src/lyxfind.cpp

   1 /**
   2  * \file lyxfind.cpp
   3  * This file is part of LyX, the document processor.
   4  * License details can be found in the file COPYING.
   5  *
   6  * \author Lars Gullik Bjønnes
   7  * \author John Levon
   8  * \author Jürgen Vigna
   9  * \author Alfredo Braunstein
  10  * \author Tommaso Cucinotta
  11  * \author Kornel Benko
  12  *
  13  * Full author contact details are available in file CREDITS.
  14  */
  15
  16 #include <config.h>
  17
  18 #include "lyxfind.h"
  19
  20 #include "Buffer.h"
  21 #include "BufferList.h"
  22 #include "BufferParams.h"
  23 #include "BufferView.h"
  24 #include "Changes.h"
  25 #include "Cursor.h"
  26 #include "CutAndPaste.h"
  27 #include "FuncRequest.h"
  28 #include "LyX.h"
  29 #include "output_latex.h"
  30 #include "OutputParams.h"
  31 #include "Paragraph.h"
  32 #include "Text.h"
  33 #include "Encoding.h"
  34 #include "Language.h"
  35
  36 #include "frontends/Application.h"
  37 #include "frontends/alert.h"
  38
  39 #include "mathed/InsetMath.h"
  40 #include "mathed/InsetMathHull.h"
  41 #include "mathed/MathData.h"
  42 #include "mathed/MathStream.h"
  43 #include "mathed/MathSupport.h"
  44
  45 #include "support/debug.h"
  46 #include "support/docstream.h"
  47 #include "support/FileName.h"
  48 #include "support/gettext.h"
  49 #include "support/lassert.h"
  50 #include "support/lstrings.h"
  51 #include "support/textutils.h"
  52
  53 #include <unordered_map>
  54 #include <regex>
  55
  56 //#define ResultsDebug
  57 #define USE_QT_FOR_SEARCH
  58 #if defined(USE_QT_FOR_SEARCH)
  59         #include <QtCore>       // sets QT_VERSION
  60         #if (QT_VERSION >= 0x050000)
  61                 #include <QRegularExpression>
  62                 #define QTSEARCH 1
  63         #else
  64                 #define QTSEARCH 0
  65         #endif
  66 #else
  67         #define QTSEARCH 0
  68 #endif
  69
  70 using namespace std;
  71 using namespace lyx::support;
  72
  73 namespace lyx {
  74
  75 typedef unordered_map<string, string> AccentsMap;
  76 typedef unordered_map<string,string>::const_iterator AccentsIterator;
  77 static AccentsMap accents = unordered_map<string, string>();
  78
  79 // Helper class for deciding what should be ignored
  80 class IgnoreFormats {
  81  public:
  82         ///
  83         IgnoreFormats() = default;
  84         ///
  85         bool getFamily() const { return ignoreFamily_; }
  86         ///
  87         bool getSeries() const { return ignoreSeries_; }
  88         ///
  89         bool getShape() const { return ignoreShape_; }
  90         ///
  91         bool getSize() const { return ignoreSize_; }
  92         ///
  93         bool getUnderline() const { return ignoreUnderline_; }
  94         ///
  95         bool getMarkUp() const { return ignoreMarkUp_; }
  96         ///
  97         bool getStrikeOut() const { return ignoreStrikeOut_; }
  98         ///
  99         bool getSectioning() const { return ignoreSectioning_; }
 100         ///
 101         bool getFrontMatter() const { return ignoreFrontMatter_; }
 102         ///
 103         bool getColor() const { return ignoreColor_; }
 104         ///
 105         bool getLanguage() const { return ignoreLanguage_; }
 106         ///
 107         bool getDeleted() const { return ignoreDeleted_; }
 108         ///
 109         void setIgnoreDeleted(bool value);
 110         ///
 111         bool getNonContent() const { return searchNonContent_; }
 112         ///
 113         void setIgnoreFormat(string const & type, bool value, bool fromUser = true);
 114
 115 private:
 116         ///
 117         bool ignoreFamily_ = false;
 118         ///
 119         bool ignoreSeries_ = false;
 120         ///
 121         bool ignoreShape_ = false;
 122         ///
 123         bool ignoreSize_ = true;
 124         ///
 125         bool ignoreUnderline_ = false;
 126         ///
 127         bool ignoreMarkUp_ = false;
 128         ///
 129         bool ignoreStrikeOut_ = false;
 130         ///
 131         bool ignoreSectioning_ = false;
 132         ///
 133         bool ignoreFrontMatter_ = false;
 134         ///
 135         bool ignoreColor_ = false;
 136         ///
 137         bool ignoreLanguage_ = false;
 138         bool userSelectedIgnoreLanguage_ = false;
 139         ///
 140         bool ignoreDeleted_ = true;
 141         ///
 142         bool searchNonContent_ = true;
 143 };
 144
 145 void IgnoreFormats::setIgnoreFormat(string const & type, bool value, bool fromUser)
 146 {
 147         if (type == "color") {
 148                 ignoreColor_ = value;
 149         }
 150         else if (type == "language") {
 151                 if (fromUser) {
 152                         userSelectedIgnoreLanguage_ = value;
 153                         ignoreLanguage_ = value;
 154                 }
 155                 else
 156                         ignoreLanguage_ = (value || userSelectedIgnoreLanguage_);
 157         }
 158         else if (type == "sectioning") {
 159                 ignoreSectioning_ = value;
 160                 ignoreFrontMatter_ = value;
 161         }
 162         else if (type == "font") {
 163                 ignoreSeries_ = value;
 164                 ignoreShape_ = value;
 165                 ignoreFamily_ = value;
 166         }
 167         else if (type == "series") {
 168                 ignoreSeries_ = value;
 169         }
 170         else if (type == "shape") {
 171                 ignoreShape_ = value;
 172         }
 173         else if (type == "size") {
 174                 ignoreSize_ = value;
 175         }
 176         else if (type == "family") {
 177                 ignoreFamily_ = value;
 178         }
 179         else if (type == "markup") {
 180                 ignoreMarkUp_ = value;
 181         }
 182         else if (type == "underline") {
 183                 ignoreUnderline_ = value;
 184         }
 185         else if (type == "strike") {
 186                 ignoreStrikeOut_ = value;
 187         }
 188         else if (type == "deleted") {
 189                 ignoreDeleted_ = value;
 190         }
 191         else if (type == "non-output-content") {
 192                 searchNonContent_ = !value;
 193         }
 194 }
 195
 196 // The global variable that can be changed from outside
 197 IgnoreFormats ignoreFormats;
 198
 199
 200 void setIgnoreFormat(string const & type, bool value, bool fromUser)
 201 {
 202         ignoreFormats.setIgnoreFormat(type, value, fromUser);
 203 }
 204
 205
 206 namespace {
 207
 208 bool parse_bool(docstring & howto, bool const defvalue = false)
 209 {
 210         if (howto.empty())
 211                 return defvalue;
 212         docstring var;
 213         howto = split(howto, var, ' ');
 214         return var == "1";
 215 }
 216
 217
 218 class MatchString
 219 {
 220 public:
 221         MatchString(docstring const & s, bool cs, bool mw)
 222                 : str(s), case_sens(cs), whole_words(mw)
 223         {}
 224
 225         // returns true if the specified string is at the specified position
