4082                 if (len != -1 && endpos > cur.pos() + len)
4083                         endpos = cur.pos() + len;
4084                 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
4085                           string(), cur.pos(), endpos, true);
4086                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Latexified text from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4087                 return(ods.str());
4088         } else if (cur.inMathed()) {
4089                 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
4090                 for (int s = cur.depth() - 1; s >= 0; --s) {
4091                         CursorSlice const & cs = cur[s];
4092                         if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
4093                                 TeXMathStream ws(os);
4094                                 cs.asInsetMath()->asHullInset()->header_write(ws);
4095                                 break;
4096                         }
4097                 }
4098
4099                 CursorSlice const & cs = cur.top();
4100                 MathData md = cs.cell();
4101                 MathData::const_iterator it_end =
4102                                 ((len == -1 || cs.pos() + len > int(md.size()))
4103                                  ? md.end()
4104                                  : md.begin() + cs.pos() + len);
4105                 MathData md2;
4106                 for (MathData::const_iterator it = md.begin() + cs.pos();
4107                      it != it_end; ++it)
4108                         md2.push_back(*it);
4109
4110                 ods << asString(md2);
4111                 // Retrieve the math environment type, and add '$' or '$]'
4112                 // or others (\end{equation}) accordingly
4113                 for (int s = cur.depth() - 1; s >= 0; --s) {
4114                         CursorSlice const & cs2 = cur[s];
4115                         InsetMath * inset = cs2.asInsetMath();
4116                         if (inset && inset->asHullInset()) {
4117                                 TeXMathStream ws(os);
4118                                 inset->asHullInset()->footer_write(ws);
4119                                 break;
4120                         }
4121                 }
4122                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Latexified math from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4123         } else {
4124                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
4125         }
4126         return ods.str();
4127 }
4128
4129 #if defined(ResultsDebug)
4130 // Debugging output
4131 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
4132 {
4133         LYXERR0( "from:\t\t\t" << from);
4134         string status;
4135         if (mres.pos_len > 0) {
4136                 // Set in finalize
4137                 status = "FINALSEARCH";
4138         }
4139         else {
4140                 if (mres.match_len > 0) {
4141                         if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
4142                                 status = "Good Match";
4143                         else
4144                                 status = "Matched in";
4145                 }
4146                 else
4147                         status = "MissedSearch";
4148         }
4149
4150         LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
4151         if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
4152                 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
4153         if ((mres.pos > 0) || (mres.match_prefix > 0))
4154                 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
4155         for (size_t i = 0; i < mres.result.size(); i++)
4156                 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
4157 }
4158 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
4159 #else
4160 #define displayMres(s, txt, cur)
4161 #endif
4162
4163 /** Finalize an advanced find operation, advancing the cursor to the innermost
4164  ** position that matches, plus computing the length of the matching text to
4165  ** be selected
4166  ** Return the cur.pos() difference between start and end of found match
4167  **/
4168 MatchResult findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
4169 {
4170         // Search the foremost position that matches (avoids find of entire math
4171         // inset when match at start of it)
4172         DocIterator old_cur(cur.buffer());
4173         MatchResult mres;
4174         static MatchResult fail = MatchResult();
4175         MatchResult max_match;
4176         // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
4177         // Happens with e.g. hyperlinks
4178         // either one sees "http://www.bla.bla" or nothing
4179         // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
4180         // and although we search for only 3 chars, we find the whole hyperlink inset
4181         bool at_begin = (expected.match_prefix == 0);
4182         if (!match.opt.forward && match.opt.ignoreformat) {
4183                 if (expected.pos > 0)
4184                         return fail;
4185         }
4186         LASSERT(at_begin, /**/);
4187         if (expected.match_len > 0 && at_begin) {
4188                 // Search for deepest match
4189                 old_cur = cur;
4190                 max_match = expected;
4191                 do {
4192                         size_t d = cur.depth();
4193                         cur.forwardPos();
4194                         if (!cur)
4195                                 break;
4196                         if (cur.depth() < d)
4197                                 break;
4198                         if (cur.depth() == d)
4199                                 break;
4200                         size_t lastd = d;
4201                         while (cur && cur.depth() > lastd) {
4202                                 lastd = cur.depth();
4203                                 mres = match(cur, -1, at_begin);
4204                                 displayMres(mres, "Checking innermost", cur);
4205                                 if (mres.match_len > 0)
4206                                         break;
4207                                 // maybe deeper?
