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 static string getRegexSpaceCount(int count)
 816 {
 817         if (count > 0) {
 818                 if (count > 1)
 819                         return "\\s{" + std::to_string(count) + "}";
 820                 else
 821                         return "\\s";
 822         }
 823         return "";
 824 }
 825
 826 string string2regex(string in)
 827 {
 828         static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\$\\])" };
 829         string tempx = std::regex_replace(in, specialChars,  R"(\$&)" );
 830         // Special handling for ' '
 831         string temp("");
 832         int blanks = 0;
 833         for (unsigned i = 0; i < tempx.size(); i++) {
 834                 if (tempx[i] == ' ' || tempx[i] == '~' ) {
 835                         // normal blanks
 836                         blanks++;
 837                 }
 838                 else if ((tempx[i] == '\302' && tempx[i+1] == '\240')
 839                         || (tempx[i] == '\342' && tempx[i+1] == '\200')) {
 840                         // protected space
 841                         // thin space
 842                         blanks++;
 843                         i++;
 844                 }
 845                 else {
 846                         if (blanks > 0) {
 847                                 temp += getRegexSpaceCount(blanks);
 848                         }
 849                         temp += tempx[i];
 850                         blanks = 0;
 851                 }
 852         }
 853         if (blanks > 0) {
 854                 temp += getRegexSpaceCount(blanks);
 855         }
 856
 857         string temp2("");
 858         size_t lastpos = 0;
 859         size_t fl_pos = 0;
 860         int offset = 1;
 861         while (fl_pos < temp.size()) {
 862                 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
 863                 if (fl_pos == string::npos)
 864                         break;
 865                 offset = 16;
 866                 temp2 += temp.substr(lastpos, fl_pos - lastpos);
 867                 temp2 += "\\n";
 868                 lastpos = fl_pos;
 869         }
 870         if (lastpos == 0)
 871                 return(temp);
 872         if (lastpos < temp.size()) {
 873                 temp2 += temp.substr(lastpos, temp.size() - lastpos);
 874         }
 875         return temp2;
 876 }
 877
 878 static void buildAccentsMap();
 879
 880 string correctRegex(string t, bool withformat)
 881 {
 882         /* Convert \backslash => \
 883          * and \{, \}, \[, \] => {, }, [, ]
 884          */
 885         string s("");
 886         static std::regex wordre("(\\\\)*(\\\\(( |[A-Za-z]+|[\\{\\}%])( |\\{\\})?|[\\[\\]\\{\\}]))");
 887         static std::regex protectedSpace { R"(~)" };
 888         size_t lastpos = 0;
 889         smatch sub;
 890         bool backslashed = false;
 891         if (accents.empty())
 892                 buildAccentsMap();
 893
 894         //LYXERR0("correctRegex input '" << t << "'");
 895         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
 896                 sub = *it;
 897                 string replace;
 898                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
 899                         continue;
 900                 }
 901                 else {
 902                         if (sub.str(4) == "backslash") {
 903                                 replace = "\\";
 904                                 {
 905                                         // transforms '\backslash \{' into '\{'
 906                                         string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
 907                                         if ((next == "\\{") || (next == "\\}") || (next == "\\ ")) {
 908                                                 replace = "";
 909                                                 backslashed = true;
 910                                         }
 911                                 }
 912                         }
 913                         else if (sub.str(4) == "mathcircumflex")
 914                                 replace = "^";
 915                         else if (backslashed) {
 916                                 backslashed = false;
 917                                 if (withformat) {
 918                                         if (sub.str(3) == "{")
 919                                                 replace = accents["braceleft"];
 920                                         else if (sub.str(3) == "}")
 921                                                 replace = accents["braceright"];
 922                                         else if (sub.str(3) == " ")
 923                                                 replace = "\\ ";
 924                                         else {
 925                                                 // else part should not exist
 926                                                 LASSERT(0, /**/);
 927                                         }
 928                                 }
 929                                 else {
 930                                         if (sub.str(3) == "{")
 931                                                 replace = "\\{";
 932                                         else if (sub.str(3) == "}")
 933                                                 replace = "\\}";
 934                                         else if (sub.str(3) == " ")
 935                                                 replace = "\\ ";
 936                                         else {
 937                                                 // else part should not exist
 938                                                 LASSERT(0, /**/);
 939                                         }
 940                                 }
 941                         }
 942                         else if (sub.str(4) == "{") // transforms '\{' into '{'
 943                                 replace = "{";
 944                         else if (sub.str(4) == "}")
 945                                 replace = "}";
 946                         else if (sub.str(4) == "%")
 947                                 replace = "%";
 948                         else if (sub.str(4) == " ")
 949                                 replace = " ";
 950                         else {
 951                                 AccentsIterator it_ac = accents.find(sub.str(4));
 952                                 if (it_ac == accents.end()) {
 953                                         replace = sub.str(2);
 954                                 }
 955                                 else {
 956                                         replace = it_ac->second;
 957                                 }
 958                         }
 959                 }
 960                 if (lastpos < (size_t) sub.position(2))
 961                         s += std::regex_replace(t.substr(lastpos, sub.position(2) - lastpos), protectedSpace, R"( )");
 962                 s += replace;
 963                 lastpos = sub.position(2) + sub.length(2);
 964         }
 965         if (lastpos == 0)
 966                 s = std::regex_replace(t, protectedSpace, R"( )");
 967         else if (lastpos < t.length())
 968                 s += std::regex_replace(t.substr(lastpos, t.length() - lastpos), protectedSpace, R"( )");
 969         // Handle quotes in regex
 970         // substitute all '„', '“', '»', '«' with '"'
 971         // and all '‚', '‘', '›', '‹' with "\'"
 972         static std::regex plainquotes { R"(„|“|»|«)" };
 973         static std::regex innerquotes { R"(‚|‘|›|‹)" };
 974         t = std::regex_replace(s, plainquotes, R"(")");
 975         s = std::regex_replace(t, innerquotes, R"(')");
 976         //LYXERR0("correctRegex output '" << s << "'");
 977         return s;
 978 }
 979
 980 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
 981 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
 982 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
 983 string escape_for_regex(string s, bool withformat)
 984 {
 985         size_t lastpos = 0;
 986         string result = "";
 987         while (lastpos < s.size()) {
 988                 size_t regex_pos = s.find("\\regexp{", lastpos);
 989                 if (regex_pos == string::npos) {
 990                         regex_pos = s.size();
 991                 }
 992                 if (regex_pos > lastpos) {
 993                         result += string2regex(s.substr(lastpos, regex_pos-lastpos));
 994                         lastpos = regex_pos;
 995                         if (lastpos == s.size())
 996                                 break;
 997                 }
 998                 size_t end_pos = s.find("\\endregexp", regex_pos + 8);
 999                 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
1000                 lastpos = end_pos + 13;
1001         }
1002         return result;
1003 }
1004
1005
1006 /// Wrapper for lyx::regex_replace with simpler interface
1007 bool regex_replace(string const & s, string & t, string const & searchstr,
1008                    string const & replacestr)
1009 {
1010         regex e(searchstr, regex_constants::ECMAScript);
1011         ostringstream oss;
1012         ostream_iterator<char, char> it(oss);
1013         regex_replace(it, s.begin(), s.end(), e, replacestr);
1014         // tolerate t and s be references to the same variable
1015         bool rv = (s != oss.str());
1016         t = oss.str();
1017         return rv;
1018 }
1019
1020 class MatchResult {
1021 public:
1022         enum range {
1023                 newIsTooFar,
1024                 newIsBetter,
1025                 newIsInvalid
1026         };
1027         int match_len;
1028         int match_prefix;
1029         int match2end;
1030         int pos;
1031         int leadsize;
1032         int pos_len;
1033         int searched_size;
1034         vector <string> result = vector <string>();
1035         MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {}
1036 };
1037
1038 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
1039 {
1040         if (newres.match2end < oldres.match2end)
1041                 return MatchResult::newIsTooFar;
1042         if (newres.match_len < oldres.match_len)
1043                 return MatchResult::newIsTooFar;
1044
1045         if (newres.match_len == oldres.match_len) {
1046                 if (newres.match2end == oldres.match2end)
1047                         return MatchResult::newIsBetter;
1048         }
1049         return MatchResult::newIsInvalid;
1050 }
1051
1052 /** The class performing a match between a position in the document and the FindAdvOptions.
1053  **/
1054
1055 class MatchStringAdv {
1056 public:
1057         MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
1058
1059         /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
1060          ** constructor as opt.search, under the opt.* options settings.
1061          **
1062          ** @param at_begin
1063          **     If set, then match is searched only against beginning of text starting at cur.
1064          **     If unset, then match is searched anywhere in text starting at cur.
1065          **
1066          ** @return
1067          ** The length of the matching text, or zero if no match was found.
1068          **/
1069         MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1070 #if QTSEARCH
1071         bool regexIsValid;
1072         string regexError;
1073 #endif
1074
1075 public:
1076         /// buffer
1077         lyx::Buffer * p_buf;
1078         /// first buffer on which search was started
1079         lyx::Buffer * const p_first_buf;
1080         /// options
1081         FindAndReplaceOptions const & opt;
1082
1083 private:
1084         /// Auxiliary find method (does not account for opt.matchword)
1085         MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1086         void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
1087
1088         /** Normalize a stringified or latexified LyX paragraph.
1089          **
1090          ** Normalize means:
1091          ** <ul>
1092          **   <li>if search is not casesensitive, then lowercase the string;
1093          **   <li>remove any newline at begin or end of the string;
1094          **   <li>replace any newline in the middle of the string with a simple space;
1095          **   <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
1096          ** </ul>
1097          **
1098          ** @todo Normalization should also expand macros, if the corresponding
1099          ** search option was checked.
1100          **/
1101         string normalize(docstring const & s, bool ignore_fomat) const;
1102         // normalized string to search
1103         string par_as_string;
1104         // regular expression to use for searching
1105         // regexp2 is same as regexp, but prefixed with a ".*?"
1106 #if QTSEARCH
1107         QRegularExpression regexp;
1108         QRegularExpression regexp2;
1109 #else
1110         regex regexp;
1111         regex regexp2;
1112 #endif
1113         // leading format material as string
1114         string lead_as_string;
1115         // par_as_string after removal of lead_as_string
1116         string par_as_string_nolead;
1117         // unmatched open braces in the search string/regexp
1118         int open_braces;
1119         // number of (.*?) subexpressions added at end of search regexp for closing
1120         // environments, math mode, styles, etc...
1121         int close_wildcards;
1122 public:
1123         // Are we searching with regular expressions ?
1124         bool use_regexp = false;
1125         static int valid_matches;
1126         static vector <string> matches;
1127         void FillResults(MatchResult &found_mr);
1128 };
1129
1130 int MatchStringAdv::valid_matches = 0;
1131 vector <string> MatchStringAdv::matches = vector <string>(10);
1132
1133 void MatchStringAdv::FillResults(MatchResult &found_mr)
1134 {
1135         if (found_mr.match_len > 0) {
1136                 valid_matches = found_mr.result.size();
1137                 for (size_t i = 0; i < found_mr.result.size(); i++)
1138                         matches[i] = found_mr.result[i];
1139         } else
1140                 valid_matches = 0;
1141 }
1142
1143 static void setFindParams(OutputParams &runparams)
1144 {
1145         runparams.flavor = Flavor::XeTeX;
1146         //runparams.use_polyglossia = true;
1147         runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1148         // No side effect of file copying and image conversion
1149         runparams.dryrun = true;
1150 }
1151
1152 static docstring buffer_to_latex(Buffer & buffer)
1153 {
1154         //OutputParams runparams(&buffer.params().encoding());
1155         OutputParams runparams(encodings.fromLyXName("utf8"));
1156         odocstringstream ods;
1157         otexstream os(ods);
1158         runparams.nice = true;
1159         setFindParams(runparams);
1160         if (ignoreFormats.getDeleted())
1161                 runparams.find_set_feature(OutputParams::SearchWithoutDeleted);
1162         else
1163                 runparams.find_set_feature(OutputParams::SearchWithDeleted);
1164         if (ignoreFormats.getNonContent()) {
1165                 runparams.find_add_feature(OutputParams::SearchNonOutput);
1166         }
1167         pit_type const endpit = buffer.paragraphs().size();
1168         for (pit_type pit = 0; pit != endpit; ++pit) {
1169                 TeXOnePar(buffer, buffer.text(), pit, os, runparams, string(), -1, -1, true);
1170                 LYXERR(Debug::FINDVERBOSE, "searchString up to here: " << ods.str());
1171         }
1172         return ods.str();
1173 }
1174
1175 static string latexNamesToUtf8(docstring strIn, bool withformat)
1176 {
1177         string addtmp = to_utf8(strIn);
1178         static regex const rmAcc("(\\\\)*("
1179                                          "\\\\([A-Za-z]+\\{.\\})"       // e.g. "ddot{A}" == sub.str(3)
1180                                         "|\\\\([A-Za-z]+)( |\\{\\})?"   // e.g. "LyX", "LyX{}", "LyX " == sub.str(4)
1181                                         ")"
1182                                 );
1183         size_t lastpos = 0;
1184         smatch sub;
1185         string replace;
1186         string add("");
1187         if (accents.empty())
1188                 buildAccentsMap();
1189         for (sregex_iterator it_add(addtmp.begin(), addtmp.end(), rmAcc), end; it_add != end; ++it_add) {
1190                 sub = *it_add;
1191                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
1192                         continue;
1193                 }
1194                 else {
1195                         string key;
1196                         if (sub.length(3) > 0)
1197                                 key = sub.str(3);
1198                         else
1199                                 key = sub.str(4);
1200                         AccentsIterator it_ac = accents.find(key);
1201                         if (it_ac == accents.end()) {
1202                                 replace = sub.str(2);
1203                         }
1204                         else {
1205                                 replace = it_ac->second;
1206                         }
1207                 }
1208                 if (lastpos < (size_t) sub.position(2))
1209                         add += addtmp.substr(lastpos, sub.position(2) - lastpos);
1210                 add += replace;
1211                 lastpos = sub.position(2) + sub.length(2);
1212         }
1213         if (lastpos == 0)
1214                 add = addtmp;
1215         else if (addtmp.length() > lastpos)
1216                 add += addtmp.substr(lastpos, addtmp.length() - lastpos);
1217         if (!withformat) {
1218                 static std::regex repltilde { R"(~)" };
1219                 add = std::regex_replace(add, repltilde, accents["lyxtilde"]);
1220         }
1221         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1222                         << add << "'");
1223         return add;
1224 }
1225
1226 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1227 {
1228         docstring str;
1229         if (!opt.ignoreformat) {
1230                 str = buffer_to_latex(buffer);
1231         } else {
1232                 // OutputParams runparams(&buffer.params().encoding());
1233                 OutputParams runparams(encodings.fromLyXName("utf8"));
1234                 runparams.nice = true;
1235                 setFindParams(runparams);
1236                 int option = AS_STR_INSETS |AS_STR_PLAINTEXT;
1237                 if (ignoreFormats.getDeleted()) {
1238                         option |= AS_STR_SKIPDELETE;
1239                         runparams.find_set_feature(OutputParams::SearchWithoutDeleted);
1240                 }
1241                 else {
1242                         runparams.find_set_feature(OutputParams::SearchWithDeleted);
1243                 }
1244                 if (ignoreFormats.getNonContent()) {
1245                         runparams.find_add_feature(OutputParams::SearchNonOutput);
1246                 }
1247                 string t("");
1248                 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1249                         Paragraph const & par = buffer.paragraphs().at(pit);
1250                         string add = latexNamesToUtf8(par.asString(pos_type(0), par.size(),
1251                                                                 option,
1252                                                                 &runparams), !opt.ignoreformat);
1253                         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1254                                 << add << "'");
1255                         t += add;
1256                 }
1257                 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1258                 while (regex_replace(t, t, "\\\\(text|lyxmathsym|ensuremath)\\{([^\\}]*)\\}", "$2"));
1259                 str = from_utf8(t);
1260         }
1261         return str;
1262 }
1263
1264
1265 /// Return separation pos between the leading material and the rest
1266 static size_t identifyLeading(string const & s)
1267 {
1268         string t = s;
1269         // @TODO Support \item[text]
1270         // Kornel: Added textsl, textsf, textit, texttt and noun
1271         // + allow to search for colored text too
1272         while (regex_replace(t, t, "^\\\\(("
1273                              "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1274                              "lyxaddress|lyxrightaddress|"
1275                              "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1276                              "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1277                              "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1278                              "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1279                || regex_replace(t, t, "^\\$", "")
1280                || regex_replace(t, t, "^\\\\\\[", "")
1281                || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1282                || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1283                ;
1284         LYXERR(Debug::FINDVERBOSE, "  after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1285         return s.size() - t.size();
1286 }
1287
1288 /*
1289  * Given a latexified string, retrieve some handled features
1290  * The features of the regex will later be compared with the features
1291  * of the searched text. If the regex features are not a
1292  * subset of the analized, then, in not format ignoring search
1293  * we can early stop the search in the relevant inset.
1294  */
1295 typedef map<string, bool> Features;
1296
1297 static Features identifyFeatures(string const & s)
1298 {
1299         static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1300         static regex const valid("^("
1301                 "("
1302                         "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1303                                 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1304                                 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1305                         "(u|uu)line|(s|x)out|uwave|"
1306                         "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1307                 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1308         smatch sub;
1309         bool displ = true;
1310         Features info;
1311
1312         for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1313                 sub = *it;
1314                 if (displ) {
1315                         if (sub.str(1).compare("regexp") == 0) {
1316                                 displ = false;
1317                                 continue;
1318                         }
1319                         string token = sub.str(1);
1320                         smatch sub2;
1321                         if (regex_match(token, sub2, valid)) {
1322                                 info[token] = true;
1323                         }
1324                         else {
1325                                 // ignore
1326                         }
1327                 }
1328                 else {
1329                         if (sub.str(1).compare("endregexp") == 0) {
1330                                 displ = true;
1331                                 continue;
1332                         }
1333                 }
1334         }
1335         return info;
1336 }
1337
1338 /*
1339  * defines values features of a key "\\[a-z]+{"
1340  */
1341 class KeyInfo {
1342 public:
1343         enum KeyType {
1344                 /* Char type with content discarded
1345                  * like \hspace{1cm} */
1346                 noContent,
1347                 /* Char, like \backslash */
1348                 isChar,
1349                 /* replace starting backslash with '#' */
1350                 isText,
1351                 /* \part, \section*, ... */
1352                 isSectioning,
1353                 /* title, author etc */
1354                 isTitle,
1355                 /* \foreignlanguage{ngerman}, ... */
1356                 isMain,
1357                 /* inside \code{}
1358                  * to discard language in content */
1359                 noMain,
1360                 isRegex,
1361                 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1362                 isMath,
1363                 /* fonts, colors, markups, ... */
1364                 isStandard,
1365                 /* footnotesize, ... large, ...