 226         // del specifies whether deleted strings in ct mode will be considered
 227         int operator()(Paragraph const & par, pos_type pos, bool del = true) const
 228         {
 229                 return par.find(str, case_sens, whole_words, pos, del);
 230         }
 231
 232 private:
 233         // search string
 234         docstring str;
 235         // case sensitive
 236         bool case_sens;
 237         // match whole words only
 238         bool whole_words;
 239 };
 240
 241
 242 int findForward(DocIterator & cur, DocIterator const endcur,
 243                 MatchString const & match,
 244                 bool find_del = true, bool onlysel = false)
 245 {
 246         for (; cur; cur.forwardChar()) {
 247                 if (onlysel && endcur.pit() == cur.pit()
 248                     && endcur.idx() == cur.idx() && endcur.pos() < cur.pos())
 249                         break;
 250                 if (cur.inTexted()) {
 251                         int len = match(cur.paragraph(), cur.pos(), find_del);
 252                         if (len > 0)
 253                                 return len;
 254                 }
 255         }
 256         return 0;
 257 }
 258
 259
 260 int findBackwards(DocIterator & cur, DocIterator const endcur,
 261                   MatchString const & match,
 262                   bool find_del = true, bool onlysel = false)
 263 {
 264         while (cur) {
 265                 cur.backwardChar();
 266                 if (onlysel && endcur.pit() == cur.pit()
 267                     && endcur.idx() == cur.idx() && endcur.pos() > cur.pos())
 268                         break;
 269                 if (cur.inTexted()) {
 270                         int len = match(cur.paragraph(), cur.pos(), find_del);
 271                         if (len > 0)
 272                                 return len;
 273                 }
 274         }
 275         return 0;
 276 }
 277
 278
 279 bool searchAllowed(docstring const & str)
 280 {
 281         if (str.empty()) {
 282                 frontend::Alert::error(_("Search error"), _("Search string is empty"));
 283                 return false;
 284         }
 285         return true;
 286 }
 287
 288 } // namespace
 289
 290
 291 bool findOne(BufferView * bv, docstring const & searchstr,
 292              bool case_sens, bool whole, bool forward,
 293              bool find_del, bool check_wrap, bool const auto_wrap,
 294              bool instant, bool onlysel)
 295 {
 296         // Clean up previous selections with empty searchstr on instant
 297         if (searchstr.empty() && instant) {
 298                 if (bv->cursor().selection()) {
 299                         bv->setCursor(bv->cursor().selectionBegin());
 300                         bv->clearSelection();
 301                 }
 302                 return true;
 303         }
 304
 305         if (!searchAllowed(searchstr))
 306                 return false;
 307
 308         DocIterator const endcur = forward ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin();
 309
 310         if (onlysel && bv->cursor().selection()) {
 311                 docstring const matchstring = bv->cursor().selectionAsString(false);
 312                 docstring const lcmatchsting = support::lowercase(matchstring);
 313                 if (matchstring == searchstr || (!case_sens && lcmatchsting == lowercase(searchstr))) {
 314                         docstring q = _("The search string matches the selection, and search is limited to selection.\n"
 315                                         "Continue search outside?");
 316                         int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
 317                                 q, 0, 1, _("&Yes"), _("&No"));
 318                         if (search_answer == 0) {
 319                                 bv->clearSelection();
 320                                 if (findOne(bv, searchstr, case_sens, whole, forward,
 321                                             find_del, check_wrap, auto_wrap, false, false))
 322                                         return true;
 323                         }
 324                         return false;
 325                 }
 326         }
 327
 328         DocIterator cur = forward
 329                 ? ((instant || onlysel) ? bv->cursor().selectionBegin() : bv->cursor().selectionEnd())
 330                 : ((instant || onlysel) ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin());
 331
 332         MatchString const match(searchstr, case_sens, whole);
 333
 334         int match_len = forward
 335                 ? findForward(cur, endcur, match, find_del, onlysel)
 336                 : findBackwards(cur, endcur, match, find_del, onlysel);
 337
 338         if (match_len > 0)
 339                 bv->putSelectionAt(cur, match_len, !forward);
 340         else if (onlysel) {
 341                 docstring q = _("The search string was not found within the selection.\n"
 342                                 "Continue search outside?");
 343                 int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
 344                         q, 0, 1, _("&Yes"), _("&No"));
 345                 if (search_answer == 0) {
 346                         bv->clearSelection();
 347                         if (findOne(bv, searchstr, case_sens, whole, forward,
 348                                     find_del, check_wrap, auto_wrap, false, false))
 349                                 return true;
 350                 }
 351                 return false;
 352         }
 353         else if (check_wrap) {
 354                 DocIterator cur_orig(bv->cursor());
 355                 bool wrap = auto_wrap;
 356                 if (!auto_wrap) {
 357                         docstring q;
 358                         if (forward)
 359                                 q = _("End of file reached while searching forward.\n"
 360                                   "Continue searching from the beginning?");
 361                         else
 362                                 q = _("Beginning of file reached while searching backward.\n"
 363                                   "Continue searching from the end?");
 364                         int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 365                                 q, 0, 1, _("&Yes"), _("&No"));
 366                         wrap = wrap_answer == 0;
 367                 }
 368                 if (wrap) {
 369                         if (forward) {
 370                                 bv->cursor().clear();
 371                                 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
 372                         } else {
 373                                 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
 374                                 bv->cursor().backwardPos();
 375                         }
 376                         if (auto_wrap) {
 377                                 docstring const msg = forward
 378                                   ? _("Search reached end of document, continuing from beginning.")