4208                                 cur.forwardPos();
4209                         }
4210                         if (mres.match_len < expected.match_len)
4211                                 break;
4212                         max_match = mres;
4213                         old_cur = cur;;
4214                 } while(1);
4215                 cur = old_cur;
4216         }
4217         else {
4218                 // (expected.match_len <= 0)
4219                 mres = match(cur);      /* match valid only if not searching whole words */
4220                 displayMres(mres, "Start with negative match", cur);
4221                 max_match = mres;
4222         }
4223         if (max_match.match_len <= 0) return fail;
4224         LYXERR(Debug::FINDVERBOSE, "Ok");
4225
4226         // Compute the match length
4227         int len = 1;
4228         if (cur.pos() + len > cur.lastpos())
4229                 return fail;
4230
4231         LASSERT(match.use_regexp, /**/);
4232         {
4233                 int minl = 1;
4234                 int maxl = cur.lastpos() - cur.pos();
4235                 // Greedy behaviour while matching regexps
4236                 while (maxl > minl) {
4237                         MatchResult mres2;
4238                         mres2 = match(cur, len, at_begin);
4239                         displayMres(mres2, "Finalize loop", cur);
4240                         int actual_match_len = mres2.match_len;
4241                         if (actual_match_len >= max_match.match_len) {
4242                                 // actual_match_len > max_match _can_ happen,
4243                                 // if the search area splits
4244                                 // some following word so that the regex
4245                                 // (e.g. 'r.*r\b' matches 'r' from the middle of the
4246                                 // splitted word)
4247                                 // This means, the len value is too big
4248                                 actual_match_len = max_match.match_len;
4249                                 max_match = mres2;
4250                                 max_match.match_len = actual_match_len;
4251                                 maxl = len;
4252                                 if (maxl - minl < 4)
4253                                         len = (maxl + minl)/2;
4254                                 else
4255                                         len = minl + (maxl - minl + 3)/4;
4256                         }
4257                         else {
4258                                 // (actual_match_len < max_match.match_len)
4259                                 minl = len + 1;
4260                                 len = (maxl + minl)/2;
4261                         }
4262                 }
4263                 len = minl;
4264                 old_cur = cur;
4265                 // Search for real start of matched characters
4266                 while (len > 1) {
4267                         MatchResult actual_match;
4268                         do {
4269                                 cur.forwardPos();
4270                         } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
4271                         if (cur.depth() < old_cur.depth()) {
4272                                 // Outer inset?
4273                                 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
4274                                 break;
4275                         }
4276                         if (cur.pos() != old_cur.pos()) {
4277                                 // OK, forwarded 1 pos in actual inset
4278                                 actual_match = match(cur, len-1, at_begin);
4279                                 if (actual_match.match_len == max_match.match_len) {
4280                                         // Ha, got it! The shorter selection has the same match length
4281                                         len--;
4282                                         old_cur = cur;
4283                                         max_match = actual_match;
4284                                 }
4285                                 else {
4286                                         // OK, the shorter selection matches less chars, revert to previous value
4287                                         cur = old_cur;
4288                                         break;
4289                                 }
4290                         }
4291                         else {
4292                                 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
4293                                 actual_match = match(cur, len, at_begin);
4294                                 if (actual_match.match_len == max_match.match_len) {
4295                                         old_cur = cur;
4296                                         max_match = actual_match;
4297                                 }
4298                         }
4299                 }
4300                 if (len == 0)
4301                         return fail;
4302                 else {
4303                         max_match.pos_len = len;
4304                         displayMres(max_match, "SEARCH RESULT", cur)
4305                                         return max_match;
4306                 }
4307         }
4308 }
4309
4310 /// Finds forward
4311 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
4312 {
4313         if (!cur)
4314                 return 0;
4315         bool repeat = false;
4316         DocIterator orig_cur;   // to be used if repeat not successful
4317         MatchResult orig_mres;
4318         while (!theApp()->longOperationCancelled() && cur) {
4319                 //(void) findAdvForwardInnermost(cur);
4320                 LYXERR(Debug::FINDVERBOSE, "findForwardAdv() cur: " << cur);
4321                 MatchResult mres = match(cur, -1, false);
4322                 string msg = "Starting";
4323                 if (repeat)
4324                         msg = "Repeated";
4325                 displayMres(mres, msg + " findForwardAdv", cur)
4326                                 int match_len = mres.match_len;
4327                 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
4328                         LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
4329                         match_len = 0;
4330                 }
4331                 if (match_len <= 0) {
4332                         // This should exit nested insets, if any, or otherwise undefine the currsor.