1366                  * Ignore all of them */
1367                 isSize,
1368                 invalid,
1369                 /* inputencoding, ...
1370                  * Discard also content, because they do not help in search */
1371                 doRemove,
1372                 /* twocolumns, ...
1373                  * like remove, but also all arguments */
1374                 removeWithArg,
1375                 /* item, listitem */
1376                 isList,
1377                 /* tex, latex, ... like isChar */
1378                 isIgnored,
1379                 /* like \lettrine[lines=5]{}{} */
1380                 cleanToStart,
1381                 // like isStandard, but always remove head
1382                 headRemove,
1383                 /* End of arguments marker for lettrine,
1384                  * so that they can be ignored */
1385                 endArguments
1386         };
1387         KeyInfo() = default;
1388         KeyInfo(KeyType type, int parcount, bool disable)
1389                 : keytype(type),
1390                   parenthesiscount(parcount),
1391                   disabled(disable) {}
1392         KeyType keytype = invalid;
1393         string head;
1394         int _tokensize = -1;
1395         int _tokenstart = -1;
1396         int _dataStart = -1;
1397         int _dataEnd = -1;
1398         int parenthesiscount = 1;
1399         bool disabled = false;
1400         bool used = false; /* by pattern */
1401 };
1402
1403 class Border {
1404 public:
1405         Border(int l=0, int u=0) : low(l), upper(u) {}
1406         int low;
1407         int upper;
1408 };
1409
1410 #define MAXOPENED 30
1411 class Intervall {
1412         bool isPatternString_;
1413 public:
1414         explicit Intervall(bool isPattern, string const & p)
1415                 : isPatternString_(isPattern), par(p), ignoreidx(-1),
1416                   actualdeptindex(0), hasTitle(false), langcount(0)
1417         {
1418                 depts[0] = 0;
1419                 closes[0] = 0;
1420         }
1421
1422         string par;
1423         int ignoreidx;
1424         static vector<Border> borders;
1425         int depts[MAXOPENED];
1426         int closes[MAXOPENED];
1427         int actualdeptindex;
1428         int previousNotIgnored(int) const;
1429         int nextNotIgnored(int) const;
1430         void handleOpenP(int i);
1431         void handleCloseP(int i, bool closingAllowed);
1432         void resetOpenedP(int openPos);
1433         void addIntervall(int upper);
1434         void addIntervall(int low, int upper); /* if explicit */
1435         void removeAccents();
1436         void setForDefaultLang(KeyInfo const & defLang) const;
1437         int findclosing(int start, int end, char up, char down, int repeat);
1438         void handleParentheses(int lastpos, bool closingAllowed);
1439         bool hasTitle;
1440         // Number of disabled language specs up
1441         // to current position in actual interval
1442         int langcount;
1443         int isOpeningPar(int pos) const;
1444         string titleValue;
1445         void output(ostringstream &os, int lastpos);
1446         // string show(int lastpos);
1447 };
1448
1449 vector<Border> Intervall::borders = vector<Border>(30);
1450
1451 int Intervall::isOpeningPar(int pos) const
1452 {
1453         if ((pos < 0) || (size_t(pos) >= par.size()))
1454                 return 0;
1455         if (par[pos] != '{')
1456                 return 0;
1457         if (size_t(pos) + 2 >= par.size())
1458                 return 1;
1459         if (par[pos+2] != '}')
1460                 return 1;
1461         if (par[pos+1] == '[' || par[pos+1] == ']')
1462                 return 3;
1463         return 1;
1464 }
1465
1466 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1467 {
1468         // Enable the use of first token again
1469         if (ignoreidx >= 0) {
1470                 int value = defLang._tokenstart + defLang._tokensize;
1471                 int borderidx = 0;
1472                 if (hasTitle)
1473                         borderidx = 1;
1474                 if (value > 0) {
1475                         if (borders[borderidx].low < value)
1476                                 borders[borderidx].low = value;
1477                         if (borders[borderidx].upper < value)
1478                                 borders[borderidx].upper = value;
1479                 }
1480         }
1481 }
1482
1483 static void checkDepthIndex(int val)
1484 {
1485         static int maxdepthidx = MAXOPENED-2;
1486         static int lastmaxdepth = 0;
1487         if (val > lastmaxdepth) {
1488                 LYXERR(Debug::INFO, "Depth reached " << val);
1489                 lastmaxdepth = val;
1490         }
1491         if (val > maxdepthidx) {
1492                 maxdepthidx = val;
1493                 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1494         }
1495 }
1496
1497 #if 0
1498 // Not needed, because borders are now dynamically expanded
1499 static void checkIgnoreIdx(int val)
1500 {
1501         static int lastmaxignore = -1;
1502         if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1503                 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1504                 lastmaxignore = val;
1505         }
1506 }
1507 #endif
1508
1509 /*
1510  * Expand the region of ignored parts of the input latex string
1511  * The region is only relevant in output()
1512  */
1513 void Intervall::addIntervall(int low, int upper)
1514 {
1515         int idx;
1516         if (low == upper) return;
1517         for (idx = ignoreidx+1; idx > 0; --idx) {
1518                 if (low > borders[idx-1].upper) {
1519                         break;
1520                 }
1521         }
1522         Border br(low, upper);
1523         if (idx > ignoreidx) {
1524                 if (borders.size() <= size_t(idx)) {
1525                         borders.push_back(br);
1526                 }
1527                 else {
1528                         borders[idx] = br;
1529                 }
1530                 ignoreidx = idx;
1531                 // checkIgnoreIdx(ignoreidx);
1532                 return;
1533         }
1534         else {
1535                 // Expand only if one of the new bound is inside the interwall
1536                 // We know here that br.low > borders[idx-1].upper
1537                 if (br.upper < borders[idx].low) {
1538                         // We have to insert at this pos
1539                         if (size_t(ignoreidx+1) >= borders.size()) {
1540                                 borders.push_back(borders[ignoreidx]);
1541                         }
1542                         else {
1543                                 borders[ignoreidx+1] = borders[ignoreidx];
1544                         }
1545                         for (int i = ignoreidx; i > idx; --i) {
1546                                 borders[i] = borders[i-1];
1547                         }
1548                         borders[idx] = br;
1549                         ignoreidx += 1;
1550                         // checkIgnoreIdx(ignoreidx);
1551                         return;
1552                 }
1553                 // Here we know, that we are overlapping
1554                 if (br.low > borders[idx].low)
1555                         br.low = borders[idx].low;
1556                 // check what has to be concatenated
1557                 int count = 0;
1558                 for (int i = idx; i <= ignoreidx; i++) {
1559                         if (br.upper >= borders[i].low) {
1560                                 count++;
1561                                 if (br.upper < borders[i].upper)
1562                                         br.upper = borders[i].upper;
1563                         }
1564                         else {
1565                                 break;
1566                         }
1567                 }
1568                 // count should be >= 1 here
1569                 borders[idx] = br;
1570                 if (count > 1) {
1571                         for (int i = idx + count; i <= ignoreidx; i++) {
1572                                 borders[i-count+1] = borders[i];
1573                         }
1574                         ignoreidx -= count - 1;
1575                         return;
1576                 }
1577         }
1578 }
1579
1580 static void buildaccent(string n, string param, string values)
1581 {
1582         stringstream s(n);
1583         string name;
1584         const char delim = '|';
1585         while (getline(s, name, delim)) {
1586                 size_t start = 0;
1587                 for (char c : param) {
1588                         string key = name + "{" + c + "}";
1589                         // get the corresponding utf8-value
1590                         if ((values[start] & 0xc0) != 0xc0) {
1591                                 // should not happen, utf8 encoding starts at least with 11xxxxxx
1592                                 // but value for '\dot{i}' is 'i', which is ascii
1593                                 if ((values[start] & 0x80) == 0) {
1594                                         // is ascii
1595                                         accents[key] = values.substr(start, 1);
1596                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1597                                 }
1598                                 start++;
1599                                 continue;
1600                         }
1601                         for (int j = 1; ;j++) {
1602                                 if (start + j >= values.size()) {
1603                                         accents[key] = values.substr(start, j);
1604                                         start = values.size() - 1;
1605                                         break;
1606                                 }
1607                                 else if ((values[start+j] & 0xc0) != 0x80) {
1608                                         // This is the first byte of following utf8 char
1609                                         accents[key] = values.substr(start, j);
1610                                         start += j;
1611                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1612                                         break;
1613                                 }
1614                         }
1615                 }
1616         }
1617 }
1618
1619 // Helper function
1620 static string getutf8(unsigned uchar)
1621 {
1622 #define maxc 5
1623         string ret = string();
1624         char c[maxc] = {0};
1625         if (uchar <= 0x7f) {
1626                 c[maxc-1] = uchar & 0x7f;
1627         }
1628         else {
1629                 unsigned char rest = 0x40;
1630                 unsigned char first = 0x80;
1631                 int start = maxc-1;
1632                 for (int i = start; i >=0; --i) {
1633                         if (uchar < rest) {
1634                                 c[i] = first + uchar;
1635                                 break;
1636                         }
1637                         c[i] = 0x80 | (uchar &  0x3f);
1638                         uchar >>= 6;
1639                         rest >>= 1;
1640                         first >>= 1;
1641                         first |= 0x80;
1642                 }
1643         }
1644         for (int i = 0; i < maxc; i++) {
1645                 if (c[i] == 0) continue;
1646                 ret += c[i];
1647         }
1648         return(ret);
1649 }
1650
1651 static void addAccents(string latex_in, string unicode_out)
1652 {
1653         latex_in = latex_in.substr(1);
1654         AccentsIterator it_ac = accents.find(latex_in);
1655         if (it_ac == accents.end()) {
1656                 accents[latex_in] = unicode_out;
1657         }
1658         else {
1659                 LYXERR0("Key " << latex_in  << " already set");
1660         }
1661 }
1662
1663 void static fillMissingUnicodesymbols()
1664 {
1665         addAccents("\\pounds", getutf8(0x00a3));
1666         addAccents("\\textsterling", getutf8(0x00a3));
1667         addAccents("\\textyen", getutf8(0x00a5));
1668         addAccents("\\yen", getutf8(0x00a5));
1669         addAccents("\\textsection", getutf8(0x00a7));
1670         addAccents("\\mathsection", getutf8(0x00a7));
1671         addAccents("\\textcopyright", getutf8(0x00a9));
1672         addAccents("\\copyright", getutf8(0x00a9));
1673         addAccents("\\textlnot", getutf8(0x00ac));
1674         addAccents("\\neg", getutf8(0x00ac));
1675         addAccents("\\textregistered", getutf8(0x00ae));
1676         addAccents("\\circledR", getutf8(0x00ae));
1677         addAccents("\\textpm", getutf8(0x00b1));
1678         addAccents("\\pm", getutf8(0x00b1));
1679         addAccents("\\textparagraph", getutf8(0x00b6));
1680         addAccents("\\mathparagraph", getutf8(0x00b6));
1681         addAccents("\\textperiodcentered", getutf8(0x00b7));
1682         addAccents("\\texttimes", getutf8(0x00d7));
1683         addAccents("\\times", getutf8(0x00d7));
1684         addAccents("\\O", getutf8(0x00d8));
1685         addAccents("\\dh", getutf8(0x00f0));
1686         addAccents("\\eth", getutf8(0x00f0));
1687         addAccents("\\textdiv", getutf8(0x00f7));
1688         addAccents("\\div", getutf8(0x00f7));
1689         addAccents("\\o", getutf8(0x00f8));
1690         addAccents("\\textcrlambda", getutf8(0x019b));
1691         addAccents("\\j", getutf8(0x0237));
1692         addAccents("\\textrevepsilon", getutf8(0x025c));
1693         addAccents("\\textbaru", getutf8(0x0289));
1694         addAccents("\\textquoteleft", getutf8(0x02bb));
1695         addAccents("\\textGamma", getutf8(0x0393));
1696         addAccents("\\Gamma", getutf8(0x0393));
1697         addAccents("\\textDelta", getutf8(0x0394));
1698         addAccents("\\Delta", getutf8(0x0394));
1699         addAccents("\\textTheta", getutf8(0x0398));
1700         addAccents("\\Theta", getutf8(0x0398));
1701         addAccents("\\textLambda", getutf8(0x039b));
1702         addAccents("\\Lambda", getutf8(0x039b));
1703         addAccents("\\textXi", getutf8(0x039e));
1704         addAccents("\\Xi", getutf8(0x039e));
1705         addAccents("\\textPi", getutf8(0x03a0));
1706         addAccents("\\Pi", getutf8(0x03a0));
1707         addAccents("\\textSigma", getutf8(0x03a3));
1708         addAccents("\\Sigma", getutf8(0x03a3));
1709         addAccents("\\textUpsilon", getutf8(0x03a5));
1710         addAccents("\\Upsilon", getutf8(0x03a5));
1711         addAccents("\\textPhi", getutf8(0x03a6));
1712         addAccents("\\Phi", getutf8(0x03a6));
1713         addAccents("\\textPsi", getutf8(0x03a8));
1714         addAccents("\\Psi", getutf8(0x03a8));
1715         addAccents("\\textOmega", getutf8(0x03a9));
1716         addAccents("\\Omega", getutf8(0x03a9));
1717         addAccents("\\textalpha", getutf8(0x03b1));
1718         addAccents("\\alpha", getutf8(0x03b1));
1719         addAccents("\\textbeta", getutf8(0x03b2));
1720         addAccents("\\beta", getutf8(0x03b2));
1721         addAccents("\\textgamma", getutf8(0x03b3));
1722         addAccents("\\gamma", getutf8(0x03b3));
1723         addAccents("\\textdelta", getutf8(0x03b4));
1724         addAccents("\\delta", getutf8(0x03b4));
1725         addAccents("\\textepsilon", getutf8(0x03b5));
1726         addAccents("\\varepsilon", getutf8(0x03b5));
1727         addAccents("\\textzeta", getutf8(0x03b6));
1728         addAccents("\\zeta", getutf8(0x03b6));
1729         addAccents("\\texteta", getutf8(0x03b7));
1730         addAccents("\\eta", getutf8(0x03b7));
1731         addAccents("\\texttheta", getutf8(0x03b8));
1732         addAccents("\\theta", getutf8(0x03b8));
1733         addAccents("\\textiota", getutf8(0x03b9));
1734         addAccents("\\iota", getutf8(0x03b9));
1735         addAccents("\\textkappa", getutf8(0x03ba));
1736         addAccents("\\kappa", getutf8(0x03ba));
1737         addAccents("\\textlambda", getutf8(0x03bb));
1738         addAccents("\\lambda", getutf8(0x03bb));
1739         addAccents("\\textmu", getutf8(0x03bc));
1740         addAccents("\\mu", getutf8(0x03bc));
1741         addAccents("\\textnu", getutf8(0x03bd));
1742         addAccents("\\nu", getutf8(0x03bd));
1743         addAccents("\\textxi", getutf8(0x03be));
1744         addAccents("\\xi", getutf8(0x03be));
1745         addAccents("\\textpi", getutf8(0x03c0));
1746         addAccents("\\pi", getutf8(0x03c0));
1747         addAccents("\\textrho", getutf8(0x03c1));
1748         addAccents("\\rho", getutf8(0x03c1));
1749         addAccents("\\textfinalsigma", getutf8(0x03c2));
1750         addAccents("\\varsigma", getutf8(0x03c2));
1751         addAccents("\\textsigma", getutf8(0x03c3));
1752         addAccents("\\sigma", getutf8(0x03c3));
1753         addAccents("\\texttau", getutf8(0x03c4));
1754         addAccents("\\tau", getutf8(0x03c4));
1755         addAccents("\\textupsilon", getutf8(0x03c5));
1756         addAccents("\\upsilon", getutf8(0x03c5));
1757         addAccents("\\textphi", getutf8(0x03c6));
1758         addAccents("\\varphi", getutf8(0x03c6));
1759         addAccents("\\textchi", getutf8(0x03c7));
1760         addAccents("\\chi", getutf8(0x03c7));
1761         addAccents("\\textpsi", getutf8(0x03c8));
1762         addAccents("\\psi", getutf8(0x03c8));
1763         addAccents("\\textomega", getutf8(0x03c9));
1764         addAccents("\\omega", getutf8(0x03c9));
1765         addAccents("\\textdigamma", getutf8(0x03dd));
1766         addAccents("\\digamma", getutf8(0x03dd));
1767         addAccents("\\hebalef", getutf8(0x05d0));
1768         addAccents("\\aleph", getutf8(0x05d0));
1769         addAccents("\\hebbet", getutf8(0x05d1));
1770         addAccents("\\beth", getutf8(0x05d1));
1771         addAccents("\\hebgimel", getutf8(0x05d2));
1772         addAccents("\\gimel", getutf8(0x05d2));
1773         addAccents("\\hebdalet", getutf8(0x05d3));
1774         addAccents("\\daleth", getutf8(0x05d3));
1775         addAccents("\\hebhe", getutf8(0x05d4));
1776         addAccents("\\hebvav", getutf8(0x05d5));
1777         addAccents("\\hebzayin", getutf8(0x05d6));
1778         addAccents("\\hebhet", getutf8(0x05d7));
1779         addAccents("\\hebtet", getutf8(0x05d8));
1780         addAccents("\\hebyod", getutf8(0x05d9));
1781         addAccents("\\hebfinalkaf", getutf8(0x05da));
1782         addAccents("\\hebkaf", getutf8(0x05db));
1783         addAccents("\\heblamed", getutf8(0x05dc));
1784         addAccents("\\hebfinalmem", getutf8(0x05dd));
1785         addAccents("\\hebmem", getutf8(0x05de));
1786         addAccents("\\hebfinalnun", getutf8(0x05df));
1787         addAccents("\\hebnun", getutf8(0x05e0));
1788         addAccents("\\hebsamekh", getutf8(0x05e1));
1789         addAccents("\\hebayin", getutf8(0x05e2));
1790         addAccents("\\hebfinalpe", getutf8(0x05e3));
1791         addAccents("\\hebpe", getutf8(0x05e4));
1792         addAccents("\\hebfinaltsadi", getutf8(0x05e5));
1793         addAccents("\\hebtsadi", getutf8(0x05e6));
1794         addAccents("\\hebqof", getutf8(0x05e7));
1795         addAccents("\\hebresh", getutf8(0x05e8));
1796         addAccents("\\hebshin", getutf8(0x05e9));
1797         addAccents("\\hebtav", getutf8(0x05ea));
1798
1799         // Thai characters
1800         addAccents("\\thaiKoKai", getutf8(0x0e01));
1801         addAccents("\\thaiKhoKhai", getutf8(0x0e02));
1802         addAccents("\\thaiKhoKhuat", getutf8(0x0e03));
1803         addAccents("\\thaiKhoKhwai", getutf8(0x0e04));
1804         addAccents("\\thaiKhoKhon", getutf8(0x0e05));
1805         addAccents("\\thaiKhoRakhang", getutf8(0x0e06));
1806         addAccents("\\thaiNgoNgu", getutf8(0x0e07));
1807         addAccents("\\thaiChoChan", getutf8(0x0e08));
1808         addAccents("\\thaiChoChing", getutf8(0x0e09));
1809         addAccents("\\thaiChoChang", getutf8(0x0e0a));
1810         addAccents("\\thaiSoSo", getutf8(0x0e0b));
1811         addAccents("\\thaiChoChoe", getutf8(0x0e0c));
1812         addAccents("\\thaiYoYing", getutf8(0x0e0d));
1813         addAccents("\\thaiDoChada", getutf8(0x0e0e));
1814         addAccents("\\thaiToPatak", getutf8(0x0e0f));
1815         addAccents("\\thaiThoThan", getutf8(0x0e10));
1816         addAccents("\\thaiThoNangmontho", getutf8(0x0e11));
1817         addAccents("\\thaiThoPhuthao", getutf8(0x0e12));
1818         addAccents("\\thaiNoNen", getutf8(0x0e13));
1819         addAccents("\\thaiDoDek", getutf8(0x0e14));
1820         addAccents("\\thaiToTao", getutf8(0x0e15));
1821         addAccents("\\thaiThoThung", getutf8(0x0e16));
1822         addAccents("\\thaiThoThahan", getutf8(0x0e17));
1823         addAccents("\\thaiThoThong", getutf8(0x0e18));
1824         addAccents("\\thaiNoNu", getutf8(0x0e19));
1825         addAccents("\\thaiBoBaimai", getutf8(0x0e1a));
1826         addAccents("\\thaiPoPla", getutf8(0x0e1b));
1827         addAccents("\\thaiPhoPhung", getutf8(0x0e1c));
1828         addAccents("\\thaiFoFa", getutf8(0x0e1d));
1829         addAccents("\\thaiPhoPhan", getutf8(0x0e1e));
1830         addAccents("\\thaiFoFan", getutf8(0x0e1f));
1831         addAccents("\\thaiPhoSamphao", getutf8(0x0e20));
1832         addAccents("\\thaiMoMa", getutf8(0x0e21));
1833         addAccents("\\thaiYoYak", getutf8(0x0e22));
1834         addAccents("\\thaiRoRua", getutf8(0x0e23));
1835         addAccents("\\thaiRu", getutf8(0x0e24));
1836         addAccents("\\thaiLoLing", getutf8(0x0e25));
1837         addAccents("\\thaiLu", getutf8(0x0e26));
1838         addAccents("\\thaiWoWaen", getutf8(0x0e27));
1839         addAccents("\\thaiSoSala", getutf8(0x0e28));
1840         addAccents("\\thaiSoRusi", getutf8(0x0e29));
1841         addAccents("\\thaiSoSua", getutf8(0x0e2a));
1842         addAccents("\\thaiHoHip", getutf8(0x0e2b));
1843         addAccents("\\thaiLoChula", getutf8(0x0e2c));
1844         addAccents("\\thaiOAng", getutf8(0x0e2d));
1845         addAccents("\\thaiHoNokhuk", getutf8(0x0e2e));
1846         addAccents("\\thaiPaiyannoi", getutf8(0x0e2f));
1847         addAccents("\\thaiSaraA", getutf8(0x0e30));
1848         addAccents("\\thaiMaiHanakat", getutf8(0x0e31));
1849         addAccents("\\thaiSaraAa", getutf8(0x0e32));
1850         addAccents("\\thaiSaraAm", getutf8(0x0e33));
1851         addAccents("\\thaiSaraI", getutf8(0x0e34));
1852         addAccents("\\thaiSaraIi", getutf8(0x0e35));
1853         addAccents("\\thaiSaraUe", getutf8(0x0e36));
1854         addAccents("\\thaiSaraUee", getutf8(0x0e37));
1855         addAccents("\\thaiSaraU", getutf8(0x0e38));
1856         addAccents("\\thaiSaraUu", getutf8(0x0e39));
1857         addAccents("\\thaiPhinthu", getutf8(0x0e3a));