 379                                   : _("Search reached beginning of document, continuing from end.");
 380                                 bv->message(msg);
 381                         }
 382                         bv->clearSelection();
 383                         if (findOne(bv, searchstr, case_sens, whole, forward,
 384                                     find_del, false, false, false, false))
 385                                 return true;
 386                 }
 387                 bv->cursor().setCursor(cur_orig);
 388                 return false;
 389         }
 390
 391         return match_len > 0;
 392 }
 393
 394
 395 namespace {
 396
 397 int replaceAll(BufferView * bv,
 398                docstring const & searchstr, docstring const & replacestr,
 399                bool case_sens, bool whole, bool onlysel)
 400 {
 401         Buffer & buf = bv->buffer();
 402
 403         if (!searchAllowed(searchstr) || buf.isReadonly())
 404                 return 0;
 405
 406         DocIterator startcur = bv->cursor().selectionBegin();
 407         DocIterator endcur = bv->cursor().selectionEnd();
 408         bool const had_selection = bv->cursor().selection();
 409
 410         MatchString const match(searchstr, case_sens, whole);
 411         int num = 0;
 412
 413         int const rsize = replacestr.size();
 414         int const ssize = searchstr.size();
 415
 416         Cursor cur(*bv);
 417         cur.setCursor(doc_iterator_begin(&buf));
 418         int match_len = findForward(cur, endcur, match, false, onlysel);
 419         while (match_len > 0) {
 420                 // Backup current cursor position and font.
 421                 pos_type const pos = cur.pos();
 422                 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
 423                 cur.recordUndo();
 424                 int ct_deleted_text = ssize -
 425                         cur.paragraph().eraseChars(pos, pos + match_len,
 426                                                    buf.params().track_changes);
 427                 cur.paragraph().insert(pos, replacestr, font,
 428                                        Change(buf.params().track_changes
 429                                               ? Change::INSERTED
 430                                               : Change::UNCHANGED));
 431                 for (int i = 0; i < rsize + ct_deleted_text
 432                      && cur.pos() < cur.lastpos(); ++i)
 433                         cur.forwardPos();
 434                 if (onlysel && cur.pit() == endcur.pit() && cur.idx() == endcur.idx()) {
 435                         // Adjust end of selection for replace-all in selection
 436                         if (rsize > ssize) {
 437                                 int const offset = rsize - ssize;
 438                                 for (int i = 0; i < offset + ct_deleted_text
 439                                      && endcur.pos() < endcur.lastpos(); ++i)
 440                                         endcur.forwardPos();
 441                         } else {
 442                                 int const offset = ssize - rsize;
 443                                 for (int i = 0; i < offset && endcur.pos() > 0; ++i)
 444                                         endcur.backwardPos();
 445                                 for (int i = 0; i < ct_deleted_text
 446                                      && endcur.pos() < endcur.lastpos(); ++i)
 447                                         endcur.forwardPos();
 448                         }
 449                 }
 450                 ++num;
 451                 match_len = findForward(cur, endcur, match, false, onlysel);
 452         }
 453
 454         bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
 455
 456         startcur.fixIfBroken();
 457         bv->setCursor(startcur);
 458
 459         // Reset selection, accounting for changes in selection
 460         if (had_selection) {
 461                 endcur.fixIfBroken();
 462                 bv->cursor().resetAnchor();
 463                 bv->setCursorSelectionTo(endcur);
 464         }
 465
 466         return num;
 467 }
 468
 469
 470 // the idea here is that we are going to replace the string that
 471 // is selected IF it is the search string.
 472 // if there is a selection, but it is not the search string, then
 473 // we basically ignore it. (FIXME We ought to replace only within
 474 // the selection.)
 475 // if there is no selection, then:
 476 //  (i) if some search string has been provided, then we find it.
 477 //      (think of how the dialog works when you hit "replace" the
 478 //      first time.)
 479 // (ii) if no search string has been provided, then we treat the
 480 //      word the cursor is in as the search string. (why? i have no
 481 //      idea.) but this only works in text?
 482 //
 483 // returns the number of replacements made (one, if any) and
 484 // whether anything at all was done.
 485 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
 486                            docstring const & replacestr, bool case_sens,
 487                            bool whole, bool forward, bool findnext, bool wrap,
 488                            bool onlysel)
 489 {
 490         Cursor & cur = bv->cursor();
 491         if (!cur.selection() || onlysel) {
 492                 // no selection, non-empty search string: find it
 493                 if (!searchstr.empty()) {
 494                         bool const found = findOne(bv, searchstr, case_sens, whole,
 495                                                    forward, true, findnext, wrap, false, onlysel);
 496                         return make_pair(found, 0);
 497                 }
 498                 // empty search string
 499                 if (!cur.inTexted())
 500                         // bail in math
 501                         return make_pair(false, 0);
 502                 // select current word and treat it as the search string.
 503                 // This causes a minor bug as undo will restore this selection,
 504                 // which the user did not create (#8986).
 505                 cur.innerText()->selectWord(cur, WHOLE_WORD);
 506                 searchstr = cur.selectionAsString(false, true);
 507         }
 508
 509         // if we still don't have a search string, report the error
 510         // and abort.
 511         if (!searchAllowed(searchstr))
 512                 return make_pair(false, 0);
 513
 514         bool have_selection = cur.selection();
 515         docstring const selected = cur.selectionAsString(false, true);
 516         bool match =
 517                 case_sens
 518                 ? searchstr == selected
 519                 : compare_no_case(searchstr, selected) == 0;
 520
 521         // no selection or current selection is not search word:
 522         // just find the search word
 523         if (!have_selection || !match) {
 524                 bool const found = findOne(bv, searchstr, case_sens, whole, forward,
 525                                            true, findnext, wrap, false, onlysel);
 526                 return make_pair(found, 0);
 527         }
 528
 529         // we're now actually ready to replace. if the buffer is
 530         // read-only, we can't, though.