4333                         cur.pos() = cur.lastpos();
4334                         LYXERR(Debug::FINDVERBOSE, "Advancing pos: cur=" << cur);
4335                         cur.forwardPos();
4336                 }
4337                 else {  // match_len > 0
4338                         // Try to find the begin of searched string
4339                         int increment;
4340                         int firstInvalid = cur.lastpos() - cur.pos();
4341                         {
4342                                 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
4343                                 int incrcur = (firstInvalid + 1 )*3/4;
4344                                 if (incrcur < incrmatch)
4345                                         increment = incrcur;
4346                                 else
4347                                         increment = incrmatch;
4348                                 if (increment < 1)
4349                                         increment = 1;
4350                         }
4351                         LYXERR(Debug::FINDVERBOSE, "Set increment to " << increment);
4352                         while (increment > 0) {
4353                                 DocIterator old_cur = cur;
4354                                 if (cur.pos() + increment >= cur.lastpos()) {
4355                                         increment /= 2;
4356                                         continue;
4357                                 }
4358                                 cur.pos() = cur.pos() + increment;
4359                                 MatchResult mres2 = match(cur, -1, false);
4360                                 displayMres(mres2, "findForwardAdv loop", cur)
4361                                                 switch (interpretMatch(mres, mres2)) {
4362                                         case MatchResult::newIsTooFar:
4363                                                 // behind the expected match
4364                                                 firstInvalid = increment;
4365                                                 cur = old_cur;
4366                                                 increment /= 2;
4367                                                 break;
4368                                         case MatchResult::newIsBetter:
4369                                                 // not reached yet, but cur.pos()+increment is bettert
4370                                                 mres = mres2;
4371                                                 firstInvalid -= increment;
4372                                                 if (increment > firstInvalid*3/4)
4373                                                         increment = firstInvalid*3/4;
4374                                                 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
4375                                                         if (increment >= mres2.match_prefix)
4376                                                                 increment = (mres2.match_prefix+1)*3/4;
4377                                                 }
4378                                                 break;
4379                                         default:
4380                                                 // Todo@
4381                                                 // Handle not like MatchResult::newIsTooFar
4382                                                 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
4383                                                 firstInvalid--;
4384                                                 increment = increment*3/4;
4385                                                 cur = old_cur;
4386                                                 break;
4387                                 }
4388                         }
4389                         if (mres.match_len > 0) {
4390                                 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
4391                                         // The match seems to indicate some deeper level
4392                                         repeat = true;
4393                                         orig_cur = cur;
4394                                         orig_mres = mres;
4395                                         cur.forwardPos();
4396                                         continue;
4397                                 }
4398                         }
4399                         else if (repeat) {
4400                                 // should never be reached.
4401                                 cur = orig_cur;
4402                                 mres = orig_mres;
4403                         }
4404                         // LYXERR0("Leaving first loop");
4405                         LYXERR(Debug::FINDVERBOSE, "Finalizing 1");
4406                         MatchResult found_match = findAdvFinalize(cur, match, mres);
4407                         if (found_match.match_len > 0) {
4408                                 match.FillResults(found_match);
4409                                 return found_match.pos_len;
4410                         }
4411                         else {
4412                                 // try next possible match
4413                                 cur.forwardPos();
4414                                 repeat = false;
4415                                 continue;
4416                         }
4417                 }
4418         }
4419         return 0;
4420 }
4421
4422
4423 /// Find the most backward consecutive match within same paragraph while searching backwards.