1858         addAccents("\\thaiSaraE", getutf8(0x0e40));
1859         addAccents("\\thaiSaraAe", getutf8(0x0e41));
1860         addAccents("\\thaiSaraO", getutf8(0x0e42));
1861         addAccents("\\thaiSaraAiMaimuan", getutf8(0x0e43));
1862         addAccents("\\thaiSaraAiMaimalai", getutf8(0x0e44));
1863         addAccents("\\thaiLakkhangyao", getutf8(0x0e45));
1864         addAccents("\\thaiMaiyamok", getutf8(0x0e46));
1865         addAccents("\\thaiMaitaikhu", getutf8(0x0e47));
1866         addAccents("\\thaiMaiEk", getutf8(0x0e48));
1867         addAccents("\\thaiMaiTho", getutf8(0x0e49));
1868         addAccents("\\thaiMaiTri", getutf8(0x0e4a));
1869         addAccents("\\thaiMaiChattawa", getutf8(0x0e4b));
1870         addAccents("\\thaiThanthakhat", getutf8(0x0e4c));
1871         addAccents("\\thaiNikhahit", getutf8(0x0e4d));
1872         addAccents("\\thaiYamakkan", getutf8(0x0e4e));
1873         addAccents("\\thaiFongman", getutf8(0x0e4f));
1874         addAccents("\\thaizero", getutf8(0x0e50));
1875         addAccents("\\thaione", getutf8(0x0e51));
1876         addAccents("\\thaitwo", getutf8(0x0e52));
1877         addAccents("\\thaithree", getutf8(0x0e53));
1878         addAccents("\\thaifour", getutf8(0x0e54));
1879         addAccents("\\thaifive", getutf8(0x0e55));
1880         addAccents("\\thaisix", getutf8(0x0e56));
1881         addAccents("\\thaiseven", getutf8(0x0e57));
1882         addAccents("\\thaieight", getutf8(0x0e58));
1883         addAccents("\\thainine", getutf8(0x0e59));
1884         addAccents("\\thaiAngkhankhu", getutf8(0x0e5a));
1885         addAccents("\\thaiKhomut", getutf8(0x0e5b));
1886         addAccents("\\dag", getutf8(0x2020));
1887         addAccents("\\dagger", getutf8(0x2020));
1888         addAccents("\\textdagger", getutf8(0x2020));
1889         addAccents("\\ddag", getutf8(0x2021));
1890         addAccents("\\ddagger", getutf8(0x2021));
1891         addAccents("\\textdaggerdbl", getutf8(0x2021));
1892         addAccents("\\textbullet", getutf8(0x2022));
1893         addAccents("\\bullet", getutf8(0x2022));
1894         addAccents("\\dots", getutf8(0x2026));
1895         addAccents("\\ldots", getutf8(0x2026));
1896         addAccents("\\textellipsis", getutf8(0x2026));
1897         addAccents("\\textasciiacute", getutf8(0x2032));
1898         addAccents("\\prime", getutf8(0x2032));
1899         addAccents("\\textacutedbl", getutf8(0x2033));
1900         addAccents("\\dprime", getutf8(0x2033));
1901         addAccents("\\textasciigrave", getutf8(0x2035));
1902         addAccents("\\backprime", getutf8(0x2035));
1903         addAccents("\\textsubcircum{ }", getutf8(0x2038));
1904         addAccents("\\caretinsert", getutf8(0x2038));
1905         addAccents("\\textasteriskcentered", getutf8(0x204e));
1906         addAccents("\\ast", getutf8(0x204e));
1907         addAccents("\\textmho", getutf8(0x2127));
1908         addAccents("\\mho", getutf8(0x2127));
1909         addAccents("\\textleftarrow", getutf8(0x2190));
1910         addAccents("\\leftarrow", getutf8(0x2190));
1911         addAccents("\\textuparrow", getutf8(0x2191));
1912         addAccents("\\uparrow", getutf8(0x2191));
1913         addAccents("\\textrightarrow", getutf8(0x2192));
1914         addAccents("\\rightarrow", getutf8(0x2192));
1915         addAccents("\\textdownarrow", getutf8(0x2193));
1916         addAccents("\\downarrow", getutf8(0x2193));
1917         addAccents("\\textglobrise", getutf8(0x2197));
1918         addAccents("\\nearrow", getutf8(0x2197));
1919         addAccents("\\textglobfall", getutf8(0x2198));
1920         addAccents("\\searrow", getutf8(0x2198));
1921         addAccents("\\textsurd", getutf8(0x221a));
1922         addAccents("\\surd", getutf8(0x221a));
1923         addAccents("\\textbigcircle", getutf8(0x25ef));
1924         addAccents("\\bigcirc", getutf8(0x25ef));
1925         addAccents("\\FiveStar", getutf8(0x2605));
1926         addAccents("\\bigstar", getutf8(0x2605));
1927         addAccents("\\FiveStarOpen", getutf8(0x2606));
1928         addAccents("\\bigwhitestar", getutf8(0x2606));
1929         addAccents("\\Checkmark", getutf8(0x2713));
1930         addAccents("\\checkmark", getutf8(0x2713));
1931         addAccents("\\CrossMaltese", getutf8(0x2720));
1932         addAccents("\\maltese", getutf8(0x2720));
1933         addAccents("\\textlangle", getutf8(0x27e8));
1934         addAccents("\\langle", getutf8(0x27e8));
1935         addAccents("\\textrangle", getutf8(0x27e9));
1936         addAccents("\\rangle", getutf8(0x27e9));
1937 }
1938
1939 static void buildAccentsMap()
1940 {
1941         accents["imath"] = "ı";
1942         accents["i"] = "ı";
1943         accents["jmath"] = "ȷ";
1944         accents["cdot"] = "·";
1945         accents["textasciicircum"] = "^";
1946         accents["mathcircumflex"] = "^";
1947         accents["guillemotright"] = "»";
1948         accents["guillemotleft"] = "«";
1949         accents["hairspace"]     = getutf8(0xf0000);    // select from free unicode plane 15
1950         accents["thinspace"]     = getutf8(0xf0002);    // and used _only_ by findadv
1951         accents["negthinspace"]  = getutf8(0xf0003);    // to omit backslashed latex macros
1952         accents["medspace"]      = getutf8(0xf0004);    // See https://en.wikipedia.org/wiki/Private_Use_Areas
1953         accents["negmedspace"]   = getutf8(0xf0005);
1954         accents["thickspace"]    = getutf8(0xf0006);
1955         accents["negthickspace"] = getutf8(0xf0007);
1956         accents["lyx"]           = getutf8(0xf0010);    // Used logos
1957         accents["LyX"]           = getutf8(0xf0010);
1958         accents["tex"]           = getutf8(0xf0011);
1959         accents["TeX"]           = getutf8(0xf0011);
1960         accents["latex"]         = getutf8(0xf0012);
1961         accents["LaTeX"]         = getutf8(0xf0012);
1962         accents["latexe"]        = getutf8(0xf0013);
1963         accents["LaTeXe"]        = getutf8(0xf0013);
1964         accents["lyxarrow"]      = getutf8(0xf0020);
1965         accents["braceleft"]     = getutf8(0xf0030);
1966         accents["braceright"]    = getutf8(0xf0031);
1967         accents["lyxtilde"]      = getutf8(0xf0032);
1968         accents["sim"]           = getutf8(0xf0032);
1969         accents["backslash lyx"]           = getutf8(0xf0010);  // Used logos inserted with starting \backslash
1970         accents["backslash LyX"]           = getutf8(0xf0010);
1971         accents["backslash tex"]           = getutf8(0xf0011);
1972         accents["backslash TeX"]           = getutf8(0xf0011);
1973         accents["backslash latex"]         = getutf8(0xf0012);
1974         accents["backslash LaTeX"]         = getutf8(0xf0012);
1975         accents["backslash latexe"]        = getutf8(0xf0013);
1976         accents["backslash LaTeXe"]        = getutf8(0xf0013);
1977         accents["backslash lyxarrow"]      = getutf8(0xf0020);
1978         accents["ddot{\\imath}"] = "ï";
1979         buildaccent("ddot", "aAeEhHiIoOtuUwWxXyY",
1980                     "äÄëËḧḦïÏöÖẗüÜẅẄẍẌÿŸ");   // umlaut
1981         buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1982                     "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ");      // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1983         accents["acute{\\imath}"] = "í";
1984         buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1985                     "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1986         buildaccent("dacute|H|h", "oOuU", "őŐűŰ");  // double acute
1987         buildaccent("mathring|r", "aAuUwy",
1988                     "åÅůŮẘẙ");  // ring
1989         accents["check{\\imath}"] = "ǐ";
1990         accents["check{\\jmath}"] = "ǰ";
1991         buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1992                     "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ");        // caron
1993         accents["hat{\\imath}"] = "î";
1994         accents["hat{\\jmath}"] = "ĵ";
1995         buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1996                     "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ");  // circ
1997         accents["bar{\\imath}"] = "ī";
1998         buildaccent("bar|=", "aAeEiIoOuUyY",
1999                     "āĀēĒīĪōŌūŪȳȲ");        // macron
2000         accents["tilde{\\imath}"] = "ĩ";
2001         buildaccent("tilde", "aAeEiInNoOuUvVyY",
2002                     "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ");  // tilde
2003         accents["breve{\\imath}"] = "ĭ";
2004         buildaccent("breve|u", "aAeEgGiIoOuU",
2005                     "ăĂĕĔğĞĭĬŏŎŭŬ");        // breve
2006         accents["grave{\\imath}"] = "ì";
2007         buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
2008                     "àÀèÈìÌòÒùÙǹǸẁẀỳỲ");    // grave
2009         buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
2010                     "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ");      // dot below
2011         buildaccent("ogonek|k", "AaEeIiUuOo",
2012                     "ĄąĘęĮįŲųǪǫ");    // ogonek
2013         buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
2014                     "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ");        // cedilla
2015         buildaccent("subring|textsubring", "Aa",
2016                     "Ḁḁ");  // subring
2017         buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
2018                     "ḒḓḘḙḼḽṊṋṰṱṶṷ");    // subcircum
2019         buildaccent("subtilde|textsubtilde", "EeIiUu",
2020                     "ḚḛḬḭṴṵ");      // subtilde
2021         accents["dgrave{\\imath}"] = "ȉ";
2022         accents["textdoublegrave{\\i}"] = "ȉ";
2023         buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
2024                     "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
2025         accents["rcap{\\imath}"] = "ȋ";
2026         accents["textroundcap{\\i}"] = "ȋ";
2027         buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
2028                     "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
2029         buildaccent("slashed", "oO",
2030                     "øØ"); // slashed
2031         fillMissingUnicodesymbols(); // Add some still not handled entries contained in 'unicodesynbols'
2032         // LYXERR0("Number of accents " << accents.size());
2033 }
2034
2035 /*
2036  * Created accents in math or regexp environment
2037  * are macros, but we need the utf8 equivalent
2038  */
2039 void Intervall::removeAccents()
2040 {
2041         if (accents.empty())
2042                 buildAccentsMap();
2043         static regex const accre("\\\\("
2044                                  "([\\S]|[A-Za-z]+)\\{[^\\{\\}]+\\}"
2045                                  "|("
2046                                  "(backslash ([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))"
2047                                  "|[A-Za-z]+"
2048                                  ")"
2049                                  "(?![a-zA-Z]))");
2050         smatch sub;
2051         for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
2052                 sub = *itacc;
2053                 string key = sub.str(1);
2054                 AccentsIterator it_ac = accents.find(key);
2055                 if (it_ac != accents.end()) {
2056                         string val = it_ac->second;
2057                         size_t pos = sub.position(size_t(0));
2058                         for (size_t i = 0; i < val.size(); i++) {
2059                                 par[pos+i] = val[i];
2060                         }
2061                         // Remove possibly following space too
2062                         if (par[pos+sub.str(0).size()] == ' ')
2063                                 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
2064                         else
2065                                 addIntervall(pos+val.size(), pos + sub.str(0).size());
2066                         for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
2067                                 // remove traces of any remaining chars
2068                                 par[i] = ' ';
2069                         }
2070                 }
2071                 else {
2072                         LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
2073                 }
2074         }
2075 }
2076
2077 void Intervall::handleOpenP(int i)
2078 {
2079         actualdeptindex++;
2080         depts[actualdeptindex] = i+1;
2081         closes[actualdeptindex] = -1;
2082         checkDepthIndex(actualdeptindex);
2083 }
2084
2085 void Intervall::handleCloseP(int i, bool closingAllowed)
2086 {
2087         if (actualdeptindex <= 0) {
2088                 if (! closingAllowed)
2089                         LYXERR(Debug::FINDVERBOSE, "Bad closing parenthesis in latex");  /* should not happen, but the latex input may be wrong */
2090                 // if we are at the very end
2091                 addIntervall(i, i+1);
2092         }
2093         else {
2094                 closes[actualdeptindex] = i+1;
2095                 actualdeptindex--;
2096         }
2097 }
2098
2099 void Intervall::resetOpenedP(int openPos)
2100 {
2101         // Used as initializer for foreignlanguage entry
2102         actualdeptindex = 1;
2103         depts[1] = openPos+1;
2104         closes[1] = -1;
2105 }
2106
2107 int Intervall::previousNotIgnored(int start) const
2108 {
2109         int idx = 0;                          /* int intervalls */
2110         for (idx = ignoreidx; idx >= 0; --idx) {
2111                 if (start > borders[idx].upper)
2112                         return start;
2113                 if (start >= borders[idx].low)
2114                         start = borders[idx].low-1;
2115         }
2116         return start;
2117 }
2118
2119 int Intervall::nextNotIgnored(int start) const
2120 {
2121         int idx = 0;                          /* int intervalls */
2122         for (idx = 0; idx <= ignoreidx; idx++) {
2123                 if (start < borders[idx].low)
2124                         return start;
2125                 if (start < borders[idx].upper)
2126                         start = borders[idx].upper;
2127         }
2128         return start;
2129 }
2130
2131 typedef unordered_map<string, KeyInfo> KeysMap;
2132 typedef unordered_map<string, KeyInfo>::const_iterator KeysIterator;
2133 typedef vector< KeyInfo> Entries;
2134 static KeysMap keys = unordered_map<string, KeyInfo>();
2135
2136 class LatexInfo {
2137 private:
2138         int entidx_;
2139         Entries entries_;
2140         Intervall interval_;
2141         void buildKeys(bool);
2142         void buildEntries(bool);
2143         void makeKey(const string &, KeyInfo, bool isPatternString);
2144         void processRegion(int start, int region_end); /*  remove {} parts */
2145         void removeHead(KeyInfo const &, int count=0);
2146
2147 public:
2148         LatexInfo(string const & par, bool isPatternString)
2149                 : entidx_(-1), interval_(isPatternString, par)
2150         {
2151                 buildKeys(isPatternString);
2152                 entries_ = vector<KeyInfo>();
2153                 buildEntries(isPatternString);
2154         }
2155         int getFirstKey() {
2156                 entidx_ = 0;
2157                 if (entries_.empty()) {
2158                         return -1;
2159                 }
2160                 if (entries_[0].keytype == KeyInfo::isTitle) {
2161                         interval_.hasTitle = true;
2162                         if (! entries_[0].disabled) {
2163                                 interval_.titleValue = entries_[0].head;
2164                         }
2165                         else {
2166                                 interval_.titleValue = "";
2167                         }
2168                         removeHead(entries_[0]);
2169                         if (entries_.size() > 1)
2170                                 return 1;
2171                         else
2172                                 return -1;
2173                 }
2174                 return 0;
2175         }
2176         int getNextKey() {
2177                 entidx_++;
2178                 if (int(entries_.size()) > entidx_) {
2179                         return entidx_;
2180                 }
2181                 else {
2182                         return -1;
2183                 }
2184         }
2185         bool setNextKey(int idx) {
2186                 if ((idx == entidx_) && (entidx_ >= 0)) {
2187                         entidx_--;
2188                         return true;
2189                 }
2190                 else
2191                         return false;
2192         }
2193         int find(int start, KeyInfo::KeyType keytype) const {
2194                 if (start < 0)
2195                         return -1;
2196                 int tmpIdx = start;
2197                 while (tmpIdx < int(entries_.size())) {
2198                         if (entries_[tmpIdx].keytype == keytype)
2199                                 return tmpIdx;
2200                         tmpIdx++;
2201                 }
2202                 return -1;
2203         }
2204         int process(ostringstream & os, KeyInfo const & actual);
2205         int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
2206         // string show(int lastpos) { return interval.show(lastpos);}
2207         int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);}
2208         KeyInfo &getKeyInfo(int keyinfo) {
2209                 static KeyInfo invalidInfo = KeyInfo();
2210                 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
2211                         return invalidInfo;
2212                 else
2213                         return entries_[keyinfo];
2214         }
2215         void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);}
2216         void addIntervall(int low, int up) { interval_.addIntervall(low, up); }
2217 };
2218
2219
2220 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
2221 {
2222         int skip = 0;
2223         int depth = 0;
2224         for (int i = start; i < end; i += 1 + skip) {
2225                 char c;
2226                 c = par[i];
2227                 skip = 0;
2228                 if (c == '\\') skip = 1;
2229                 else if (c == up) {
2230                         depth++;
2231                 }
2232                 else if (c == down) {
2233                         if (depth == 0) {
2234                                 repeat--;
2235                                 if ((repeat <= 0) || (par[i+1] != up))
2236                                         return i;
2237                         }
2238                         --depth;
2239                 }
2240         }
2241         return end;
2242 }
2243
2244 class MathInfo {
2245         class MathEntry {
2246         public:
2247                 string wait;
2248                 size_t mathEnd;
2249                 size_t mathpostfixsize;
2250                 size_t mathStart;
2251                 size_t mathprefixsize;
2252                 size_t mathSize;
2253         };
2254         size_t actualIdx_;
2255         vector<MathEntry> entries_;
2256 public:
2257         MathInfo() {
2258                 actualIdx_ = 0;
2259         }
2260         void insert(string const & wait, size_t start, size_t prefixsize, size_t end, size_t postfixsize) {
2261                 MathEntry m = MathEntry();
2262                 m.wait = wait;
2263                 m.mathStart = start;
2264                 m.mathprefixsize = prefixsize;
2265                 m.mathEnd = end + postfixsize;
2266                 m.mathpostfixsize = postfixsize;
2267                 m.mathSize = m.mathEnd - m.mathStart;
2268                 entries_.push_back(m);
2269         }
2270         bool empty() const { return entries_.empty(); }
2271         size_t getEndPos() const {
2272                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2273                         return 0;
2274                 }
2275                 return entries_[actualIdx_].mathEnd;
2276         }
2277         size_t getStartPos() const {
2278                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2279                         return 100000;                    /*  definitely enough? */
2280                 }
2281                 return entries_[actualIdx_].mathStart;
2282         }
2283         size_t getPrefixSize() const {
2284                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2285                         return 0;
2286                 }
2287                 return entries_[actualIdx_].mathprefixsize;
2288         }
2289         size_t getPostfixSize() const {
2290                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2291                         return 0;
2292                 }
2293                 return entries_[actualIdx_].mathpostfixsize;
2294         }
2295         size_t getFirstPos() {
2296                 actualIdx_ = 0;
2297                 return getStartPos();
2298         }
2299         size_t getSize() const {
2300                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2301                         return size_t(0);
2302                 }
2303                 return entries_[actualIdx_].mathSize;
2304         }
2305         void incrEntry() { actualIdx_++; }
2306 };
2307
2308 void LatexInfo::buildEntries(bool isPatternString)
2309 {
2310         static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|x?x?alignat)\\*?\\})(\\{[0-9]+\\})?)");
2311         static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
2312         static bool disableLanguageOverride = false;
2313         smatch sub, submath;
2314         bool evaluatingRegexp = false;
2315         MathInfo mi;
2316         bool evaluatingMath = false;
2317         bool evaluatingCode = false;
2318         size_t codeEnd = 0;
2319         bool evaluatingOptional = false;
2320         size_t optionalEnd = 0;
2321         int codeStart = -1;
2322         KeyInfo found;
2323         bool math_end_waiting = false;
2324         size_t math_pos = 10000;
2325         size_t math_prefix_size = 1;
2326         string math_end;
2327         static vector<string> usedText = vector<string>();
2328         static bool removeMathHull = false;
2329
2330         interval_.removeAccents();
2331
2332         for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
2333                 submath = *itmath;
2334                 if ((submath.position(2) - submath.position(0)) %2 == 1) {
2335                         // prefixed by odd count of '\\'
2336                         continue;