 531         if (bv->buffer().isReadonly())
 532                 return make_pair(false, 0);
 533
 534         cap::replaceSelectionWithString(cur, replacestr);
 535         if (forward) {
 536                 cur.pos() += replacestr.length();
 537                 LASSERT(cur.pos() <= cur.lastpos(),
 538                         cur.pos() = cur.lastpos());
 539         }
 540         if (findnext)
 541                 findOne(bv, searchstr, case_sens, whole,
 542                         forward, false, findnext, wrap, false, onlysel);
 543
 544         return make_pair(true, 1);
 545 }
 546
 547 } // namespace
 548
 549
 550 docstring const find2string(docstring const & search,
 551                             bool casesensitive, bool matchword,
 552                             bool forward, bool wrap, bool instant,
 553                             bool onlysel)
 554 {
 555         odocstringstream ss;
 556         ss << search << '\n'
 557            << int(casesensitive) << ' '
 558            << int(matchword) << ' '
 559            << int(forward) << ' '
 560            << int(wrap) << ' '
 561            << int(instant) << ' '
 562            << int(onlysel);
 563         return ss.str();
 564 }
 565
 566
 567 docstring const replace2string(docstring const & replace,
 568                                docstring const & search,
 569                                bool casesensitive, bool matchword,
 570                                bool all, bool forward, bool findnext,
 571                                bool wrap, bool onlysel)
 572 {
 573         odocstringstream ss;
 574         ss << replace << '\n'
 575            << search << '\n'
 576            << int(casesensitive) << ' '
 577            << int(matchword) << ' '
 578            << int(all) << ' '
 579            << int(forward) << ' '
 580            << int(findnext) << ' '
 581            << int(wrap) << ' '
 582            << int(onlysel);
 583         return ss.str();
 584 }
 585
 586
 587 docstring const string2find(docstring const & argument,
 588                               bool &casesensitive,
 589                               bool &matchword,
 590                               bool &forward,
 591                               bool &wrap,
 592                               bool &instant,
 593                               bool &onlysel)
 594 {
 595         // data is of the form
 596         // "<search>
 597         //  <casesensitive> <matchword> <forward> <wrap> <onlysel>"
 598         docstring search;
 599         docstring howto = split(argument, search, '\n');
 600
 601         casesensitive = parse_bool(howto);
 602         matchword     = parse_bool(howto);
 603         forward       = parse_bool(howto, true);
 604         wrap          = parse_bool(howto);
 605         instant       = parse_bool(howto);
 606         onlysel       = parse_bool(howto);
 607
 608         return search;
 609 }
 610
 611
 612 bool lyxfind(BufferView * bv, FuncRequest const & ev)
 613 {
 614         if (!bv || ev.action() != LFUN_WORD_FIND)
 615                 return false;
 616
 617         //lyxerr << "find called, cmd: " << ev << endl;
 618         bool casesensitive;
 619         bool matchword;
 620         bool forward;
 621         bool wrap;
 622         bool instant;
 623         bool onlysel;
 624
 625         docstring search = string2find(ev.argument(), casesensitive,
 626                                        matchword, forward, wrap, instant, onlysel);
 627
 628         return findOne(bv, search, casesensitive, matchword, forward,
 629                        false, true, wrap, instant, onlysel);
 630 }
 631
 632
 633 bool lyxreplace(BufferView * bv, FuncRequest const & ev)
 634 {
 635         if (!bv || ev.action() != LFUN_WORD_REPLACE)
 636                 return false;
 637
 638         // data is of the form
 639         // "<search>
 640         //  <replace>
 641         //  <casesensitive> <matchword> <all> <forward> <findnext> <wrap> <onlysel>"
 642         docstring search;
 643         docstring rplc;
 644         docstring howto = split(ev.argument(), rplc, '\n');
 645         howto = split(howto, search, '\n');
 646
 647         bool casesensitive = parse_bool(howto);
 648         bool matchword     = parse_bool(howto);
 649         bool all           = parse_bool(howto);
 650         bool forward       = parse_bool(howto, true);
 651         bool findnext      = parse_bool(howto, true);
 652         bool wrap          = parse_bool(howto);
 653         bool onlysel       = parse_bool(howto);
 654
 655         if (!bv->cursor().selection())
 656                 // only selection only makes sense with selection
 657                 onlysel = false;
 658
 659         bool update = false;
 660
 661         int replace_count = 0;
 662         if (all) {
 663                 replace_count = replaceAll(bv, search, rplc, casesensitive,
 664                                            matchword, onlysel);
 665                 update = replace_count > 0;
 666         } else {
 667                 pair<bool, int> rv =
 668                         replaceOne(bv, search, rplc, casesensitive, matchword,
 669                                    forward, findnext, wrap, onlysel);
 670                 update = rv.first;
 671                 replace_count = rv.second;
 672         }
 673
 674         Buffer const & buf = bv->buffer();
 675         if (!update) {
 676                 // emit message signal.
 677                 if (onlysel)
 678                         buf.message(_("String not found in selection."));
 679                 else
 680                         buf.message(_("String not found."));
 681         } else {
 682                 if (replace_count == 0) {
 683                         buf.message(_("String found."));
 684                 } else if (replace_count == 1) {
 685                         buf.message(_("String has been replaced."));
 686                 } else {
 687                         docstring const str = onlysel
 688                                         ? bformat(_("%1$d strings have been replaced in the selection."), replace_count)
 689                                         : bformat(_("%1$d strings have been replaced."), replace_count);
 690                         buf.message(str);
 691                 }
 692         }
 693         return update;
 694 }
 695
 696
 697 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
 698 {
 699         for (; cur; cur.forwardPos())
 700                 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
 701                         return true;
 702
 703         if (check_wrap) {
 704                 DocIterator cur_orig(bv->cursor());
 705                 docstring q = _("End of file reached while searching forward.\n"
 706                           "Continue searching from the beginning?");
 707                 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 708                         q, 0, 1, _("&Yes"), _("&No"));
 709                 if (wrap_answer == 0) {
 710                         bv->cursor().clear();
 711                         bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
 712                         bv->clearSelection();
 713                         cur.setCursor(bv->cursor().selectionBegin());
 714                         if (findNextChange(bv, cur, false))
 715                                 return true;
 716                 }
 717                 bv->cursor().setCursor(cur_orig);
 718         }
 719
 720         return false;
 721 }
 722
 723
 724 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
 725 {
 726         for (cur.backwardPos(); cur; cur.backwardPos()) {
 727                 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
 728                         return true;
 729         }
 730
 731         if (check_wrap) {
 732                 DocIterator cur_orig(bv->cursor());
 733                 docstring q = _("Beginning of file reached while searching backward.\n"
 734                           "Continue searching from the end?");
 735                 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
 736                         q, 0, 1, _("&Yes"), _("&No"));
 737                 if (wrap_answer == 0) {
 738                         bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
 739                         bv->cursor().backwardPos();
 740                         bv->clearSelection();
 741                         cur.setCursor(bv->cursor().selectionBegin());
 742                         if (findPreviousChange(bv, cur, false))
 743                                 return true;