4424 MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, MatchResult &expected)
4425 {
4426         DocIterator cur_begin = cur;
4427         cur_begin.pos() = 0;
4428         DocIterator tmp_cur = cur;
4429         MatchResult mr = findAdvFinalize(tmp_cur, match, expected);
4430         Inset & inset = cur.inset();
4431         for (; cur != cur_begin; cur.backwardPos()) {
4432                 LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): cur=" << cur);
4433                 DocIterator new_cur = cur;
4434                 new_cur.backwardPos();
4435                 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
4436                         break;
4437                 MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
4438                 if (new_mr.match_len == mr.match_len)
4439                         break;
4440                 mr = new_mr;
4441         }
4442         LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): exiting with cur=" << cur);
4443         return mr;
4444 }
4445
4446
4447 /// Finds backwards
4448 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
4449 {
4450         if (! cur)
4451                 return 0;
4452         // Backup of original position
4453         DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4454         if (cur == cur_begin)
4455                 return 0;
4456         cur.backwardPos();
4457         DocIterator cur_orig(cur);
4458         bool pit_changed = false;
4459         do {
4460                 cur.pos() = 0;
4461                 MatchResult found_match = match(cur, -1, false);
4462
4463                 if (found_match.match_len > 0) {
4464                         if (pit_changed)
4465                                 cur.pos() = cur.lastpos();
4466                         else
4467                                 cur.pos() = cur_orig.pos();
4468                         LYXERR(Debug::FINDVERBOSE, "findBackAdv2: cur: " << cur);
4469                         DocIterator cur_prev_iter;
4470                         do {
4471                                 found_match = match(cur);
4472                                 LYXERR(Debug::FINDVERBOSE, "findBackAdv3: found_match="
4473                                        << (found_match.match_len > 0) << ", cur: " << cur);
4474                                 if (found_match.match_len > 0) {
4475                                         MatchResult found_mr = findMostBackwards(cur, match, found_match);
4476                                         if (found_mr.pos_len > 0) {
4477                                                 match.FillResults(found_mr);
4478                                                 return found_mr.pos_len;
4479                                         }
4480                                 }
4481
4482                                 // Stop if begin of document reached
4483                                 if (cur == cur_begin)
4484                                         break;
4485                                 cur_prev_iter = cur;
4486                                 cur.backwardPos();
4487                         } while (true);
4488                 }
4489                 if (cur == cur_begin)
4490                         break;
4491                 if (cur.pit() > 0)
4492                         --cur.pit();
4493                 else
4494                         cur.backwardPos();
4495                 pit_changed = true;
4496         } while (!theApp()->longOperationCancelled());
4497         return 0;
4498 }
4499
4500
4501 } // namespace
4502
4503
4504 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4505                                  DocIterator const & cur, int len)
4506 {
4507         if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4508                 return docstring();
4509         if (!opt.ignoreformat)
4510                 return latexifyFromCursor(cur, len);
4511         else
4512                 return stringifyFromCursor(cur, len);
4513 }
4514
4515
4516 FindAndReplaceOptions::FindAndReplaceOptions(
4517                 docstring const & _find_buf_name, bool _casesensitive,
4518                 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4519                 docstring const & _repl_buf_name, bool _keep_case,
4520                 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4521         : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4522           forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4523           repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4524 {
4525 }
4526
4527
4528 namespace {
4529
4530
4531 /** Check if 'len' letters following cursor are all non-lowercase */
4532 static bool allNonLowercase(Cursor const & cur, int len)
4533 {
4534         pos_type beg_pos = cur.selectionBegin().pos();
4535         pos_type end_pos = cur.selectionBegin().pos() + len;
4536         if (len > cur.lastpos() + 1 - beg_pos) {
4537                 LYXERR(Debug::FINDVERBOSE, "This should not happen, more debug needed");
4538                 len = cur.lastpos() + 1 - beg_pos;
4539                 end_pos = beg_pos + len;
4540         }
4541         for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4542                 if (isLowerCase(cur.paragraph().getChar(pos)))
4543                         return false;
4544         return true;
4545 }
4546
4547
4548 /** Check if first letter is upper case and second one is lower case */
4549 static bool firstUppercase(Cursor const & cur)
4550 {
4551         char_type ch1, ch2;
4552         pos_type pos = cur.selectionBegin().pos();
4553         if (pos >= cur.lastpos() - 1) {
4554                 LYXERR(Debug::FINDVERBOSE, "No upper-case at cur: " << cur);
4555                 return false;
4556         }
4557         ch1 = cur.paragraph().getChar(pos);
4558         ch2 = cur.paragraph().getChar(pos + 1);
4559         bool result = isUpperCase(ch1) && isLowerCase(ch2);
4560         LYXERR(Debug::FINDVERBOSE, "firstUppercase(): "
4561                << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4562                << ch2 << "(" << char(ch2) << ")"
4563                << ", result=" << result << ", cur=" << cur);
4564         return result;
4565 }
4566
4567
4568 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4569  **
4570  ** \fixme What to do with possible further paragraphs in replace buffer ?