2337                 }
2338                 if (math_end_waiting) {
2339                         size_t pos = submath.position(size_t(2));
2340                         if ((math_end == "$") &&
2341                                         (submath.str(2) == "$")) {
2342                                 mi.insert("$", math_pos, 1, pos, 1);
2343                                 math_end_waiting = false;
2344                         }
2345                         else if ((math_end == "\\]") &&
2346                                  (submath.str(2) == "\\]")) {
2347                                 mi.insert("\\]", math_pos, 2, pos, 2);
2348                                 math_end_waiting = false;
2349                         }
2350                         else if ((submath.str(3).compare("end") == 0) &&
2351                                  (submath.str(5).compare(math_end) == 0)) {
2352                                 mi.insert(math_end, math_pos, math_prefix_size, pos, submath.str(2).length());
2353                                 math_end_waiting = false;
2354                         }
2355                         else
2356                                 continue;
2357                 }
2358                 else {
2359                         if (submath.str(3).compare("begin") == 0) {
2360                                 math_end_waiting = true;
2361                                 math_end = submath.str(5);
2362                                 math_pos = submath.position(size_t(2));
2363                                 math_prefix_size = submath.str(2).length();
2364                         }
2365                         else if (submath.str(2).compare("\\[") == 0) {
2366                                 math_end_waiting = true;
2367                                 math_end = "\\]";
2368                                 math_pos = submath.position(size_t(2));
2369                         }
2370                         else if (submath.str(2) == "$") {
2371                                 size_t pos = submath.position(size_t(2));
2372                                 math_end_waiting = true;
2373                                 math_end = "$";
2374                                 math_pos = pos;
2375                         }
2376                 }
2377         }
2378         // Ignore language if there is math somewhere in pattern-string
2379         if (isPatternString) {
2380                 for (auto s: usedText) {
2381                         // Remove entries created in previous search runs
2382                         keys.erase(s);
2383                 }
2384                 usedText = vector<string>();
2385                 if (! mi.empty()) {
2386                         // Disable language
2387                         keys["foreignlanguage"].disabled = true;
2388                         disableLanguageOverride = true;
2389                         removeMathHull = false;
2390                 }
2391                 else {
2392                         removeMathHull = true;  // used later if not isPatternString
2393                         disableLanguageOverride = false;
2394                 }
2395         }
2396         else {
2397                 if (disableLanguageOverride) {
2398                         keys["foreignlanguage"].disabled = true;
2399                 }
2400         }
2401         math_pos = mi.getFirstPos();
2402         for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
2403                 sub = *it;
2404                 if ((sub.position(2) - sub.position(0)) %2 == 1) {
2405                         // prefixed by odd count of '\\'
2406                         continue;
2407                 }
2408                 string key = sub.str(5);
2409                 if (key == "") {
2410                         if (sub.str(2)[0] == '\\')
2411                                 key = sub.str(2)[1];
2412                         else {
2413                                 key = sub.str(2);
2414                         }
2415                 }
2416                 KeysIterator it_key = keys.find(key);
2417                 if (it_key != keys.end()) {
2418                         if (it_key->second.keytype == KeyInfo::headRemove) {
2419                                 KeyInfo found1 = it_key->second;
2420                                 found1.disabled = true;
2421                                 found1.head = "\\" + key + "{";
2422                                 found1._tokenstart = sub.position(size_t(2));
2423                                 found1._tokensize = found1.head.length();
2424                                 found1._dataStart = found1._tokenstart + found1.head.length();
2425                                 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
2426                                 found1._dataEnd = endpos;
2427                                 removeHead(found1);
2428                                 continue;
2429                         }
2430                 }
2431                 if (evaluatingRegexp) {
2432                         if (sub.str(3).compare("endregexp") == 0) {
2433                                 evaluatingRegexp = false;
2434                                 // found._tokenstart already set
2435                                 found._dataEnd = sub.position(size_t(2)) + 13;
2436                                 found._dataStart = found._dataEnd;
2437                                 found._tokensize = found._dataEnd - found._tokenstart;
2438                                 found.parenthesiscount = 0;
2439                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2440                         }
2441                         else {
2442                                 continue;
2443                         }
2444                 }
2445                 else {
2446                         if (evaluatingMath) {
2447                                 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
2448                                         continue;
2449                                 evaluatingMath = false;
2450                                 mi.incrEntry();
2451                                 math_pos = mi.getStartPos();
2452                         }
2453                         if (it_key == keys.end()) {
2454                                 found = KeyInfo(KeyInfo::isStandard, 0, true);
2455                                 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
2456                                 found = KeyInfo(KeyInfo::isText, 0, false);
2457                                 if (isPatternString) {
2458                                         found.keytype = KeyInfo::isChar;
2459                                         found.disabled = false;
2460                                         found.used = true;
2461                                 }
2462                                 keys[key] = found;
2463                                 usedText.push_back(key);
2464                         }
2465                         else
2466                                 found = keys[key];
2467                         if (key.compare("regexp") == 0) {
2468                                 evaluatingRegexp = true;
2469                                 found._tokenstart = sub.position(size_t(2));
2470                                 found._tokensize = 0;
2471                                 continue;
2472                         }
2473                 }
2474                 // Handle the other params of key
2475                 if (found.keytype == KeyInfo::isIgnored)
2476                         continue;
2477                 else if (found.keytype == KeyInfo::isMath) {
2478                         if (size_t(sub.position(size_t(2))) == math_pos) {
2479                                 found = keys[key];
2480                                 found._tokenstart = sub.position(size_t(2));
2481                                 found._tokensize = mi.getSize();
2482                                 found._dataEnd = found._tokenstart + found._tokensize;
2483                                 found._dataStart = found._dataEnd;
2484                                 found.parenthesiscount = 0;
2485                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2486                                 if (removeMathHull) {
2487                                         interval_.addIntervall(found._tokenstart, found._tokenstart + mi.getPrefixSize());
2488                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize(), found._dataEnd);
2489                                 }
2490                                 else {
2491                                         // Treate all math constructs as simple math
2492                                         interval_.par[found._tokenstart] = '$';
2493                                         interval_.par[found._dataEnd - mi.getPostfixSize()] = '$';
2494                                         interval_.addIntervall(found._tokenstart + 1, found._tokenstart + mi.getPrefixSize());
2495                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize() + 1, found._dataEnd);
2496                                 }
2497                                 evaluatingMath = true;
2498                         }
2499                         else {
2500                                 // begin|end of unknown env, discard
2501                                 // First handle tables
2502                                 // longtable|tabular
2503                                 bool discardComment;
2504                                 found = keys[key];
2505                                 found.keytype = KeyInfo::doRemove;
2506                                 if ((sub.str(7).compare("longtable") == 0) ||
2507                                                 (sub.str(7).compare("tabular") == 0)) {
2508                                         discardComment = true;        /* '%' */
2509                                 }
2510                                 else {
2511                                         discardComment = false;
2512                                         static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
2513                                         smatch sub2;
2514                                         string token = sub.str(7);
2515                                         if (regex_match(token, sub2, removeArgs)) {
2516                                                 found.keytype = KeyInfo::removeWithArg;
2517                                         }
2518                                 }
2519                                 // discard spaces before pos(2)
2520                                 int pos = sub.position(size_t(2));
2521                                 int count;
2522                                 for (count = 0; pos - count > 0; count++) {
2523                                         char c = interval_.par[pos-count-1];
2524                                         if (discardComment) {
2525                                                 if ((c != ' ') && (c != '%'))
2526                                                         break;
2527                                         }
2528                                         else if (c != ' ')
2529                                                 break;
2530                                 }
2531                                 found._tokenstart = pos - count;
2532                                 if (sub.str(3).compare(0, 5, "begin") == 0) {
2533                                         size_t pos1 = pos + sub.str(2).length();
2534                                         if (sub.str(7).compare("cjk") == 0) {
2535                                                 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2536                                                 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2537                                                         pos1 += 2;
2538                                                 found.keytype = KeyInfo::isMain;
2539                                                 found._dataStart = pos1;
2540                                                 found._dataEnd = interval_.par.length();
2541                                                 found.disabled = keys["foreignlanguage"].disabled;
2542                                                 found.used = keys["foreignlanguage"].used;
2543                                                 found._tokensize = pos1 - found._tokenstart;
2544                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2545                                         }
2546                                         else {
2547                                                 // Swallow possible optional params
2548                                                 while (interval_.par[pos1] == '[') {
2549                                                         pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2550                                                 }
2551                                                 // Swallow also the eventual parameter
2552                                                 if (interval_.par[pos1] == '{') {
2553                                                         found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2554                                                 }
2555                                                 else {
2556                                                         found._dataEnd = pos1;
2557                                                 }
2558                                                 found._dataStart = found._dataEnd;
2559                                                 found._tokensize = count + found._dataEnd - pos;
2560                                                 found.parenthesiscount = 0;
2561                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2562                                                 found.disabled = true;
2563                                         }
2564                                 }
2565                                 else {
2566                                         // Handle "\end{...}"
2567                                         found._dataStart = pos + sub.str(2).length();
2568                                         found._dataEnd = found._dataStart;
2569                                         found._tokensize = count + found._dataEnd - pos;
2570                                         found.parenthesiscount = 0;
2571                                         found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2572                                         found.disabled = true;
2573                                 }
2574                         }
2575                 }
2576                 else if (found.keytype != KeyInfo::isRegex) {
2577                         found._tokenstart = sub.position(size_t(2));
2578                         if (found.parenthesiscount == 0) {
2579                                 // Probably to be discarded
2580                                 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2581                                 char following = interval_.par[following_pos];
2582                                 if (following == ' ')
2583                                         found.head = "\\" + sub.str(5) + " ";
2584                                 else if (following == '=') {
2585                                         // like \uldepth=1000pt
2586                                         found.head = sub.str(2);
2587                                 }
2588                                 else
2589                                         found.head = "\\" + key;
2590                                 found._tokensize = found.head.length();
2591                                 found._dataEnd = found._tokenstart + found._tokensize;
2592                                 found._dataStart = found._dataEnd;
2593                         }
2594                         else {
2595                                 int params = found._tokenstart + key.length() + 1;
2596                                 if (evaluatingOptional) {
2597                                         if (size_t(found._tokenstart) > optionalEnd) {
2598                                                 evaluatingOptional = false;
2599                                         }
2600                                         else {
2601                                                 found.disabled = true;
2602                                         }
2603                                 }
2604                                 int optend = params;
2605                                 while (interval_.par[optend] == '[') {
2606                                         // discard optional parameters
2607                                         optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2608                                 }
2609                                 if (optend > params) {
2610                                         key += interval_.par.substr(params, optend-params);
2611                                         evaluatingOptional = true;
2612                                         optionalEnd = optend;
2613                                         if (found.keytype == KeyInfo::isSectioning) {
2614                                                 // Remove optional values (but still keep in header)
2615                                                 interval_.addIntervall(params, optend);
2616                                         }
2617                                 }
2618                                 string token = sub.str(7);
2619                                 int closings;
2620                                 if (interval_.par[optend] != '{') {
2621                                         closings = 0;
2622                                         found.parenthesiscount = 0;
2623                                         found.head = "\\" + key;
2624                                 }
2625                                 else
2626                                         closings = found.parenthesiscount;
2627                                 if (found.parenthesiscount == 1) {
2628                                         found.head = "\\" + key + "{";
2629                                 }
2630                                 else if (found.parenthesiscount > 1) {
2631                                         if (token != "") {
2632                                                 found.head = sub.str(2) + "{";
2633                                                 closings = found.parenthesiscount - 1;
2634                                         }
2635                                         else {
2636                                                 found.head = "\\" + key + "{";
2637                                         }
2638                                 }
2639                                 found._tokensize = found.head.length();
2640                                 found._dataStart = found._tokenstart + found.head.length();
2641                                 if (found.keytype == KeyInfo::doRemove) {
2642                                         if (closings > 0) {
2643                                                 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2644                                                 if (endpar >= interval_.par.length())
2645                                                         found._dataStart = interval_.par.length();
2646                                                 else
2647                                                         found._dataStart = endpar;
2648                                                 found._tokensize = found._dataStart - found._tokenstart;
2649                                         }
2650                                         else {
2651                                                 found._dataStart = found._tokenstart + found._tokensize;
2652                                         }
2653                                         closings = 0;
2654                                 }
2655                                 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2656                                         found._dataStart += 15;
2657                                 }
2658                                 size_t endpos;
2659                                 if (closings < 1)
2660                                         endpos = found._dataStart - 1;
2661                                 else
2662                                         endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2663                                 if (found.keytype == KeyInfo::isList) {
2664                                         // Check if it really is list env
2665                                         static regex const listre("^([a-z]+)$");
2666                                         smatch sub2;
2667                                         if (!regex_match(token, sub2, listre)) {
2668                                                 // Change the key of this entry. It is not in a list/item environment
2669                                                 found.keytype = KeyInfo::endArguments;
2670                                         }
2671                                 }
2672                                 if (found.keytype == KeyInfo::noMain) {
2673                                         evaluatingCode = true;
2674                                         codeEnd = endpos;
2675                                         codeStart = found._dataStart;
2676                                 }
2677                                 else if (evaluatingCode) {
2678                                         if (size_t(found._dataStart) > codeEnd)
2679                                                 evaluatingCode = false;
2680                                         else if (found.keytype == KeyInfo::isMain) {
2681                                                 // Disable this key, treate it as standard
2682                                                 found.keytype = KeyInfo::isStandard;
2683                                                 found.disabled = true;
2684                                                 if ((codeEnd +1 >= interval_.par.length()) &&
2685                                                                 (found._tokenstart == codeStart)) {
2686                                                         // trickery, because the code inset starts
2687                                                         // with \selectlanguage ...