 744                 }
 745                 bv->cursor().setCursor(cur_orig);
 746         }
 747
 748         return false;
 749 }
 750
 751
 752 bool selectChange(Cursor & cur, bool forward)
 753 {
 754         if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
 755                 return false;
 756         Change ch = cur.paragraph().lookupChange(cur.pos());
 757
 758         CursorSlice tip1 = cur.top();
 759         for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
 760                 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
 761                 if (!ch2.isSimilarTo(ch))
 762                         break;
 763         }
 764         CursorSlice tip2 = cur.top();
 765         for (; tip2.pit() > 0 || tip2.pos() > 0;) {
 766                 tip2.backwardPos();
 767                 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
 768                 if (!ch2.isSimilarTo(ch)) {
 769                         // take a step forward to correctly set the selection
 770                         tip2.forwardPos();
 771                         break;
 772                 }
 773         }
 774         if (forward)
 775                 swap(tip1, tip2);
 776         cur.top() = tip1;
 777         cur.bv().mouseSetCursor(cur, false);
 778         cur.top() = tip2;
 779         cur.bv().mouseSetCursor(cur, true);
 780         return true;
 781 }
 782
 783
 784 namespace {
 785
 786
 787 bool findChange(BufferView * bv, bool forward)
 788 {
 789         Cursor cur(*bv);
 790         cur.setCursor(forward ? bv->cursor().selectionEnd()
 791                       : bv->cursor().selectionBegin());
 792         forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
 793         return selectChange(cur, forward);
 794 }
 795
 796 } // namespace
 797
 798 bool findNextChange(BufferView * bv)
 799 {
 800         return findChange(bv, true);
 801 }
 802
 803
 804 bool findPreviousChange(BufferView * bv)
 805 {
 806         return findChange(bv, false);
 807 }
 808
 809
 810
 811 namespace {
 812
 813 typedef vector<pair<string, string> > Escapes;
 814
 815 string string2regex(string in)
 816 {
 817         static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\$\\])" };
 818         string tempx = std::regex_replace(in, specialChars,  R"(\$&)" );
 819         // Special handling for ' '
 820         string temp("");
 821         int blanks = 0;
 822         for (unsigned i = 0; i < tempx.size(); i++) {
 823                 if (tempx[i] == ' ' || tempx[i] == '~' ) {
 824                         // normal blanks
 825                         blanks++;
 826                 }
 827                 else if ((tempx[i] == '\302' && tempx[i+1] == '\240')
 828                         || (tempx[i] == '\342' && tempx[i+1] == '\200')) {
 829                         // protected space
 830                         // thin space
 831                         blanks++;
 832                         i++;
 833                 }
 834                 else {
 835                         if (blanks > 0) {
 836                                 temp += "\\s+";
 837                         }
 838                         temp += tempx[i];
 839                         blanks = 0;
 840                 }
 841         }
 842         if (blanks > 0) {
 843                 temp += "\\s+";
 844         }
 845
 846         string temp2("");
 847         size_t lastpos = 0;
 848         size_t fl_pos = 0;
 849         int offset = 1;
 850         while (fl_pos < temp.size()) {
 851                 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
 852                 if (fl_pos == string::npos)
 853                         break;
 854                 offset = 16;
 855                 temp2 += temp.substr(lastpos, fl_pos - lastpos);
 856                 temp2 += "\\n";
 857                 lastpos = fl_pos;
 858         }
 859         if (lastpos == 0)
 860                 return(temp);
 861         if (lastpos < temp.size()) {
 862                 temp2 += temp.substr(lastpos, temp.size() - lastpos);
 863         }
 864         return temp2;
 865 }
 866
 867 static void buildAccentsMap();
 868
 869 string correctRegex(string t, bool withformat)
 870 {
 871         /* Convert \backslash => \
 872          * and \{, \}, \[, \] => {, }, [, ]
 873          */
 874         string s("");
 875         regex wordre("(\\\\)*(\\\\(([A-Za-z]+|[\\{\\}%])( |\\{\\})?|[\\[\\]\\{\\}]))");
 876         size_t lastpos = 0;
 877         smatch sub;
 878         bool backslashed = false;
 879         if (accents.empty())
 880                 buildAccentsMap();
 881
 882         //LYXERR0("correctRegex input '" << t << "'");
 883         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
 884                 sub = *it;
 885                 string replace;
 886                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
 887                         continue;
 888                 }
 889                 else {
 890                         if (sub.str(4) == "backslash") {
 891                                 replace = "\\";
 892                                 {
 893                                         // transforms '\backslash \{' into '\{'
 894                                         string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
 895                                         if ((next == "\\{") || (next == "\\}")) {
 896                                                 replace = "";
 897                                                 backslashed = true;
 898                                         }
 899                                 }
 900                         }
 901                         else if (sub.str(4) == "mathcircumflex")
 902                                 replace = "^";
 903                         else if (backslashed) {
 904                                 backslashed = false;
 905                                 if (withformat) {
 906                                         if (sub.str(3) == "{")
 907                                                 replace = accents["braceleft"];
 908                                         else if (sub.str(3) == "}")
 909                                                 replace = accents["braceright"];
 910                                         else {
 911                                                 // else part should not exist
 912                                                 LASSERT(0, /**/);
 913                                         }
 914                                 }
 915                                 else {
 916                                         if (sub.str(3) == "{")
 917                                                 replace = "\\{";
 918                                         else if (sub.str(3) == "}")
 919                                                 replace = "\\}";
 920                                         else {
 921                                                 // else part should not exist
 922                                                 LASSERT(0, /**/);
 923                                         }
 924                                 }
 925                         }
 926                         else if (sub.str(4) == "{") // transforms '\{' into '{'
 927                                 replace = "{";
 928                         else if (sub.str(4) == "}")
 929                                 replace = "}";
 930                         else if (sub.str(4) == "%")
 931                                 replace = "%";
 932                         else {
 933                                 AccentsIterator it_ac = accents.find(sub.str(4));
 934                                 if (it_ac == accents.end()) {
 935                                         replace = sub.str(2);
 936                                 }
 937                                 else {
 938                                         replace = it_ac->second;
 939                                 }
 940                         }
 941                 }
 942                 if (lastpos < (size_t) sub.position(2))
 943                         s += t.substr(lastpos, sub.position(2) - lastpos);
 944                 s += replace;
 945                 lastpos = sub.position(2) + sub.length(2);
 946         }
 947         if (lastpos == 0)
 948                 s = t;
 949         else if (lastpos < t.length())
 950                 s += t.substr(lastpos, t.length() - lastpos);
 951         // Handle quotes in regex
 952         // substitute all '„', '“', '»', '«' with '"'
 953         // and all '‚', '‘', '›', '‹' with "\'"