4571  **/
4572 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4573 {
4574         ParagraphList::iterator pit = buffer.paragraphs().begin();
4575         LASSERT(!pit->empty(), /**/);
4576         pos_type right = pos_type(1);
4577         pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4578         right = pit->size();
4579         pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4580 }
4581 } // namespace
4582
4583 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4584 {
4585         // Should replace the string "$" + std::to_string(matchnum) with replacement
4586         // if the char '$' is not prefixed with odd number of char '\\'
4587         static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4588         string s;
4589         size_t lastpos = 0;
4590         smatch sub;
4591         for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4592                 sub = *it;
4593                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4594                         continue;
4595                 int num = stoi(sub.str(3), nullptr, 10);
4596                 if (num >= maxmatchnum)
4597                         continue;
4598                 if (lastpos < (size_t) sub.position(2))
4599                         s += t.substr(lastpos, sub.position(2) - lastpos);
4600                 s += replacements[num];
4601                 lastpos = sub.position(2) + sub.length(2);
4602         }
4603         if (lastpos == 0)
4604                 return false;
4605         else if (lastpos < t.length())
4606                 s += t.substr(lastpos, t.length() - lastpos);
4607         t = s;
4608         return true;
4609 }
4610
4611 ///
4612 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4613 {
4614         Cursor & cur = bv->cursor();
4615         if (opt.repl_buf_name.empty()
4616                         || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4617                         || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4618                 return 0;
4619
4620         DocIterator sel_beg = cur.selectionBegin();
4621         DocIterator sel_end = cur.selectionEnd();
4622         if (&sel_beg.inset() != &sel_end.inset()
4623                         || sel_beg.pit() != sel_end.pit()
4624                         || sel_beg.idx() != sel_end.idx())
4625                 return 0;
4626         int sel_len = sel_end.pos() - sel_beg.pos();
4627         LYXERR(Debug::FINDVERBOSE, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4628                << ", sel_len: " << sel_len << endl);
4629         if (sel_len == 0)
4630                 return 0;
4631         LASSERT(sel_len > 0, return 0);
4632
4633         if (!matchAdv(sel_beg, sel_len).match_len)
4634                 return 0;
4635
4636         // Build a copy of the replace buffer, adapted to the KeepCase option
4637         Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4638         ostringstream oss;
4639         repl_buffer_orig.write(oss);
4640         string lyx = oss.str();
4641         if (matchAdv.valid_matches > 0)
4642                 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4643         Buffer repl_buffer(string(), false);
4644         repl_buffer.setInternal(true);
4645         repl_buffer.setUnnamed(true);
4646         LASSERT(repl_buffer.readString(lyx), return 0);
4647         if (opt.keep_case && sel_len >= 2) {
4648                 LYXERR(Debug::FINDVERBOSE, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4649                 if (cur.inTexted()) {
4650                         if (firstUppercase(cur))
4651                                 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4652                         else if (allNonLowercase(cur, sel_len))
4653                                 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4654                 }
4655         }
4656         cap::cutSelection(cur, false);
4657         if (cur.inTexted()) {
4658                 repl_buffer.changeLanguage(
4659                                         repl_buffer.language(),
4660                                         cur.getFont().language());
4661                 LYXERR(Debug::FINDVERBOSE, "Replacing by pasteParagraphList()ing repl_buffer");
4662                 LYXERR(Debug::FINDVERBOSE, "Before pasteParagraphList() cur=" << cur << endl);
4663                 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4664                                         repl_buffer.params().documentClassPtr(),
4665                                         repl_buffer.params().authors(),
4666                                         bv->buffer().errorList("Paste"));
4667                 LYXERR(Debug::FINDVERBOSE, "After pasteParagraphList() cur=" << cur << endl);
4668                 sel_len = repl_buffer.paragraphs().begin()->size();
4669         } else if (cur.inMathed()) {
4670                 odocstringstream ods;
4671                 otexstream os(ods);
4672                 // OutputParams runparams(&repl_buffer.params().encoding());
4673                 OutputParams runparams(encodings.fromLyXName("utf8"));
4674                 runparams.nice = false;
4675                 setFindParams(runparams);
4676                 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams, string(), -1, -1, true);
4677                 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4678                 docstring repl_latex = ods.str();
4679                 LYXERR(Debug::FINDVERBOSE, "Latexified replace_buffer: '" << repl_latex << "'");
4680                 string s;
4681                 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4682                 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4683                 repl_latex = from_utf8(s);
4684                 LYXERR(Debug::FINDVERBOSE, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4685                 MathData ar(cur.buffer());
4686                 asArray(repl_latex, ar, Parse::NORMAL);
4687                 cur.insert(ar);
4688                 sel_len = ar.size();
4689                 LYXERR(Debug::FINDVERBOSE, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4690         }
4691         if (cur.pos() >= sel_len)
4692                 cur.pos() -= sel_len;
4693         else
4694                 cur.pos() = 0;
4695         LYXERR(Debug::FINDVERBOSE, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4696         bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4697         bv->processUpdateFlags(Update::Force);
4698         return 1;
4699 }
4700
4701
4702 /// Perform a FindAdv operation.