2688                                                         codeEnd = endpos;
2689                                                         if (entries_.size() > 1) {
2690                                                                 entries_[entries_.size()-1]._dataEnd = codeEnd;
2691                                                         }
2692                                                 }
2693                                         }
2694                                 }
2695                                 if ((endpos == interval_.par.length()) &&
2696                                                 (found.keytype == KeyInfo::doRemove)) {
2697                                         // Missing closing => error in latex-input?
2698                                         // therefore do not delete remaining data
2699                                         found._dataStart -= 1;
2700                                         found._dataEnd = found._dataStart;
2701                                 }
2702                                 else
2703                                         found._dataEnd = endpos;
2704                         }
2705                         if (isPatternString) {
2706                                 keys[key].used = true;
2707                         }
2708                 }
2709                 entries_.push_back(found);
2710         }
2711 }
2712
2713 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2714 {
2715         stringstream s(keysstring);
2716         string key;
2717         const char delim = '|';
2718         while (getline(s, key, delim)) {
2719                 KeyInfo keyII(keyI);
2720                 if (isPatternString) {
2721                         keyII.used = false;
2722                 }
2723                 else if ( !keys[key].used)
2724                         keyII.disabled = true;
2725                 keys[key] = keyII;
2726         }
2727 }
2728
2729 void LatexInfo::buildKeys(bool isPatternString)
2730 {
2731
2732         static bool keysBuilt = false;
2733         if (keysBuilt && !isPatternString) return;
2734
2735         // Keys to ignore in any case
2736         makeKey("text|textcyrillic|lyxmathsym|ensuremath", KeyInfo(KeyInfo::headRemove, 1, true), true);
2737         // Known standard keys with 1 parameter.
2738         // Split is done, if not at start of region
2739         makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2740         makeKey("textbf",               KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2741         makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2742         makeKey("uuline|uline|uwave",   KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2743         makeKey("emph|noun",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2744         makeKey("sout|xout",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2745
2746         makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2747                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2748         makeKey("section*|subsection*|subsubsection*|paragraph*",
2749                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2750         makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2751         makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2752         // Regex
2753         makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2754
2755         // Split is done, if not at start of region
2756         makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2757         makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2758
2759         // Split is done always.
2760         makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2761
2762         // Known charaters
2763         // No split
2764         makeKey("backslash|textbackslash|slash",  KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2765         makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2766         makeKey("textasciiacute|texemdash",       KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2767         makeKey("dots|ldots",                     KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2768         // Spaces
2769         makeKey("quad|qquad|hfill|dotfill",               KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2770         makeKey("textvisiblespace|nobreakspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2771         makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2772         makeKey("thickspace|medspace|thinspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2773         // Skip
2774         // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2775         // Custom space/skip, remove the content (== length value)
2776         makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2777         // Found in fr/UserGuide.lyx
2778         makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2779         // quotes
2780         makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2781         makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2782         // Known macros to remove (including their parameter)
2783         // No split
2784         makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2785         makeKey("addtocounter|setlength",                 KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2786         // handle like standard keys with 1 parameter.
2787         makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2788
2789         if (ignoreFormats.getDeleted()) {
2790                 // Ignore deleted text
2791                 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2792         }
2793         else {
2794                 // but preserve added text
2795                 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2796         }
2797         makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2798
2799         // Macros to remove, but let the parameter survive
2800         // No split
2801         makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2802
2803         // Remove language spec from content of these insets
2804         makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2805
2806         // Same effect as previous, parameter will survive (because there is no one anyway)
2807         // No split
2808         makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2809         // Remove table decorations
2810         makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2811         // Discard shape-header.
2812         // For footnote or shortcut too, because of lang settings
2813         // and wrong handling if used 'KeyInfo::noMain'
2814         makeKey("circlepar|diamondpar|heartpar|nutpar",  KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2815         makeKey("trianglerightpar|hexagonpar|starpar",   KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2816         makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2817         makeKey("triangleleftpar|shapepar|dropuppar",    KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2818         makeKey("hphantom|vphantom|note|footnote|shortcut|include|includegraphics",     KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2819         makeKey("textgreek|textcyrillic", KeyInfo(KeyInfo::isStandard, 1, true), false);
2820         makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2821         // like ('tiny{}' or '\tiny ' ... )
2822         makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, ignoreFormats.getSize()), isPatternString);
2823
2824         // Survives, like known character
2825         // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2826         makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2827         makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2828
2829         makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2830         makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2831         makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2832
2833         makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2834         // Remove RTL/LTR marker
2835         makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2836         makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2837         makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2838         makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2839         makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2840         makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2841         makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2842         if (isPatternString) {
2843                 // Allow the first searched string to rebuild the keys too
2844                 keysBuilt = false;
2845         }
2846         else {
2847                 // no need to rebuild again
2848                 keysBuilt = true;
2849         }
2850 }
2851
2852 /*
2853  * Keep the list of actual opened parentheses actual
2854  * (e.g. depth == 4 means there are 4 '{' not processed yet)
2855  */
2856 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2857 {
2858         int skip = 0;
2859         for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2860                 char c;
2861                 c = par[i];
2862                 skip = 0;
2863                 if (c == '\\') skip = 1;
2864                 else if (c == '{') {
2865                         handleOpenP(i);
2866                 }
2867                 else if (c == '}') {
2868                         handleCloseP(i, closingAllowed);
2869                 }
2870         }
2871 }
2872
2873 #if (0)
2874 string Intervall::show(int lastpos)
2875 {
2876         int idx = 0;                          /* int intervalls */
2877         string s;
2878         int i = 0;
2879         if ((unsigned) lastpos > par.size())
2880                 lastpos = par.size();
2881         for (idx = 0; idx <= ignoreidx; idx++) {
2882                 while (i < lastpos) {
2883                         int printsize;
2884                         if (i <= borders[idx].low) {
2885                                 if (borders[idx].low > lastpos)
2886                                         printsize = lastpos - i;
2887                                 else
2888                                         printsize = borders[idx].low - i;
2889                                 s += par.substr(i, printsize);
2890                                 i += printsize;
2891                                 if (i >= borders[idx].low)
2892                                         i = borders[idx].upper;
2893                         }
2894                         else {
2895                                 i = borders[idx].upper;
2896                                 break;
2897                         }
2898                 }
2899         }
2900         if (lastpos > i) {
2901                 s += par.substr(i, lastpos-i);
2902         }
2903         return s;
2904 }
2905 #endif
2906
2907 void Intervall::output(ostringstream &os, int lastpos)
2908 {
2909         // get number of chars to output
2910         int idx = 0;                          /* int intervalls */
2911         int i = 0;
2912         int printed = 0;
2913         string startTitle = titleValue;
2914         for (idx = 0; idx <= ignoreidx; idx++) {
2915                 if (i < lastpos) {
2916                         if (i <= borders[idx].low) {
2917                                 int printsize;
2918                                 if (borders[idx].low > lastpos)
2919                                         printsize = lastpos - i;
2920                                 else
2921                                         printsize = borders[idx].low - i;
2922                                 if (printsize > 0) {
2923                                         os << startTitle << par.substr(i, printsize);
2924                                         i += printsize;
2925                                         printed += printsize;
2926                                         startTitle = "";
2927                                 }
2928                                 handleParentheses(i, false);
2929                                 if (i >= borders[idx].low)
2930                                         i = borders[idx].upper;
2931                         }
2932                         else {
2933                                 i = borders[idx].upper;
2934                         }
2935                 }
2936                 else
2937                         break;
2938         }
2939         if (lastpos > i) {
2940                 os << startTitle << par.substr(i, lastpos-i);
2941                 printed += lastpos-i;
2942         }
2943         handleParentheses(lastpos, false);
2944         int startindex;
2945         if (keys["foreignlanguage"].disabled)
2946                 startindex = actualdeptindex-langcount;
2947         else
2948                 startindex = actualdeptindex;
2949         for (int i = startindex; i > 0; --i) {
2950                 os << "}";
2951         }
2952         if (hasTitle && (printed > 0))
2953                 os << "}";
2954         if (! isPatternString_)
2955                 os << "\n";
2956         handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2957 }
2958
2959 void LatexInfo::processRegion(int start, int region_end)
2960 {
2961         while (start < region_end) {          /* Let {[} and {]} survive */
2962                 int cnt = interval_.isOpeningPar(start);
2963                 if (cnt == 1) {
2964                         // Closing is allowed past the region
2965                         int closing = interval_.findclosing(start+1, interval_.par.length());
2966                         interval_.addIntervall(start, start+1);
2967                         interval_.addIntervall(closing, closing+1);
2968                 }
2969                 else if (cnt == 3)
2970                         start += 2;
2971                 start = interval_.nextNotIgnored(start+1);
2972         }
2973 }
2974
2975 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2976 {
2977         if (actual.parenthesiscount == 0) {
2978                 // "{\tiny{} ...}" ==> "{{} ...}"
2979                 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2980         }
2981         else {
2982                 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2983                 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2984                 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2985         }
2986 }
2987
2988 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2989 {
2990         int nextKeyIdx = 0;
2991         switch (actual.keytype)
2992         {
2993         case KeyInfo::isTitle: {
2994                 removeHead(actual);
2995                 nextKeyIdx = getNextKey();
2996                 break;
2997         }
2998         case KeyInfo::cleanToStart: {
2999                 actual._dataEnd = actual._dataStart;
3000                 nextKeyIdx = getNextKey();
3001                 // Search for end of arguments
3002                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3003                 if (tmpIdx > 0) {
3004                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3005                                 entries_[i].disabled = true;
3006                         }
3007                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3008                 }
3009                 while (interval_.par[actual._dataEnd] == ' ')
3010                         actual._dataEnd++;
3011                 interval_.addIntervall(0, actual._dataEnd+1);
3012                 interval_.actualdeptindex = 0;
3013                 interval_.depts[0] = actual._dataEnd+1;
3014                 interval_.closes[0] = -1;
3015                 break;
3016         }
3017         case KeyInfo::isText:
3018                 interval_.par[actual._tokenstart] = '#';
3019                 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
3020                 nextKeyIdx = getNextKey();
3021                 break;
3022         case KeyInfo::noContent: {          /* char like "\hspace{2cm}" */
3023                 if (actual.disabled)
3024                         interval_.addIntervall(actual._tokenstart, actual._dataEnd);
3025                 else
3026                         interval_.addIntervall(actual._dataStart, actual._dataEnd);
3027         }
3028                 // fall through
3029         case KeyInfo::isChar: {
3030                 nextKeyIdx = getNextKey();
3031                 break;
3032         }
3033         case KeyInfo::isSize: {
3034                 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
3035                         if (actual.parenthesiscount == 0)
3036                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
3037                         else {
3038                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3039                         }
3040                         nextKeyIdx = getNextKey();
3041                 } else {
3042                         // Here _dataStart points to '{', so correct it
3043                         actual._dataStart += 1;
3044                         actual._tokensize += 1;
3045                         actual.parenthesiscount = 1;
3046                         if (interval_.par[actual._dataStart] == '}') {
3047                                 // Determine the end if used like '{\tiny{}...}'
3048                                 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
3049                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3050                         }
3051                         else {
3052                                 // Determine the end if used like '\tiny{...}'
3053                                 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
3054                         }
3055                         // Split on this key if not at start
3056                         int start = interval_.nextNotIgnored(previousStart);
3057                         if (start < actual._tokenstart) {
3058                                 interval_.output(os, actual._tokenstart);
3059                                 interval_.addIntervall(start, actual._tokenstart);
3060                         }
3061                         // discard entry if at end of actual
3062                         nextKeyIdx = process(os, actual);
3063                 }
3064                 break;
3065         }
3066         case KeyInfo::endArguments: {
3067                 // Remove trailing '{}' too
3068                 actual._dataStart += 1;
3069                 actual._dataEnd += 1;
3070                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3071                 nextKeyIdx = getNextKey();
3072                 break;
3073         }
3074         case KeyInfo::noMain:
3075                 // fall through
3076         case KeyInfo::isStandard: {
3077                 if (actual.disabled) {
3078                         removeHead(actual);
3079                         processRegion(actual._dataStart, actual._dataStart+1);
3080                         nextKeyIdx = getNextKey();
3081                 } else {
3082                         // Split on this key if not at datastart of calling entry
3083                         int start = interval_.nextNotIgnored(previousStart);
3084                         if (start < actual._tokenstart) {
3085                                 interval_.output(os, actual._tokenstart);
3086                                 interval_.addIntervall(start, actual._tokenstart);
3087                         }
3088                         // discard entry if at end of actual
3089                         nextKeyIdx = process(os, actual);
3090                 }
3091                 break;
3092         }
3093         case KeyInfo::removeWithArg: {
3094                 nextKeyIdx = getNextKey();
3095                 // Search for end of arguments
3096                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3097                 if (tmpIdx > 0) {
3098                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3099                                 entries_[i].disabled = true;
3100                         }
3101                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3102                 }
3103                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3104                 break;
3105         }
3106         case KeyInfo::doRemove: {
3107                 // Remove the key with all parameters and following spaces
3108                 size_t pos;
3109                 size_t start;
3110                 if (interval_.par[actual._dataEnd-1] == ' ' || interval_.par[actual._dataEnd-1] == '}')
3111                         start = actual._dataEnd;
3112                 else
3113                         start = actual._dataEnd+1;
3114                 for (pos = start; pos < interval_.par.length(); pos++) {
3115                         if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
3116                                 break;
3117                 }
3118                 // Remove also enclosing parentheses [] and {}
3119                 int numpars = 0;
3120                 int spaces = 0;
3121                 while (actual._tokenstart > numpars) {
3122                         if (pos+numpars >= interval_.par.size())
3123                                 break;
3124                         else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
3125                                 numpars++;
3126                         else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
3127                                 numpars++;
3128                         else
3129                                 break;
3130                 }
3131                 if (numpars > 0) {
3132                         if (interval_.par[pos+numpars] == ' ')
3133                                 spaces++;
3134                 }
3135
3136                 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
3137                 nextKeyIdx = getNextKey();
3138                 break;
3139         }
3140         case KeyInfo::isList: {
3141                 // Discard space before _tokenstart
3142                 int count;
3143                 for (count = 0; count < actual._tokenstart; count++) {
3144                         if (interval_.par[actual._tokenstart-count-1] != ' ')
3145                                 break;
3146                 }
3147                 nextKeyIdx = getNextKey();
3148                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3149                 if (tmpIdx > 0) {
3150                         // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
3151                         // with arguments
3152                         // How else can we catch this one?
3153                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3154                                 entries_[i].disabled = true;
3155                         }
3156                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3157                 }
3158                 else if (nextKeyIdx > 0) {
3159                         // Ignore any lang entries inside data region
3160                         for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
3161                                 if (entries_[i].keytype == KeyInfo::isMain)
3162                                         entries_[i].disabled = true;
3163                         }
3164                 }
3165                 if (actual.disabled) {
3166                         interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
3167                 }
3168                 else {
3169                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3170                 }
3171                 if (interval_.par[actual._dataEnd+1] == '[') {
3172                         int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
3173                         if ((interval_.par[actual._dataEnd+2] == '{') &&
3174                                         (interval_.par[posdown-1] == '}')) {
3175                                 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
3176                                 interval_.addIntervall(posdown-1, posdown+1);
3177                         }
3178                         else {
3179                                 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
3180                                 interval_.addIntervall(posdown, posdown+1);
3181                         }
3182                         int blk = interval_.nextNotIgnored(actual._dataEnd+1);
3183                         if (blk > posdown) {
3184                                 // Discard at most 1 space after empty item
3185                                 int count;
3186                                 for (count = 0; count < 1; count++) {
3187                                         if (interval_.par[blk+count] != ' ')
3188                                                 break;
3189                                 }
3190                                 if (count > 0)
3191                                         interval_.addIntervall(blk, blk+count);
3192                         }
3193                 }
3194                 break;
3195         }
3196         case KeyInfo::isSectioning: {
3197                 // Discard spaces before _tokenstart
3198                 int count;
3199                 int val = actual._tokenstart;
3200                 for (count = 0; count < actual._tokenstart;) {
3201                         val = interval_.previousNotIgnored(val-1);
3202                         if (val < 0 || interval_.par[val] != ' ')
3203                                 break;
3204                         else {
3205                                 count = actual._tokenstart - val;
3206                         }
3207                 }
3208                 if (actual.disabled) {
3209                         removeHead(actual, count);
3210                         nextKeyIdx = getNextKey();
3211                 } else {
3212                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3213                         nextKeyIdx = process(os, actual);
3214                 }
3215                 break;
3216         }
3217         case KeyInfo::isMath: {
3218                 // Same as regex, use the content unchanged
3219                 nextKeyIdx = getNextKey();
3220                 break;
3221         }
3222         case KeyInfo::isRegex: {
3223                 // DO NOT SPLIT ON REGEX
3224                 // Do not disable
3225                 nextKeyIdx = getNextKey();
3226                 break;
3227         }
3228         case KeyInfo::isIgnored: {
3229                 // Treat like a character for now
3230                 nextKeyIdx = getNextKey();
3231                 break;
3232         }
3233         case KeyInfo::isMain: {
3234                 if (interval_.par.substr(actual._dataStart, 2) == "% ")
3235                         interval_.addIntervall(actual._dataStart, actual._dataStart+2);
3236                 if (actual._tokenstart > 0) {
3237                         int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
3238                         if ((prev >= 0) && interval_.par[prev] == '%')
3239                                 interval_.addIntervall(prev, prev+1);
3240                 }
3241                 if (actual.disabled) {
3242                         removeHead(actual);
3243                         interval_.langcount++;
3244                         if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
3245                                         (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
3246                                 // Discard also the space before math-equation
3247                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3248                         }
3249                         nextKeyIdx = getNextKey();
3250                         // interval.resetOpenedP(actual._dataStart-1);
3251                 }
3252                 else {
3253                         if (actual._tokenstart < 26) {
3254                                 // for the first (and maybe dummy) language
3255                                 interval_.setForDefaultLang(actual);
3256                         }
3257                         interval_.resetOpenedP(actual._dataStart-1);
3258                 }
3259                 break;
3260         }
3261         case KeyInfo::invalid:
3262         case KeyInfo::headRemove:
3263                 // These two cases cannot happen, already handled
3264                 // fall through
3265         default: {
3266                 // LYXERR(Debug::INFO, "Unhandled keytype");
3267                 nextKeyIdx = getNextKey();
3268                 break;
3269         }
3270         }
3271         return nextKeyIdx;
3272 }
3273
3274 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
3275 {
3276         int end = interval_.nextNotIgnored(actual._dataEnd);
3277         int oldStart = actual._dataStart;
3278         int nextKeyIdx = getNextKey();
3279         while (true) {
3280                 if ((nextKeyIdx < 0) ||
3281                                 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
3282                                 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
3283                         if (oldStart <= end) {
3284                                 processRegion(oldStart, end);
3285                                 oldStart = end+1;
3286                         }
3287                         break;
3288                 }
3289                 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
3290
3291                 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
3292                         (void) dispatch(os, actual._dataStart, nextKey);
3293                         end = nextKey._tokenstart;
3294                         break;
3295                 }
3296                 processRegion(oldStart, nextKey._tokenstart);
3297                 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
3298
3299                 oldStart = nextKey._dataEnd+1;
3300         }
3301         // now nextKey is either invalid or is outside of actual._dataEnd
3302         // output the remaining and discard myself
3303         if (oldStart <= end) {
3304                 processRegion(oldStart, end);
3305         }
3306         if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
3307                 end += 1;
3308                 // This is the normal case.