 954         static std::regex plainquotes { R"(„|“|»|«)" };
 955         static std::regex innerquotes { R"(‚|‘|›|‹)" };
 956         t = std::regex_replace(s, plainquotes, R"(")");
 957         s = std::regex_replace(t, innerquotes, R"(')");
 958         //LYXERR0("correctRegex output '" << s << "'");
 959         return s;
 960 }
 961
 962 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
 963 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
 964 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
 965 string escape_for_regex(string s, bool withformat)
 966 {
 967         size_t lastpos = 0;
 968         string result = "";
 969         while (lastpos < s.size()) {
 970                 size_t regex_pos = s.find("\\regexp{", lastpos);
 971                 if (regex_pos == string::npos) {
 972                         regex_pos = s.size();
 973                 }
 974                 if (regex_pos > lastpos) {
 975                         result += string2regex(s.substr(lastpos, regex_pos-lastpos));
 976                         lastpos = regex_pos;
 977                         if (lastpos == s.size())
 978                                 break;
 979                 }
 980                 size_t end_pos = s.find("\\endregexp", regex_pos + 8);
 981                 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
 982                 lastpos = end_pos + 13;
 983         }
 984         return result;
 985 }
 986
 987
 988 /// Wrapper for lyx::regex_replace with simpler interface
 989 bool regex_replace(string const & s, string & t, string const & searchstr,
 990                    string const & replacestr)
 991 {
 992         regex e(searchstr, regex_constants::ECMAScript);
 993         ostringstream oss;
 994         ostream_iterator<char, char> it(oss);
 995         regex_replace(it, s.begin(), s.end(), e, replacestr);
 996         // tolerate t and s be references to the same variable
 997         bool rv = (s != oss.str());
 998         t = oss.str();
 999         return rv;
1000 }
1001
1002 class MatchResult {
1003 public:
1004         enum range {
1005                 newIsTooFar,
1006                 newIsBetter,
1007                 newIsInvalid
1008         };
1009         int match_len;
1010         int match_prefix;
1011         int match2end;
1012         int pos;
1013         int leadsize;
1014         int pos_len;
1015         int searched_size;
1016         vector <string> result = vector <string>();
1017         MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {}
1018 };
1019
1020 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
1021 {
1022         if (newres.match2end < oldres.match2end)
1023                 return MatchResult::newIsTooFar;
1024         if (newres.match_len < oldres.match_len)
1025                 return MatchResult::newIsTooFar;
1026
1027         if (newres.match_len == oldres.match_len) {
1028                 if (newres.match2end == oldres.match2end)
1029                         return MatchResult::newIsBetter;
1030         }
1031         return MatchResult::newIsInvalid;
1032 }
1033
1034 /** The class performing a match between a position in the document and the FindAdvOptions.
1035  **/
1036
1037 class MatchStringAdv {
1038 public:
1039         MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
1040
1041         /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
1042          ** constructor as opt.search, under the opt.* options settings.
1043          **
1044          ** @param at_begin
1045          **     If set, then match is searched only against beginning of text starting at cur.
1046          **     If unset, then match is searched anywhere in text starting at cur.
1047          **
1048          ** @return
1049          ** The length of the matching text, or zero if no match was found.
1050          **/
1051         MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1052 #if QTSEARCH
1053         bool regexIsValid;
1054         string regexError;
1055 #endif
1056
1057 public:
1058         /// buffer
1059         lyx::Buffer * p_buf;
1060         /// first buffer on which search was started
1061         lyx::Buffer * const p_first_buf;
1062         /// options
1063         FindAndReplaceOptions const & opt;
1064
1065 private:
1066         /// Auxiliary find method (does not account for opt.matchword)
1067         MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1068         void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
1069
1070         /** Normalize a stringified or latexified LyX paragraph.
1071          **
1072          ** Normalize means:
1073          ** <ul>
1074          **   <li>if search is not casesensitive, then lowercase the string;
1075          **   <li>remove any newline at begin or end of the string;
1076          **   <li>replace any newline in the middle of the string with a simple space;
1077          **   <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
1078          ** </ul>
1079          **
1080          ** @todo Normalization should also expand macros, if the corresponding
1081          ** search option was checked.
1082          **/
1083         string normalize(docstring const & s, bool ignore_fomat) const;
1084         // normalized string to search
1085         string par_as_string;
1086         // regular expression to use for searching
1087         // regexp2 is same as regexp, but prefixed with a ".*?"
1088 #if QTSEARCH
1089         QRegularExpression regexp;
1090         QRegularExpression regexp2;
1091 #else
1092         regex regexp;
1093         regex regexp2;
1094 #endif
1095         // leading format material as string
1096         string lead_as_string;
1097         // par_as_string after removal of lead_as_string
1098         string par_as_string_nolead;
1099         // unmatched open braces in the search string/regexp
1100         int open_braces;
1101         // number of (.*?) subexpressions added at end of search regexp for closing
1102         // environments, math mode, styles, etc...
1103         int close_wildcards;
1104 public:
1105         // Are we searching with regular expressions ?
1106         bool use_regexp = false;
1107         static int valid_matches;
1108         static vector <string> matches;
1109         void FillResults(MatchResult &found_mr);
1110 };
1111
1112 int MatchStringAdv::valid_matches = 0;
1113 vector <string> MatchStringAdv::matches = vector <string>(10);
1114
1115 void MatchStringAdv::FillResults(MatchResult &found_mr)
1116 {
1117         if (found_mr.match_len > 0) {
1118                 valid_matches = found_mr.result.size();
1119                 for (size_t i = 0; i < found_mr.result.size(); i++)
1120                         matches[i] = found_mr.result[i];
1121         } else
1122                 valid_matches = 0;
1123 }
1124
1125 static void setFindParams(OutputParams &runparams)
1126 {
1127         runparams.flavor = Flavor::XeTeX;
1128         //runparams.use_polyglossia = true;
1129         runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1130         // No side effect of file copying and image conversion
1131         runparams.dryrun = true;
1132 }
1133
1134 static docstring buffer_to_latex(Buffer & buffer)
1135 {
1136         //OutputParams runparams(&buffer.params().encoding());
1137         OutputParams runparams(encodings.fromLyXName("utf8"));
1138         odocstringstream ods;
1139         otexstream os(ods);
1140         runparams.nice = true;
1141         setFindParams(runparams);
1142         if (ignoreFormats.getDeleted())
1143                 runparams.for_search = OutputParams::SearchWithoutDeleted;
1144         else
1145                 runparams.for_search = OutputParams::SearchWithDeleted;
1146         if (ignoreFormats.getNonContent()) {
1147                 runparams.for_search |= OutputParams::SearchNonOutput;
1148         }
1149         pit_type const endpit = buffer.paragraphs().size();
1150         for (pit_type pit = 0; pit != endpit; ++pit) {
1151                 TeXOnePar(buffer, buffer.text(), pit, os, runparams, string(), -1, -1, true);
1152                 LYXERR(Debug::FINDVERBOSE, "searchString up to here: " << ods.str());
1153         }
1154         return ods.str();
1155 }
1156
1157 static string latexNamesToUtf8(docstring strIn)
1158 {
1159         string addtmp = to_utf8(strIn);
1160         static regex const rmAcc("(\\\\)*("
1161                                          "\\\\([A-Za-z]+\\{.\\})"       // e.g. "ddot{A}" == sub.str(3)
1162                                         "|\\\\([A-Za-z]+)( |\\{\\})?"   // e.g. "LyX", "LyX{}", "LyX " == sub.str(4)