4703 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4704 {
4705         DocIterator cur;
4706         int pos_len = 0;
4707
4708         // e.g., when invoking word-findadv from mini-buffer wither with
4709         //       wrong options syntax or before ever opening advanced F&R pane
4710         if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4711                 return false;
4712
4713         try {
4714                 MatchStringAdv matchAdv(bv->buffer(), opt);
4715 #if QTSEARCH
4716                 if (!matchAdv.regexIsValid) {
4717                         bv->message(lyx::from_utf8(matchAdv.regexError));
4718                         return(false);
4719                 }
4720 #endif
4721                 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4722                 if (length > 0)
4723                         bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4724                 num_replaced += findAdvReplace(bv, opt, matchAdv);
4725                 cur = bv->cursor();
4726                 if (opt.forward)
4727                         pos_len = findForwardAdv(cur, matchAdv);
4728                 else
4729                         pos_len = findBackwardsAdv(cur, matchAdv);
4730         } catch (exception & ex) {
4731                 bv->message(from_utf8(ex.what()));
4732                 return false;
4733         }
4734
4735         if (pos_len == 0) {
4736                 if (num_replaced > 0) {
4737                         switch (num_replaced)
4738                         {
4739                         case 1:
4740                                 bv->message(_("One match has been replaced."));
4741                                 break;
4742                         case 2:
4743                                 bv->message(_("Two matches have been replaced."));
4744                                 break;
4745                         default:
4746                                 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4747                                 break;
4748                         }
4749                         num_replaced = 0;
4750                 }
4751                 else {
4752                         bv->message(_("Match not found."));
4753                 }
4754                 return false;
4755         }
4756
4757         if (num_replaced > 0)
4758                 bv->message(_("Match has been replaced."));
4759         else
4760                 bv->message(_("Match found."));
4761
4762         if (cur.pos() + pos_len > cur.lastpos()) {
4763                 // Prevent crash in bv->putSelectionAt()
4764                 // Should never happen, maybe LASSERT() here?
4765                 pos_len = cur.lastpos() - cur.pos();
4766         }
4767         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4768         bv->putSelectionAt(cur, pos_len, !opt.forward);
4769
4770         return true;
4771 }
4772
4773
4774 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4775 {
4776         os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4777            << opt.casesensitive << ' '
4778            << opt.matchword << ' '
4779            << opt.forward << ' '
4780            << opt.expandmacros << ' '
4781            << opt.ignoreformat << ' '
4782            << opt.replace_all << ' '
4783            << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4784            << opt.keep_case << ' '
4785            << int(opt.scope) << ' '
4786            << int(opt.restr);
4787
4788         LYXERR(Debug::FINDVERBOSE, "built: " << os.str());
4789
4790         return os;
4791 }
4792
4793
4794 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4795 {
4796         // LYXERR(Debug::FINDVERBOSE, "parsing");
4797         string s;
4798         string line;
4799         getline(is, line);
4800         while (line != "EOSS") {
4801                 if (! s.empty())
4802                         s = s + "\n";
4803                 s = s + line;
4804                 if (is.eof())   // Tolerate malformed request
4805                         break;
4806                 getline(is, line);
4807         }
4808         // LYXERR(Debug::FINDVERBOSE, "file_buf_name: '" << s << "'");
4809         opt.find_buf_name = from_utf8(s);
4810         is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4811         is.get();       // Waste space before replace string
4812         s = "";
4813         getline(is, line);
4814         while (line != "EOSS") {
4815                 if (! s.empty())
4816                         s = s + "\n";
4817                 s = s + line;
4818                 if (is.eof())   // Tolerate malformed request
4819                         break;
4820                 getline(is, line);
4821         }
4822         // LYXERR(Debug::FINDVERBOSE, "repl_buf_name: '" << s << "'");
4823         opt.repl_buf_name = from_utf8(s);
4824         is >> opt.keep_case;
4825         int i;
4826         is >> i;
4827         opt.scope = FindAndReplaceOptions::SearchScope(i);
4828         is >> i;
4829         opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4830
4831         /*
4832         LYXERR(Debug::FINDVERBOSE, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4833                << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4834                << opt.scope << ' ' << opt.restr);
4835         */
4836         return is;
4837 }
4838
4839 } // namespace lyx