3309                 // But if using the firstlanguage, the closing may be missing
3310         }
3311         // get minimum of 'end' and  'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
3312         int output_end;
3313         if (actual._dataEnd < end)
3314                 output_end = interval_.nextNotIgnored(actual._dataEnd);
3315         else if (interval_.par.size() > (size_t) end)
3316                 output_end = interval_.nextNotIgnored(end);
3317         else
3318                 output_end = interval_.par.size();
3319         if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
3320                 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
3321         }
3322         // Remove possible empty data
3323         int dstart = interval_.nextNotIgnored(actual._dataStart);
3324         while (interval_.isOpeningPar(dstart) == 1) {
3325                 interval_.addIntervall(dstart, dstart+1);
3326                 int dend = interval_.findclosing(dstart+1, output_end);
3327                 interval_.addIntervall(dend, dend+1);
3328                 dstart = interval_.nextNotIgnored(dstart+1);
3329         }
3330         if (dstart < output_end)
3331                 interval_.output(os, output_end);
3332         if (nextKeyIdx < 0)
3333                 interval_.addIntervall(0, end);
3334         else
3335                 interval_.addIntervall(actual._tokenstart, end);
3336         return nextKeyIdx;
3337 }
3338
3339 string splitOnKnownMacros(string par, bool isPatternString)
3340 {
3341         ostringstream os;
3342         LatexInfo li(par, isPatternString);
3343         // LYXERR(Debug::INFO, "Berfore split: " << par);
3344         KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
3345         DummyKey.head = "";
3346         DummyKey._tokensize = 0;
3347         DummyKey._dataStart = 0;
3348         DummyKey._dataEnd = par.length();
3349         DummyKey.disabled = true;
3350         int firstkeyIdx = li.getFirstKey();
3351         string s;
3352         if (firstkeyIdx >= 0) {
3353                 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
3354                 DummyKey._tokenstart = firstKey._tokenstart;
3355                 int nextkeyIdx;
3356                 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
3357                         // Use dummy firstKey
3358                         firstKey = DummyKey;
3359                         (void) li.setNextKey(firstkeyIdx);
3360                 }
3361                 else {
3362                         if (par.substr(firstKey._dataStart, 2) == "% ")
3363                                 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
3364                 }
3365                 nextkeyIdx = li.process(os, firstKey);
3366                 while (nextkeyIdx >= 0) {
3367                         // Check for a possible gap between the last
3368                         // entry and this one
3369                         int datastart = li.nextNotIgnored(firstKey._dataStart);
3370                         KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
3371                         if ((nextKey._tokenstart > datastart)) {
3372                                 // Handle the gap
3373                                 firstKey._dataStart = datastart;
3374                                 firstKey._dataEnd = par.length();
3375                                 (void) li.setNextKey(nextkeyIdx);
3376                                 // Fake the last opened parenthesis
3377                                 li.setForDefaultLang(firstKey);
3378                                 nextkeyIdx = li.process(os, firstKey);
3379                         }
3380                         else {
3381                                 if (nextKey.keytype != KeyInfo::isMain) {
3382                                         firstKey._dataStart = datastart;
3383                                         firstKey._dataEnd = nextKey._dataEnd+1;
3384                                         (void) li.setNextKey(nextkeyIdx);
3385                                         li.setForDefaultLang(firstKey);
3386                                         nextkeyIdx = li.process(os, firstKey);
3387                                 }
3388                                 else {
3389                                         nextkeyIdx = li.process(os, nextKey);
3390                                 }
3391                         }
3392                 }
3393                 // Handle the remaining
3394                 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
3395                 firstKey._dataEnd = par.length();
3396                 // Check if ! empty
3397                 if ((firstKey._dataStart < firstKey._dataEnd) &&
3398                                 (par[firstKey._dataStart] != '}')) {
3399                         li.setForDefaultLang(firstKey);
3400                         (void) li.process(os, firstKey);
3401                 }
3402                 s = os.str();
3403                 // return string definitelly impossible to match, but should be known
3404         }
3405         else
3406                 s = par;                            /* no known macros found */
3407         // LYXERR(Debug::INFO, "After split: " << s);
3408         return s;
3409 }
3410
3411 /*
3412  * Try to unify the language specs in the latexified text.
3413  * Resulting modified string is set to "", if
3414  * the searched tex does not contain all the features in the search pattern
3415  */
3416 static string correctlanguagesetting(string par, bool isPatternString, bool withformat, lyx::Buffer *pbuf = nullptr)
3417 {
3418         static Features regex_f;
3419         static int missed = 0;
3420         static bool regex_with_format = false;
3421
3422         int parlen = par.length();
3423
3424         while ((parlen > 0) && (par[parlen-1] == '\n')) {
3425                 parlen--;
3426         }
3427 #if 0
3428         if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
3429                 // Happens to be there in case of description or labeling environment
3430                 parlen--;
3431         }
3432 #endif
3433         string result;
3434         if (withformat) {
3435                 // Split the latex input into pieces which
3436                 // can be digested by our search engine
3437                 LYXERR(Debug::FINDVERBOSE, "input: \"" << par << "\"");
3438                 if (isPatternString && (pbuf != nullptr)) { // Check if we should disable/enable test for language
3439                         // We check for polyglossia, because in runparams.flavor we use Flavor::XeTeX
3440                         string doclang = pbuf->params().language->polyglossia();
3441                         static regex langre("\\\\(foreignlanguage)\\{([^\\}]+)\\}");
3442                         smatch sub;
3443                         bool toIgnoreLang = true;
3444                         for (sregex_iterator it(par.begin(), par.end(), langre), end; it != end; ++it) {
3445                                 sub = *it;
3446                                 if (sub.str(2) != doclang) {
3447                                         toIgnoreLang = false;
3448                                         break;
3449                                 }
3450                         }
3451                         setIgnoreFormat("language", toIgnoreLang, false);
3452
3453                 }
3454                 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
3455                 LYXERR(Debug::FINDVERBOSE, "After splitOnKnownMacros:\n\"" << result << "\"");
3456         }
3457         else
3458                 result = par.substr(0, parlen);
3459         if (isPatternString) {
3460                 missed = 0;
3461                 if (withformat) {
3462                         regex_f = identifyFeatures(result);
3463                         string features = "";
3464                         for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3465                                 string a = it->first;
3466                                 regex_with_format = true;
3467                                 features += " " + a;
3468                                 // LYXERR(Debug::INFO, "Identified regex format:" << a);
3469                         }
3470                         LYXERR(Debug::FINDVERBOSE, "Identified Features" << features);
3471
3472                 }
3473         } else if (regex_with_format) {
3474                 Features info = identifyFeatures(result);
3475                 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3476                         string a = it->first;
3477                         bool b = it->second;
3478                         if (b && ! info[a]) {
3479                                 missed++;
3480                                 LYXERR(Debug::FINDVERBOSE, "Missed(" << missed << " " << a <<", srclen = " << parlen );
3481                                 return "";
3482                         }
3483                 }
3484
3485         }
3486         else {
3487                 // LYXERR(Debug::INFO, "No regex formats");
3488         }
3489         return result;
3490 }
3491
3492
3493 // Remove trailing closure of math, macros and environments, so to catch parts of them.
3494 static void identifyClosing(string & t, bool ignoreformat)
3495 {
3496         do {
3497                 LYXERR(Debug::FINDVERBOSE, "identifyClosing(): t now is '" << t << "'");
3498                 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
3499                         continue;
3500                 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
3501                         continue;
3502                 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]+\\*?\\}$", "$1"))
3503                         continue;
3504                 if (! ignoreformat) {
3505                         if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1"))
3506                                 continue;
3507                 }
3508                 break;
3509         } while (true);
3510         return;
3511 }
3512
3513 static int num_replaced = 0;
3514 static bool previous_single_replace = true;
3515
3516 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
3517 {
3518 #if QTSEARCH
3519         if (regexp_str.empty() || regexp2_str.empty()) {
3520                 regexIsValid = false;
3521                 regexError = "Invalid empty regex";
3522                 return;
3523         }
3524         // Handle \w properly
3525         QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3526         if (! opt.casesensitive) {
3527                 popts |= QRegularExpression::CaseInsensitiveOption;
3528         }
3529         regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3530         regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3531         regexError = "";
3532         if (regexp.isValid() && regexp2.isValid()) {
3533                 regexIsValid = true;
3534                 // Check '{', '}' pairs inside the regex
3535                 int balanced = 0;
3536                 int skip = 1;
3537                 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3538                         char c = par_as_string[i];
3539                         if (c == '\\') {
3540                                 skip = 2;
3541                                 continue;
3542                         }
3543                         if (c == '{')
3544                                 balanced++;
3545                         else if (c == '}') {
3546                                 balanced--;
3547                                 if (balanced < 0)
3548                                         break;
3549                         }
3550                         skip = 1;
3551                 }
3552                 if (balanced != 0) {
3553                         regexIsValid = false;
3554                         regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3555                 }
3556         }
3557         else {
3558                 regexIsValid = false;
3559                 if (!regexp.isValid())
3560                         regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3561                 else
3562                         regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3563         }
3564 #else
3565         (void)par_as_string;
3566         if (opt.casesensitive) {
3567                 regexp = regex(regexp_str);
3568                 regexp2 = regex(regexp2_str);
3569         }
3570         else {
3571                 regexp = regex(regexp_str, std::regex_constants::icase);
3572                 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3573         }
3574 #endif
3575 }
3576
3577 static void modifyRegexForMatchWord(string &t)
3578 {
3579         string s("");
3580         regex wordre("(\\\\)*((\\.|\\\\b))");
3581         size_t lastpos = 0;
3582         smatch sub;
3583         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3584                 sub = *it;
3585                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3586                         continue;
3587                 }
3588                 else if (sub.str(2) == "\\\\b")
3589                         return;
3590                 if (lastpos < (size_t) sub.position(2))
3591                         s += t.substr(lastpos, sub.position(2) - lastpos);
3592                 s += "\\S";
3593                 lastpos = sub.position(2) + sub.length(2);
3594         }
3595         if (lastpos == 0) {
3596                 s = "\\b" + t + "\\b";
3597                 t = s;
3598                 return;
3599         }
3600         else if (lastpos < t.length())
3601                 s += t.substr(lastpos, t.length() - lastpos);
3602         t = "\\b" + s + "\\b";
3603 }
3604
3605 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3606         : p_buf(&buf), p_first_buf(&buf), opt(opt)
3607 {
3608         Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3609         docstring const & ds = stringifySearchBuffer(find_buf, opt);
3610         if (ds.empty() ) {
3611                 CreateRegexp(opt, "", "", "");
3612                 return;
3613         }
3614         use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3615         if (opt.replace_all && previous_single_replace) {
3616                 previous_single_replace = false;
3617                 num_replaced = 0;
3618         }
3619         else if (!opt.replace_all) {
3620                 num_replaced = 0;       // count number of replaced strings
3621                 previous_single_replace = true;
3622         }
3623         // When using regexp, braces are hacked already by escape_for_regex()
3624         par_as_string = normalize(ds, opt.ignoreformat);
3625         open_braces = 0;
3626         close_wildcards = 0;
3627
3628         size_t lead_size = 0;
3629         // correct the language settings
3630         par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat, &buf);
3631         if (par_as_string.empty()) {
3632                 CreateRegexp(opt, "", "", "");
3633                 return;
3634         }
3635         opt.matchAtStart = false;
3636         if (!use_regexp) {
3637                 identifyClosing(par_as_string, opt.ignoreformat); // Removes math closings ($, ], ...) at end of string
3638                 if (opt.ignoreformat) {
3639                         lead_size = 0;
3640                 }
3641                 else {
3642                         lead_size = identifyLeading(par_as_string);
3643                 }
3644                 lead_as_string = par_as_string.substr(0, lead_size);
3645                 string lead_as_regex_string = string2regex(lead_as_string);
3646                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3647                 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3648                 /* Handle whole words too in this case
3649                 */
3650                 if (opt.matchword) {
3651                         par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3652                         opt.matchword = false;
3653                 }
3654                 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3655                 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3656                 CreateRegexp(opt, regexp_str, regexp2_str);
3657                 use_regexp = true;
3658                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3659                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3660                 return;
3661         }
3662
3663         if (!opt.ignoreformat) {
3664                 lead_size = identifyLeading(par_as_string);
3665                 LYXERR(Debug::FINDVERBOSE, "Lead_size: " << lead_size);
3666                 lead_as_string = par_as_string.substr(0, lead_size);
3667                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3668         }
3669
3670         // Here we are using regexp
3671         LASSERT(use_regexp, /**/);
3672         {
3673                 string lead_as_regexp;
3674                 if (lead_size > 0) {
3675                         lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3676                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "\\$$", "");
3677                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3678                         par_as_string = par_as_string_nolead;
3679                         LYXERR(Debug::FINDVERBOSE, "lead_as_regexp is '" << lead_as_regexp << "'");
3680                         LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3681                 }
3682                 // LYXERR(Debug::FINDVERBOSE, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3683                 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3684                 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3685                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3686                 ++close_wildcards;
3687                 size_t lng = par_as_string.size();
3688                 if (!opt.ignoreformat) {
3689                         // Remove extra '\}' at end if not part of \{\.\}
3690                         while(lng > 2) {
3691                                 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3692                                         if (lng >= 6) {
3693                                                 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3694                                                         break;
3695                                         }
3696                                         lng -= 2;
3697                                         open_braces++;
3698                                 }
3699                                 else
3700                                         break;
3701                         }
3702                         if (lng < par_as_string.size())
3703                                 par_as_string = par_as_string.substr(0,lng);
3704                 }
3705                 LYXERR(Debug::FINDVERBOSE, "par_as_string after correctRegex is '" << par_as_string << "'");
3706                 if ((lng > 0) && (par_as_string[0] == '^')) {
3707                         par_as_string = par_as_string.substr(1);
3708                         --lng;
3709                         opt.matchAtStart = true;
3710                 }
3711                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3712                 // LYXERR(Debug::FINDVERBOSE, "Open braces: " << open_braces);
3713                 // LYXERR(Debug::FINDVERBOSE, "Replaced text (to be used as regex): " << par_as_string);
3714
3715                 // If entered regexp must match at begin of searched string buffer
3716                 // Kornel: Added parentheses to use $1 for size of the leading string
3717                 string regexp_str;
3718                 string regexp2_str;
3719                 {
3720                         // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3721                         // Unfortunately is '\1', '\2', etc not working for strings with extra format
3722                         // so the convert has no effect in that case
3723                         for (int i = 7; i > 0; --i) {
3724                                 string orig = "\\\\" + std::to_string(i);
3725                                 string dest = "\\" + std::to_string(i+2);
3726                                 while (regex_replace(par_as_string, par_as_string, orig, dest));
3727                         }
3728                         if (opt.matchword) {
3729                                 modifyRegexForMatchWord(par_as_string);
3730                                 opt.matchword = false;
3731                         }
3732                         regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3733                         regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3734                 }
3735                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3736                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3737                 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3738         }
3739 }
3740
3741 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3742 {
3743         MatchResult mres;
3744
3745         mres.searched_size = len;
3746         if (at_begin &&
3747                         (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3748                 return mres;
3749
3750         docstring docstr = stringifyFromForSearch(opt, cur, len);
3751         string str;
3752         str = normalize(docstr, opt.ignoreformat);
3753         if (!opt.ignoreformat) {
3754                 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3755                 // remove closing '}' and '\n' to allow for use of '$' in regex
3756                 size_t lng = str.size();
3757                 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3758                         lng--;
3759                 if (lng != str.size()) {
3760                         str = str.substr(0, lng);
3761                 }
3762                 // Replace occurences of '~' to ' '
3763                 static std::regex specialChars { R"(~)" };
3764                 str = std::regex_replace(str, specialChars,  R"( )" );
3765         }
3766         if (str.empty()) {
3767                 mres.match_len = -1;
3768                 return mres;
3769         }
3770         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
3771
3772         LASSERT(use_regexp, /**/);
3773         {
3774                 // use_regexp always true
3775                 LYXERR(Debug::FINDVERBOSE, "Searching in regexp mode: at_begin=" << at_begin);
3776 #if QTSEARCH
3777                 QString qstr = QString::fromStdString(str);
3778                 QRegularExpression const *p_regexp;
3779                 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3780                 if (at_begin) {
3781                         p_regexp = &regexp;
3782                 } else {
3783                         p_regexp = &regexp2;
3784                 }
3785                 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3786                 if (!match.hasMatch())
3787                         return mres;
3788 #else
3789                 regex const *p_regexp;
3790                 regex_constants::match_flag_type flags;
3791                 if (at_begin) {
3792                         flags = regex_constants::match_continuous;
3793                         p_regexp = &regexp;
3794                 } else {
3795                         flags = regex_constants::match_default;
3796                         p_regexp = &regexp2;
3797                 }
3798                 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3799                 if (re_it == sregex_iterator())
3800                         return mres;
3801                 match_results<string::const_iterator> const & m = *re_it;
3802 #endif
3803                 // Whole found string, including the leading
3804                 // std: m[0].second - m[0].first
3805                 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3806                 //
3807                 // Size of the leading string
3808                 // std: m[1].second - m[1].first
3809                 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3810                 int leadingsize = 0;
3811 #if QTSEARCH
3812                 if (match.lastCapturedIndex() > 0) {
3813                         leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3814                 }
3815
3816 #else
3817                 if (m.size() > 2) {
3818                         leadingsize = m[1].second - m[1].first;
3819                 }
3820 #endif
3821 #if QTSEARCH
3822                 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3823                 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3824                 // because of different number of closing at end of string
3825                 // we have to 'unify' the length of the post-match.