1163                                         ")"
1164                                 );
1165         size_t lastpos = 0;
1166         smatch sub;
1167         string replace;
1168         string add("");
1169         if (accents.empty())
1170                 buildAccentsMap();
1171         for (sregex_iterator it_add(addtmp.begin(), addtmp.end(), rmAcc), end; it_add != end; ++it_add) {
1172                 sub = *it_add;
1173                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
1174                         continue;
1175                 }
1176                 else {
1177                         string key;
1178                         if (sub.length(3) > 0)
1179                                 key = sub.str(3);
1180                         else
1181                                 key = sub.str(4);
1182                         AccentsIterator it_ac = accents.find(key);
1183                         if (it_ac == accents.end()) {
1184                                 replace = sub.str(2);
1185                         }
1186                         else {
1187                                 replace = it_ac->second;
1188                         }
1189                 }
1190                 if (lastpos < (size_t) sub.position(2))
1191                         add += addtmp.substr(lastpos, sub.position(2) - lastpos);
1192                 add += replace;
1193                 lastpos = sub.position(2) + sub.length(2);
1194         }
1195         if (lastpos == 0)
1196                 add = addtmp;
1197         else if (addtmp.length() > lastpos)
1198                 add += addtmp.substr(lastpos, addtmp.length() - lastpos);
1199         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1200                         << add << "'");
1201         return add;
1202 }
1203
1204 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1205 {
1206         docstring str;
1207         if (!opt.ignoreformat) {
1208                 str = buffer_to_latex(buffer);
1209         } else {
1210                 // OutputParams runparams(&buffer.params().encoding());
1211                 OutputParams runparams(encodings.fromLyXName("utf8"));
1212                 runparams.nice = true;
1213                 setFindParams(runparams);
1214                 int option = AS_STR_INSETS |AS_STR_PLAINTEXT;
1215                 if (ignoreFormats.getDeleted()) {
1216                         option |= AS_STR_SKIPDELETE;
1217                         runparams.for_search = OutputParams::SearchWithoutDeleted;
1218                 }
1219                 else {
1220                         runparams.for_search = OutputParams::SearchWithDeleted;
1221                 }
1222                 if (ignoreFormats.getNonContent()) {
1223                         runparams.for_search |= OutputParams::SearchNonOutput;
1224                 }
1225                 string t("");
1226                 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1227                         Paragraph const & par = buffer.paragraphs().at(pit);
1228                         string add = latexNamesToUtf8(par.asString(pos_type(0), par.size(),
1229                                                                 option,
1230                                                                 &runparams));
1231                         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1232                                 << add << "'");
1233                         t += add;
1234                 }
1235                 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1236                 while (regex_replace(t, t, "\\\\(text|lyxmathsym|ensuremath)\\{([^\\}]*)\\}", "$2"));
1237                 str = from_utf8(t);
1238         }
1239         return str;
1240 }
1241
1242
1243 /// Return separation pos between the leading material and the rest
1244 static size_t identifyLeading(string const & s)
1245 {
1246         string t = s;
1247         // @TODO Support \item[text]
1248         // Kornel: Added textsl, textsf, textit, texttt and noun
1249         // + allow to search for colored text too
1250         while (regex_replace(t, t, "^\\\\(("
1251                              "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1252                              "lyxaddress|lyxrightaddress|"
1253                              "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1254                              "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1255                              "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1256                              "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1257                || regex_replace(t, t, "^\\$", "")
1258                || regex_replace(t, t, "^\\\\\\[", "")
1259                || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1260                || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1261                ;
1262         LYXERR(Debug::FINDVERBOSE, "  after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1263         return s.size() - t.size();
1264 }
1265
1266 /*
1267  * Given a latexified string, retrieve some handled features
1268  * The features of the regex will later be compared with the features
1269  * of the searched text. If the regex features are not a
1270  * subset of the analized, then, in not format ignoring search
1271  * we can early stop the search in the relevant inset.
1272  */
1273 typedef map<string, bool> Features;
1274
1275 static Features identifyFeatures(string const & s)
1276 {
1277         static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1278         static regex const valid("^("
1279                 "("
1280                         "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1281                                 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1282                                 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1283                         "(u|uu)line|(s|x)out|uwave|"
1284                         "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1285                 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1286         smatch sub;
1287         bool displ = true;
1288         Features info;
1289
1290         for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1291                 sub = *it;
1292                 if (displ) {
1293                         if (sub.str(1).compare("regexp") == 0) {
1294                                 displ = false;
1295                                 continue;
1296                         }
1297                         string token = sub.str(1);
1298                         smatch sub2;
1299                         if (regex_match(token, sub2, valid)) {
1300                                 info[token] = true;
1301                         }
1302                         else {
1303                                 // ignore
1304                         }
1305                 }
1306                 else {
1307                         if (sub.str(1).compare("endregexp") == 0) {
1308                                 displ = true;
1309                                 continue;
1310                         }
1311                 }
1312         }
1313         return info;
1314 }
1315
1316 /*
1317  * defines values features of a key "\\[a-z]+{"
1318  */
1319 class KeyInfo {
1320 public:
1321         enum KeyType {
1322                 /* Char type with content discarded
1323                  * like \hspace{1cm} */
1324                 noContent,
1325                 /* Char, like \backslash */
1326                 isChar,
1327                 /* replace starting backslash with '#' */
1328                 isText,
1329                 /* \part, \section*, ... */
1330                 isSectioning,
1331                 /* title, author etc */
1332                 isTitle,
1333                 /* \foreignlanguage{ngerman}, ... */
1334                 isMain,
1335                 /* inside \code{}
1336                  * to discard language in content */
1337                 noMain,
1338                 isRegex,
1339                 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1340                 isMath,
1341                 /* fonts, colors, markups, ... */
1342                 isStandard,
1343                 /* footnotesize, ... large, ...
1344                  * Ignore all of them */
1345                 isSize,
1346                 invalid,
1347                 /* inputencoding, ...
1348                  * Discard also content, because they do not help in search */
1349                 doRemove,
1350                 /* twocolumns, ...