3826                 // Done by ignoring closing parenthesis and linefeeds at string end
3827                 int matchend = match.capturedEnd(0);
3828                 size_t strsize = qstr.size();
3829                 if (!opt.ignoreformat) {
3830                         while (mres.match_len > 1) {
3831                                 QChar c = qstr.at(matchend - 1);
3832                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3833                                         mres.match_len--;
3834                                         matchend--;
3835                                 }
3836                                 else
3837                                         break;
3838                         }
3839                         while (strsize > (size_t) match.capturedEnd(0)) {
3840                                 QChar c = qstr.at(strsize-1);
3841                                 if ((c == '\n') || (c == '}')) {
3842                                         --strsize;
3843                                 }
3844                                 else
3845                                         break;
3846                         }
3847                 }
3848                 // LYXERR0(qstr.toStdString());
3849                 mres.match2end = strsize - matchend;
3850                 mres.pos = match.capturedStart(2);
3851 #else
3852                 mres.match_prefix = m[2].second - m[2].first;
3853                 mres.match_len = m[0].second - m[2].second;
3854                 // ignore closing parenthesis and linefeeds at string end
3855                 size_t strend = m[0].second - m[0].first;
3856                 int matchend = strend;
3857                 size_t strsize = str.size();
3858                 if (!opt.ignoreformat) {
3859                         while (mres.match_len > 1) {
3860                                 char c = str.at(matchend - 1);
3861                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3862                                         mres.match_len--;
3863                                         matchend--;
3864                                 }
3865                                 else
3866                                         break;
3867                         }
3868                         while (strsize > strend) {
3869                                 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3870                                         --strsize;
3871                                 }
3872                                 else
3873                                         break;
3874                         }
3875                 }
3876                 // LYXERR0(str);
3877                 mres.match2end = strsize - matchend;
3878                 mres.pos = m[2].first - m[0].first;;
3879 #endif
3880                 if (mres.match2end < 0)
3881                         mres.match_len = 0;
3882                 mres.leadsize = leadingsize;
3883 #if QTSEARCH
3884                 if (mres.match_len > 0) {
3885                         string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3886                         mres.result.push_back(a0);
3887                         for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3888                                 mres.result.push_back(match.captured(i).toStdString());
3889                         }
3890                 }
3891 #else
3892                 if (mres.match_len > 0) {
3893                         string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3894                         mres.result.push_back(a0);
3895                         for (size_t i = 3; i < m.size(); i++) {
3896                                 mres.result.push_back(m[i]);
3897                         }
3898                 }
3899 #endif
3900                 return mres;
3901         }
3902 }
3903
3904
3905 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3906 {
3907         MatchResult mres = findAux(cur, len, at_begin);
3908         int res = mres.match_len;
3909         LYXERR(Debug::FINDVERBOSE,
3910                "res=" << res << ", at_begin=" << at_begin
3911                << ", matchAtStart=" << opt.matchAtStart
3912                << ", inTexted=" << cur.inTexted());
3913         if (opt.matchAtStart) {
3914                 if (cur.pos() != 0)
3915                         mres.match_len = 0;
3916                 else if (mres.match_prefix > 0)
3917                         mres.match_len = 0;
3918                 return mres;
3919         }
3920         else
3921                 return mres;
3922 }
3923
3924 #if 0
3925 static bool simple_replace(string &t, string from, string to)
3926 {
3927         regex repl("(\\\\)*(" + from + ")");
3928         string s("");
3929         size_t lastpos = 0;
3930         smatch sub;
3931         for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3932                 sub = *it;
3933                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3934                         continue;
3935                 if (lastpos < (size_t) sub.position(2))
3936                         s += t.substr(lastpos, sub.position(2) - lastpos);
3937                 s += to;
3938                 lastpos = sub.position(2) + sub.length(2);
3939         }
3940         if (lastpos == 0)
3941                 return false;
3942         else if (lastpos < t.length())
3943                 s += t.substr(lastpos, t.length() - lastpos);
3944         t = s;
3945         return true;
3946 }
3947 #endif
3948
3949 string MatchStringAdv::normalize(docstring const & s, bool ignore_format) const
3950 {
3951         string t;
3952         t = lyx::to_utf8(s);
3953         // Remove \n at begin
3954         while (!t.empty() && t[0] == '\n')
3955                 t = t.substr(1);
3956         // Remove \n* at end
3957         while (!t.empty() && t[t.size() - 1] == '\n') {
3958                 t = t.substr(0, t.size() - 1);
3959         }
3960         size_t pos;
3961         // Handle all other '\n'
3962         while ((pos = t.find("\n")) != string::npos) {
3963                 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3964                         // Handle '\\\n'
3965                         if (isAlnumASCII(t[pos+1])) {
3966                                 t.replace(pos-2, 3, " ");
3967                         }
3968                         else {
3969                                 t.replace(pos-2, 3, "");
3970                         }
3971                 }
3972                 else {
3973                         if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3974                                 // '\n' adjacent to non-alpha-numerics, discard
3975                                 t.replace(pos, 1, "");
3976                         }
3977                         else {
3978                                 // Replace all other \n with spaces
3979                                 t.replace(pos, 1, " ");
3980                         }
3981                         if (!ignore_format) {
3982                                 size_t count = 0;
3983                                 while ((pos > count + 1) && (t[pos - 1 -count] == '%')) {
3984                                         count++;
3985                                 }
3986                                 if (count > 0) {
3987                                         t.replace(pos - count, count, "");
3988                                 }
3989                         }
3990                 }
3991         }
3992         // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3993         // Kornel: Added textsl, textsf, textit, texttt and noun
3994         // + allow to seach for colored text too
3995         LYXERR(Debug::FINDVERBOSE, "Removing stale empty macros from: " << t);
3996         while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3997                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3998         while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3999                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\section{}, \\part{}, \\paragraph{} macros from: " << t);
4000         while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
4001
4002         return t;
4003 }
4004
4005
4006 docstring stringifyFromCursor(DocIterator const & cur, int len)
4007 {
4008         LYXERR(Debug::FINDVERBOSE, "Stringifying with len=" << len << " from cursor at pos: " << cur);
4009         if (cur.inTexted()) {
4010                 Paragraph const & par = cur.paragraph();
4011                 // TODO what about searching beyond/across paragraph breaks ?
4012                 // TODO Try adding a AS_STR_INSERTS as last arg
4013                 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
4014                                         int(par.size()) : cur.pos() + len;
4015                 // OutputParams runparams(&cur.buffer()->params().encoding());
4016                 OutputParams runparams(encodings.fromLyXName("utf8"));
4017                 runparams.nice = true;
4018                 setFindParams(runparams);
4019                 int option = AS_STR_INSETS | AS_STR_PLAINTEXT;
4020                 if (ignoreFormats.getDeleted()) {
4021                         option |= AS_STR_SKIPDELETE;
4022                         runparams.find_set_feature(OutputParams::SearchWithoutDeleted);
4023                 }
4024                 else {
4025                         runparams.find_set_feature(OutputParams::SearchWithDeleted);
4026                 }
4027                 if (ignoreFormats.getNonContent()) {
4028                         runparams.find_add_feature(OutputParams::SearchNonOutput);
4029                 }
4030                 LYXERR(Debug::FINDVERBOSE, "Stringifying with cur: "
4031                        << cur << ", from pos: " << cur.pos() << ", end: " << end);
4032                 docstring res = from_utf8(latexNamesToUtf8(par.asString(cur.pos(), end,
4033                                                                         option,
4034                                                                         &runparams), false));
4035                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Stringified text from pos(" << cur.pos() << ") len(" << len << "): " << res);
4036                 return res;
4037         } else if (cur.inMathed()) {
4038                 CursorSlice cs = cur.top();
4039                 MathData md = cs.cell();
4040                 MathData::const_iterator it_end =
4041                                 (( len == -1 || cs.pos() + len > int(md.size()))
4042                                  ? md.end()
4043                                  : md.begin() + cs.pos() + len );
4044                 MathData md2;
4045                 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
4046                         md2.push_back(*it);
4047                 docstring res = from_utf8(latexNamesToUtf8(asString(md2), false));
4048                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Stringified math from pos(" << cur.pos() << ") len(" << len << "): " << res);
4049                 return res;
4050         }
4051         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
4052         return docstring();
4053 }
4054
4055 /** Computes the LaTeX export of buf starting from cur and ending len positions
4056  * after cur, if len is positive, or at the paragraph or innermost inset end
4057  * if len is -1.
4058  */
4059 docstring latexifyFromCursor(DocIterator const & cur, int len)
4060 {
4061         /*
4062         LYXERR(Debug::FINDVERBOSE, "Latexifying with len=" << len << " from cursor at pos: " << cur);
4063         LYXERR(Debug::FINDVERBOSE, "  with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
4064                << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
4065         */
4066         Buffer const & buf = *cur.buffer();
4067
4068         odocstringstream ods;
4069         otexstream os(ods);
4070         //OutputParams runparams(&buf.params().encoding());
4071         OutputParams runparams(encodings.fromLyXName("utf8"));
4072         runparams.nice = false;
4073         setFindParams(runparams);
4074         if (ignoreFormats.getDeleted()) {
4075                 runparams.find_set_feature(OutputParams::SearchWithoutDeleted);
4076         }
4077         else {
4078                 runparams.find_set_feature(OutputParams::SearchWithDeleted);
4079         }
4080         if (ignoreFormats.getNonContent()) {
4081                 runparams.find_add_feature(OutputParams::SearchNonOutput);
4082         }
4083
4084         if (cur.inTexted()) {
4085                 // @TODO what about searching beyond/across paragraph breaks ?
4086                 pos_type endpos = cur.paragraph().size();
4087                 if (len != -1 && endpos > cur.pos() + len)
4088                         endpos = cur.pos() + len;
4089                 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
4090                           string(), cur.pos(), endpos, true);
4091                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Latexified text from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4092                 return(ods.str());
4093         } else if (cur.inMathed()) {
4094                 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
4095                 for (int s = cur.depth() - 1; s >= 0; --s) {
4096                         CursorSlice const & cs = cur[s];
4097                         if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
4098                                 TeXMathStream ws(os);
4099                                 cs.asInsetMath()->asHullInset()->header_write(ws);
4100                                 break;
4101                         }
4102                 }
4103
4104                 CursorSlice const & cs = cur.top();
4105                 MathData md = cs.cell();
4106                 MathData::const_iterator it_end =
4107                                 ((len == -1 || cs.pos() + len > int(md.size()))
4108                                  ? md.end()
4109                                  : md.begin() + cs.pos() + len);
4110                 MathData md2;
4111                 for (MathData::const_iterator it = md.begin() + cs.pos();
4112                      it != it_end; ++it)
4113                         md2.push_back(*it);
4114
4115                 ods << asString(md2);
4116                 // Retrieve the math environment type, and add '$' or '$]'
4117                 // or others (\end{equation}) accordingly
4118                 for (int s = cur.depth() - 1; s >= 0; --s) {
4119                         CursorSlice const & cs2 = cur[s];
4120                         InsetMath * inset = cs2.asInsetMath();
4121                         if (inset && inset->asHullInset()) {
4122                                 TeXMathStream ws(os);
4123                                 inset->asHullInset()->footer_write(ws);
4124                                 break;
4125                         }
4126                 }
4127                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Latexified math from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4128         } else {
4129                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
4130         }
4131         return ods.str();
4132 }
4133
4134 #if defined(ResultsDebug)
4135 // Debugging output
4136 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
4137 {
4138         LYXERR0( "from:\t\t\t" << from);
4139         string status;
4140         if (mres.pos_len > 0) {
4141                 // Set in finalize
4142                 status = "FINALSEARCH";
4143         }
4144         else {
4145                 if (mres.match_len > 0) {
4146                         if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
4147                                 status = "Good Match";
4148                         else
4149                                 status = "Matched in";
4150                 }
4151                 else
4152                         status = "MissedSearch";
4153         }
4154
4155         LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
4156         if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
4157                 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
4158         if ((mres.pos > 0) || (mres.match_prefix > 0))
4159                 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
4160         for (size_t i = 0; i < mres.result.size(); i++)
4161                 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
4162 }
4163 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
4164 #else
4165 #define displayMres(s, txt, cur)
4166 #endif
4167
4168 /** Finalize an advanced find operation, advancing the cursor to the innermost
4169  ** position that matches, plus computing the length of the matching text to
4170  ** be selected
4171  ** Return the cur.pos() difference between start and end of found match
4172  **/
4173 MatchResult findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
4174 {
4175         // Search the foremost position that matches (avoids find of entire math
4176         // inset when match at start of it)
4177         DocIterator old_cur(cur.buffer());
4178         MatchResult mres;
4179         static MatchResult fail = MatchResult();
4180         MatchResult max_match;
4181         // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
4182         // Happens with e.g. hyperlinks
4183         // either one sees "http://www.bla.bla" or nothing
4184         // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
4185         // and although we search for only 3 chars, we find the whole hyperlink inset
4186         bool at_begin = (expected.match_prefix == 0);
4187         if (!match.opt.forward && match.opt.ignoreformat) {
4188                 if (expected.pos > 0)
4189                         return fail;
4190         }
4191         LASSERT(at_begin, /**/);
4192         if (expected.match_len > 0 && at_begin) {
4193                 // Search for deepest match
4194                 old_cur = cur;
4195                 max_match = expected;
4196                 do {
4197                         size_t d = cur.depth();
4198                         cur.forwardPos();
4199                         if (!cur)
4200                                 break;
4201                         if (cur.depth() < d)
4202                                 break;
4203                         if (cur.depth() == d)
4204                                 break;
4205                         size_t lastd = d;
4206                         while (cur && cur.depth() > lastd) {
4207                                 lastd = cur.depth();
4208                                 mres = match(cur, -1, at_begin);
4209                                 displayMres(mres, "Checking innermost", cur);
4210                                 if (mres.match_len > 0)
4211                                         break;
4212                                 // maybe deeper?
4213                                 cur.forwardPos();
4214                         }
4215                         if (mres.match_len < expected.match_len)
4216                                 break;
4217                         max_match = mres;
4218                         old_cur = cur;;
4219                 } while(1);
4220                 cur = old_cur;
4221         }
4222         else {
4223                 // (expected.match_len <= 0)
4224                 mres = match(cur);      /* match valid only if not searching whole words */
4225                 displayMres(mres, "Start with negative match", cur);
4226                 max_match = mres;
4227         }
4228         if (max_match.match_len <= 0) return fail;
4229         LYXERR(Debug::FINDVERBOSE, "Ok");
4230
4231         // Compute the match length
4232         int len = 1;
4233         if (cur.pos() + len > cur.lastpos())
4234                 return fail;
4235
4236         LASSERT(match.use_regexp, /**/);
4237         {
4238                 int minl = 1;
4239                 int maxl = cur.lastpos() - cur.pos();
4240                 // Greedy behaviour while matching regexps
4241                 while (maxl > minl) {
4242                         MatchResult mres2;
4243                         mres2 = match(cur, len, at_begin);
4244                         displayMres(mres2, "Finalize loop", cur);
4245                         int actual_match_len = mres2.match_len;
4246                         if (actual_match_len >= max_match.match_len) {
4247                                 // actual_match_len > max_match _can_ happen,
4248                                 // if the search area splits
4249                                 // some following word so that the regex
4250                                 // (e.g. 'r.*r\b' matches 'r' from the middle of the
4251                                 // splitted word)
4252                                 // This means, the len value is too big
4253                                 actual_match_len = max_match.match_len;
4254                                 max_match = mres2;
4255                                 max_match.match_len = actual_match_len;
4256                                 maxl = len;
4257                                 if (maxl - minl < 4)
4258                                         len = (maxl + minl)/2;
4259                                 else
4260                                         len = minl + (maxl - minl + 3)/4;
4261                         }
4262                         else {
4263                                 // (actual_match_len < max_match.match_len)
4264                                 minl = len + 1;
4265                                 len = (maxl + minl)/2;
4266                         }
4267                 }
4268                 len = minl;
4269                 old_cur = cur;
4270                 // Search for real start of matched characters
4271                 while (len > 1) {
4272                         MatchResult actual_match;
4273                         do {
4274                                 cur.forwardPos();
4275                         } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
4276                         if (cur.depth() < old_cur.depth()) {
4277                                 // Outer inset?
4278                                 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
4279                                 break;
4280                         }
4281                         if (cur.pos() != old_cur.pos()) {
4282                                 // OK, forwarded 1 pos in actual inset
4283                                 actual_match = match(cur, len-1, at_begin);
4284                                 if (actual_match.match_len == max_match.match_len) {
4285                                         // Ha, got it! The shorter selection has the same match length
4286                                         len--;
4287                                         old_cur = cur;
4288                                         max_match = actual_match;
4289                                 }
4290                                 else {
4291                                         // OK, the shorter selection matches less chars, revert to previous value
4292                                         cur = old_cur;
4293                                         break;
4294                                 }
4295                         }
4296                         else {
4297                                 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
4298                                 actual_match = match(cur, len, at_begin);
4299                                 if (actual_match.match_len == max_match.match_len) {
4300                                         old_cur = cur;
4301                                         max_match = actual_match;
4302                                 }
4303                         }
4304                 }
4305                 if (len == 0)
4306                         return fail;
4307                 else {
4308                         max_match.pos_len = len;
4309                         displayMres(max_match, "SEARCH RESULT", cur)
4310                                         return max_match;
4311                 }
4312         }
4313 }
4314
4315 /// Finds forward
4316 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
4317 {
4318         if (!cur)
4319                 return 0;
4320         bool repeat = false;
4321         DocIterator orig_cur;   // to be used if repeat not successful
4322         MatchResult orig_mres;
4323         while (!theApp()->longOperationCancelled() && cur) {
4324                 //(void) findAdvForwardInnermost(cur);
4325                 LYXERR(Debug::FINDVERBOSE, "findForwardAdv() cur: " << cur);
4326                 MatchResult mres = match(cur, -1, false);
4327                 string msg = "Starting";
4328                 if (repeat)
4329                         msg = "Repeated";
4330                 displayMres(mres, msg + " findForwardAdv", cur)
4331                                 int match_len = mres.match_len;
4332                 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
4333                         LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
4334                         match_len = 0;
4335                 }
4336                 if (match_len <= 0) {
4337                         // This should exit nested insets, if any, or otherwise undefine the currsor.
4338                         cur.pos() = cur.lastpos();
4339                         LYXERR(Debug::FINDVERBOSE, "Advancing pos: cur=" << cur);
4340                         cur.forwardPos();
4341                 }
4342                 else {  // match_len > 0
4343                         // Try to find the begin of searched string
4344                         int increment;
4345                         int firstInvalid = cur.lastpos() - cur.pos();
4346                         {
4347                                 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
4348                                 int incrcur = (firstInvalid + 1 )*3/4;
4349                                 if (incrcur < incrmatch)
4350                                         increment = incrcur;
4351                                 else
4352                                         increment = incrmatch;
4353                                 if (increment < 1)
4354                                         increment = 1;
4355                         }
4356                         LYXERR(Debug::FINDVERBOSE, "Set increment to " << increment);
4357                         while (increment > 0) {
4358                                 DocIterator old_cur = cur;
4359                                 if (cur.pos() + increment >= cur.lastpos()) {
4360                                         increment /= 2;
4361                                         continue;
4362                                 }
4363                                 cur.pos() = cur.pos() + increment;
4364                                 MatchResult mres2 = match(cur, -1, false);
4365                                 displayMres(mres2, "findForwardAdv loop", cur)
4366                                                 switch (interpretMatch(mres, mres2)) {
4367                                         case MatchResult::newIsTooFar:
4368                                                 // behind the expected match
4369                                                 firstInvalid = increment;
4370                                                 cur = old_cur;
4371                                                 increment /= 2;
4372                                                 break;
4373                                         case MatchResult::newIsBetter:
4374                                                 // not reached yet, but cur.pos()+increment is bettert
4375                                                 mres = mres2;
4376                                                 firstInvalid -= increment;
4377                                                 if (increment > firstInvalid*3/4)
4378                                                         increment = firstInvalid*3/4;
4379                                                 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
4380                                                         if (increment >= mres2.match_prefix)
4381                                                                 increment = (mres2.match_prefix+1)*3/4;
4382                                                 }
4383                                                 break;
4384                                         default:
4385                                                 // Todo@
4386                                                 // Handle not like MatchResult::newIsTooFar
4387                                                 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
4388                                                 firstInvalid--;
4389                                                 increment = increment*3/4;
4390                                                 cur = old_cur;
4391                                                 break;
4392                                 }
4393                         }
4394                         if (mres.match_len > 0) {
4395                                 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
4396                                         // The match seems to indicate some deeper level
4397                                         repeat = true;
4398                                         orig_cur = cur;
4399                                         orig_mres = mres;
4400                                         cur.forwardPos();
4401                                         continue;
4402                                 }
4403                         }
4404                         else if (repeat) {
4405                                 // should never be reached.
4406                                 cur = orig_cur;
4407                                 mres = orig_mres;
4408                         }
4409                         // LYXERR0("Leaving first loop");
4410                         LYXERR(Debug::FINDVERBOSE, "Finalizing 1");
4411                         MatchResult found_match = findAdvFinalize(cur, match, mres);
4412                         if (found_match.match_len > 0) {
4413                                 match.FillResults(found_match);
4414                                 return found_match.pos_len;
4415                         }
4416                         else {
4417                                 // try next possible match
4418                                 cur.forwardPos();
4419                                 repeat = false;
4420                                 continue;
4421                         }
4422                 }
4423         }
4424         return 0;
4425 }
4426
4427
4428 /// Find the most backward consecutive match within same paragraph while searching backwards.
4429 MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, MatchResult &expected)
4430 {
4431         DocIterator cur_begin = cur;
4432         cur_begin.pos() = 0;
4433         DocIterator tmp_cur = cur;
4434         MatchResult mr = findAdvFinalize(tmp_cur, match, expected);
4435         Inset & inset = cur.inset();
4436         for (; cur != cur_begin; cur.backwardPos()) {
4437                 LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): cur=" << cur);
4438                 DocIterator new_cur = cur;
4439                 new_cur.backwardPos();
4440                 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
4441                         break;
4442                 MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
4443                 if (new_mr.match_len == mr.match_len)
4444                         break;
4445                 mr = new_mr;
4446         }
4447         LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): exiting with cur=" << cur);
4448         return mr;
4449 }
4450
4451
4452 /// Finds backwards
4453 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
4454 {
4455         if (! cur)
4456                 return 0;
4457         // Backup of original position
4458         DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4459         if (cur == cur_begin)
4460                 return 0;
4461         cur.backwardPos();
4462         DocIterator cur_orig(cur);
4463         bool pit_changed = false;
4464         do {
4465                 cur.pos() = 0;
4466                 MatchResult found_match = match(cur, -1, false);
4467
4468                 if (found_match.match_len > 0) {
4469                         if (pit_changed)
4470                                 cur.pos() = cur.lastpos();
4471                         else
4472                                 cur.pos() = cur_orig.pos();
4473                         LYXERR(Debug::FINDVERBOSE, "findBackAdv2: cur: " << cur);
4474                         DocIterator cur_prev_iter;
4475                         do {
4476                                 found_match = match(cur);
4477                                 LYXERR(Debug::FINDVERBOSE, "findBackAdv3: found_match="
4478                                        << (found_match.match_len > 0) << ", cur: " << cur);
4479                                 if (found_match.match_len > 0) {
4480                                         MatchResult found_mr = findMostBackwards(cur, match, found_match);
4481                                         if (found_mr.pos_len > 0) {
4482                                                 match.FillResults(found_mr);
4483                                                 return found_mr.pos_len;
4484                                         }
4485                                 }
4486
4487                                 // Stop if begin of document reached
4488                                 if (cur == cur_begin)
4489                                         break;
4490                                 cur_prev_iter = cur;
4491                                 cur.backwardPos();
4492                         } while (true);
4493                 }
4494                 if (cur == cur_begin)
4495                         break;
4496                 if (cur.pit() > 0)
4497                         --cur.pit();
4498                 else
4499                         cur.backwardPos();
4500                 pit_changed = true;
4501         } while (!theApp()->longOperationCancelled());
4502         return 0;
4503 }
4504
4505
4506 } // namespace
4507
4508
4509 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4510                                  DocIterator const & cur, int len)
4511 {
4512         if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4513                 return docstring();
4514         if (!opt.ignoreformat)
4515                 return latexifyFromCursor(cur, len);
4516         else
4517                 return stringifyFromCursor(cur, len);
4518 }
4519
4520
4521 FindAndReplaceOptions::FindAndReplaceOptions(
4522                 docstring const & _find_buf_name, bool _casesensitive,
4523                 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4524                 docstring const & _repl_buf_name, bool _keep_case,
4525                 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4526         : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4527           forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4528           repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4529 {
4530 }
4531
4532
4533 namespace {
4534
4535
4536 /** Check if 'len' letters following cursor are all non-lowercase */
4537 static bool allNonLowercase(Cursor const & cur, int len)
4538 {
4539         pos_type beg_pos = cur.selectionBegin().pos();
4540         pos_type end_pos = cur.selectionBegin().pos() + len;
4541         if (len > cur.lastpos() + 1 - beg_pos) {
4542                 LYXERR(Debug::FINDVERBOSE, "This should not happen, more debug needed");
4543                 len = cur.lastpos() + 1 - beg_pos;
4544                 end_pos = beg_pos + len;
4545         }
4546         for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4547                 if (isLowerCase(cur.paragraph().getChar(pos)))
4548                         return false;
4549         return true;
4550 }
4551
4552
4553 /** Check if first letter is upper case and second one is lower case */
4554 static bool firstUppercase(Cursor const & cur)
4555 {
4556         char_type ch1, ch2;
4557         pos_type pos = cur.selectionBegin().pos();
4558         if (pos >= cur.lastpos() - 1) {
4559                 LYXERR(Debug::FINDVERBOSE, "No upper-case at cur: " << cur);
4560                 return false;
4561         }
4562         ch1 = cur.paragraph().getChar(pos);
4563         ch2 = cur.paragraph().getChar(pos + 1);
4564         bool result = isUpperCase(ch1) && isLowerCase(ch2);
4565         LYXERR(Debug::FINDVERBOSE, "firstUppercase(): "
4566                << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4567                << ch2 << "(" << char(ch2) << ")"
4568                << ", result=" << result << ", cur=" << cur);
4569         return result;
4570 }
4571
4572
4573 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4574  **
4575  ** \fixme What to do with possible further paragraphs in replace buffer ?
4576  **/
4577 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4578 {
4579         ParagraphList::iterator pit = buffer.paragraphs().begin();
4580         LASSERT(!pit->empty(), /**/);
4581         pos_type right = pos_type(1);
4582         pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4583         right = pit->size();
4584         pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4585 }
4586 } // namespace
4587
4588 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4589 {
4590         // Should replace the string "$" + std::to_string(matchnum) with replacement
4591         // if the char '$' is not prefixed with odd number of char '\\'
4592         static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4593         string s;
4594         size_t lastpos = 0;
4595         smatch sub;
4596         for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4597                 sub = *it;
4598                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4599                         continue;
4600                 int num = stoi(sub.str(3), nullptr, 10);
4601                 if (num >= maxmatchnum)
4602                         continue;
4603                 if (lastpos < (size_t) sub.position(2))
4604                         s += t.substr(lastpos, sub.position(2) - lastpos);
4605                 s += replacements[num];
4606                 lastpos = sub.position(2) + sub.length(2);
4607         }
4608         if (lastpos == 0)
4609                 return false;
4610         else if (lastpos < t.length())
4611                 s += t.substr(lastpos, t.length() - lastpos);
4612         t = s;
4613         return true;
4614 }
4615
4616 ///
4617 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4618 {
4619         Cursor & cur = bv->cursor();
4620         if (opt.repl_buf_name.empty()
4621                         || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4622                         || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4623                 return 0;
4624
4625         DocIterator sel_beg = cur.selectionBegin();
4626         DocIterator sel_end = cur.selectionEnd();
4627         if (&sel_beg.inset() != &sel_end.inset()
4628                         || sel_beg.pit() != sel_end.pit()
4629                         || sel_beg.idx() != sel_end.idx())
4630                 return 0;
4631         int sel_len = sel_end.pos() - sel_beg.pos();
4632         LYXERR(Debug::FINDVERBOSE, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4633                << ", sel_len: " << sel_len << endl);
4634         if (sel_len == 0)
4635                 return 0;
4636         LASSERT(sel_len > 0, return 0);
4637
4638         if (!matchAdv(sel_beg, sel_len).match_len)
4639                 return 0;
4640
4641         // Build a copy of the replace buffer, adapted to the KeepCase option
4642         Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4643         ostringstream oss;
4644         repl_buffer_orig.write(oss);
4645         string lyx = oss.str();
4646         if (matchAdv.valid_matches > 0)
4647                 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4648         Buffer repl_buffer(string(), false);
4649         repl_buffer.setInternal(true);
4650         repl_buffer.setUnnamed(true);
4651         LASSERT(repl_buffer.readString(lyx), return 0);
4652         if (opt.keep_case && sel_len >= 2) {
4653                 LYXERR(Debug::FINDVERBOSE, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4654                 if (cur.inTexted()) {
4655                         if (firstUppercase(cur))
4656                                 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4657                         else if (allNonLowercase(cur, sel_len))
4658                                 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4659                 }
4660         }
4661         cap::cutSelection(cur, false);
4662         if (cur.inTexted()) {
4663                 repl_buffer.changeLanguage(
4664                                         repl_buffer.language(),
4665                                         cur.getFont().language());
4666                 LYXERR(Debug::FINDVERBOSE, "Replacing by pasteParagraphList()ing repl_buffer");
4667                 LYXERR(Debug::FINDVERBOSE, "Before pasteParagraphList() cur=" << cur << endl);
4668                 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4669                                         repl_buffer.params().documentClassPtr(),
4670                                         repl_buffer.params().authors(),
4671                                         bv->buffer().errorList("Paste"));
4672                 LYXERR(Debug::FINDVERBOSE, "After pasteParagraphList() cur=" << cur << endl);
4673                 sel_len = repl_buffer.paragraphs().begin()->size();
4674         } else if (cur.inMathed()) {
4675                 odocstringstream ods;
4676                 otexstream os(ods);
4677                 // OutputParams runparams(&repl_buffer.params().encoding());
4678                 OutputParams runparams(encodings.fromLyXName("utf8"));
4679                 runparams.nice = false;
4680                 setFindParams(runparams);
4681                 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams, string(), -1, -1, true);
4682                 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4683                 docstring repl_latex = ods.str();
4684                 LYXERR(Debug::FINDVERBOSE, "Latexified replace_buffer: '" << repl_latex << "'");
4685                 string s;
4686                 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4687                 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4688                 repl_latex = from_utf8(s);
4689                 LYXERR(Debug::FINDVERBOSE, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4690                 MathData ar(cur.buffer());
4691                 asArray(repl_latex, ar, Parse::NORMAL);
4692                 cur.insert(ar);
4693                 sel_len = ar.size();
4694                 LYXERR(Debug::FINDVERBOSE, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4695         }
4696         if (cur.pos() >= sel_len)
4697                 cur.pos() -= sel_len;
4698         else
4699                 cur.pos() = 0;
4700         LYXERR(Debug::FINDVERBOSE, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4701         bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4702         bv->processUpdateFlags(Update::Force);
4703         return 1;
4704 }
4705
4706
4707 /// Perform a FindAdv operation.
4708 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4709 {
4710         DocIterator cur;
4711         int pos_len = 0;
4712
4713         // e.g., when invoking word-findadv from mini-buffer wither with
4714         //       wrong options syntax or before ever opening advanced F&R pane
4715         if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4716                 return false;
4717
4718         try {
4719                 MatchStringAdv matchAdv(bv->buffer(), opt);
4720 #if QTSEARCH
4721                 if (!matchAdv.regexIsValid) {
4722                         bv->message(lyx::from_utf8(matchAdv.regexError));
4723                         return(false);
4724                 }
4725 #endif
4726                 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4727                 if (length > 0)
4728                         bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4729                 num_replaced += findAdvReplace(bv, opt, matchAdv);
4730                 cur = bv->cursor();
4731                 if (opt.forward)
4732                         pos_len = findForwardAdv(cur, matchAdv);
4733                 else
4734                         pos_len = findBackwardsAdv(cur, matchAdv);
4735         } catch (exception & ex) {
4736                 bv->message(from_utf8(ex.what()));
4737                 return false;
4738         }
4739
4740         if (pos_len == 0) {
4741                 if (num_replaced > 0) {
4742                         switch (num_replaced)
4743                         {
4744                         case 1:
4745                                 bv->message(_("One match has been replaced."));
4746                                 break;
4747                         case 2:
4748                                 bv->message(_("Two matches have been replaced."));
4749                                 break;
4750                         default:
4751                                 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4752                                 break;
4753                         }
4754                         num_replaced = 0;
4755                 }
4756                 else {
4757                         bv->message(_("Match not found."));
4758                 }
4759                 return false;
4760         }
4761
4762         if (num_replaced > 0)
4763                 bv->message(_("Match has been replaced."));
4764         else
4765                 bv->message(_("Match found."));
4766
4767         if (cur.pos() + pos_len > cur.lastpos()) {
4768                 // Prevent crash in bv->putSelectionAt()
4769                 // Should never happen, maybe LASSERT() here?
4770                 pos_len = cur.lastpos() - cur.pos();
4771         }
4772         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4773         bv->putSelectionAt(cur, pos_len, !opt.forward);
4774
4775         return true;
4776 }
4777
4778
4779 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4780 {
4781         os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4782            << opt.casesensitive << ' '
4783            << opt.matchword << ' '
4784            << opt.forward << ' '
4785            << opt.expandmacros << ' '
4786            << opt.ignoreformat << ' '
4787            << opt.replace_all << ' '
4788            << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4789            << opt.keep_case << ' '
4790            << int(opt.scope) << ' '
4791            << int(opt.restr);
4792
4793         LYXERR(Debug::FINDVERBOSE, "built: " << os.str());
4794
4795         return os;
4796 }
4797
4798
4799 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4800 {
4801         // LYXERR(Debug::FINDVERBOSE, "parsing");
4802         string s;
4803         string line;
4804         getline(is, line);
4805         while (line != "EOSS") {
4806                 if (! s.empty())
4807                         s = s + "\n";
4808                 s = s + line;
4809                 if (is.eof())   // Tolerate malformed request
4810                         break;
4811                 getline(is, line);
4812         }
4813         // LYXERR(Debug::FINDVERBOSE, "file_buf_name: '" << s << "'");
4814         opt.find_buf_name = from_utf8(s);
4815         is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4816         is.get();       // Waste space before replace string
4817         s = "";
4818         getline(is, line);
4819         while (line != "EOSS") {
4820                 if (! s.empty())
4821                         s = s + "\n";
4822                 s = s + line;
4823                 if (is.eof())   // Tolerate malformed request
4824                         break;
4825                 getline(is, line);
4826         }
4827         // LYXERR(Debug::FINDVERBOSE, "repl_buf_name: '" << s << "'");
4828         opt.repl_buf_name = from_utf8(s);
4829         is >> opt.keep_case;
4830         int i;
4831         is >> i;
4832         opt.scope = FindAndReplaceOptions::SearchScope(i);
4833         is >> i;
4834         opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4835
4836         /*
4837         LYXERR(Debug::FINDVERBOSE, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4838                << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4839                << opt.scope << ' ' << opt.restr);
4840         */
4841         return is;
4842 }
4843
4844 } // namespace lyx