1351                  * like remove, but also all arguments */
1352                 removeWithArg,
1353                 /* item, listitem */
1354                 isList,
1355                 /* tex, latex, ... like isChar */
1356                 isIgnored,
1357                 /* like \lettrine[lines=5]{}{} */
1358                 cleanToStart,
1359                 // like isStandard, but always remove head
1360                 headRemove,
1361                 /* End of arguments marker for lettrine,
1362                  * so that they can be ignored */
1363                 endArguments
1364         };
1365         KeyInfo() = default;
1366         KeyInfo(KeyType type, int parcount, bool disable)
1367                 : keytype(type),
1368                   parenthesiscount(parcount),
1369                   disabled(disable) {}
1370         KeyType keytype = invalid;
1371         string head;
1372         int _tokensize = -1;
1373         int _tokenstart = -1;
1374         int _dataStart = -1;
1375         int _dataEnd = -1;
1376         int parenthesiscount = 1;
1377         bool disabled = false;
1378         bool used = false; /* by pattern */
1379 };
1380
1381 class Border {
1382 public:
1383         Border(int l=0, int u=0) : low(l), upper(u) {}
1384         int low;
1385         int upper;
1386 };
1387
1388 #define MAXOPENED 30
1389 class Intervall {
1390         bool isPatternString_;
1391 public:
1392         explicit Intervall(bool isPattern, string const & p)
1393                 : isPatternString_(isPattern), par(p), ignoreidx(-1),
1394                   actualdeptindex(0), hasTitle(false), langcount(0)
1395         {
1396                 depts[0] = 0;
1397                 closes[0] = 0;
1398         }
1399
1400         string par;
1401         int ignoreidx;
1402         static vector<Border> borders;
1403         int depts[MAXOPENED];
1404         int closes[MAXOPENED];
1405         int actualdeptindex;
1406         int previousNotIgnored(int) const;
1407         int nextNotIgnored(int) const;
1408         void handleOpenP(int i);
1409         void handleCloseP(int i, bool closingAllowed);
1410         void resetOpenedP(int openPos);
1411         void addIntervall(int upper);
1412         void addIntervall(int low, int upper); /* if explicit */
1413         void removeAccents();
1414         void setForDefaultLang(KeyInfo const & defLang) const;
1415         int findclosing(int start, int end, char up, char down, int repeat);
1416         void handleParentheses(int lastpos, bool closingAllowed);
1417         bool hasTitle;
1418         // Number of disabled language specs up
1419         // to current position in actual interval
1420         int langcount;
1421         int isOpeningPar(int pos) const;
1422         string titleValue;
1423         void output(ostringstream &os, int lastpos);
1424         // string show(int lastpos);
1425 };
1426
1427 vector<Border> Intervall::borders = vector<Border>(30);
1428
1429 int Intervall::isOpeningPar(int pos) const
1430 {
1431         if ((pos < 0) || (size_t(pos) >= par.size()))
1432                 return 0;
1433         if (par[pos] != '{')
1434                 return 0;
1435         if (size_t(pos) + 2 >= par.size())
1436                 return 1;
1437         if (par[pos+2] != '}')
1438                 return 1;
1439         if (par[pos+1] == '[' || par[pos+1] == ']')
1440                 return 3;
1441         return 1;
1442 }
1443
1444 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1445 {
1446         // Enable the use of first token again
1447         if (ignoreidx >= 0) {
1448                 int value = defLang._tokenstart + defLang._tokensize;
1449                 int borderidx = 0;
1450                 if (hasTitle)
1451                         borderidx = 1;
1452                 if (value > 0) {
1453                         if (borders[borderidx].low < value)
1454                                 borders[borderidx].low = value;
1455                         if (borders[borderidx].upper < value)
1456                                 borders[borderidx].upper = value;
1457                 }
1458         }
1459 }
1460
1461 static void checkDepthIndex(int val)
1462 {
1463         static int maxdepthidx = MAXOPENED-2;
1464         static int lastmaxdepth = 0;
1465         if (val > lastmaxdepth) {
1466                 LYXERR(Debug::INFO, "Depth reached " << val);
1467                 lastmaxdepth = val;
1468         }
1469         if (val > maxdepthidx) {
1470                 maxdepthidx = val;
1471                 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1472         }
1473 }
1474
1475 #if 0
1476 // Not needed, because borders are now dynamically expanded
1477 static void checkIgnoreIdx(int val)
1478 {
1479         static int lastmaxignore = -1;
1480         if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1481                 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1482                 lastmaxignore = val;
1483         }
1484 }
1485 #endif
1486
1487 /*
1488  * Expand the region of ignored parts of the input latex string
1489  * The region is only relevant in output()
1490  */
1491 void Intervall::addIntervall(int low, int upper)
1492 {
1493         int idx;
1494         if (low == upper) return;
1495         for (idx = ignoreidx+1; idx > 0; --idx) {
1496                 if (low > borders[idx-1].upper) {
1497                         break;
1498                 }
1499         }
1500         Border br(low, upper);
1501         if (idx > ignoreidx) {
1502                 if (borders.size() <= size_t(idx)) {
1503                         borders.push_back(br);
1504                 }
1505                 else {
1506                         borders[idx] = br;
1507                 }
1508                 ignoreidx = idx;
1509                 // checkIgnoreIdx(ignoreidx);
1510                 return;
1511         }
1512         else {
1513                 // Expand only if one of the new bound is inside the interwall
1514                 // We know here that br.low > borders[idx-1].upper
1515                 if (br.upper < borders[idx].low) {
1516                         // We have to insert at this pos
1517                         if (size_t(ignoreidx+1) >= borders.size()) {
1518                                 borders.push_back(borders[ignoreidx]);
1519                         }
1520                         else {
1521                                 borders[ignoreidx+1] = borders[ignoreidx];
1522                         }
1523                         for (int i = ignoreidx; i > idx; --i) {
1524                                 borders[i] = borders[i-1];
1525                         }
1526                         borders[idx] = br;
1527                         ignoreidx += 1;
1528                         // checkIgnoreIdx(ignoreidx);
1529                         return;
1530                 }
1531                 // Here we know, that we are overlapping
1532                 if (br.low > borders[idx].low)
1533                         br.low = borders[idx].low;
1534                 // check what has to be concatenated
1535                 int count = 0;
1536                 for (int i = idx; i <= ignoreidx; i++) {
1537                         if (br.upper >= borders[i].low) {
1538                                 count++;
1539                                 if (br.upper < borders[i].upper)
1540                                         br.upper = borders[i].upper;
1541                         }
1542                         else {
1543                                 break;
1544                         }
1545                 }
1546                 // count should be >= 1 here
1547                 borders[idx] = br;
1548                 if (count > 1) {
1549                         for (int i = idx + count; i <= ignoreidx; i++) {
1550                                 borders[i-count+1] = borders[i];
1551                         }
1552                         ignoreidx -= count - 1;
1553                         return;
1554                 }
1555         }
1556 }
1557
1558 static void buildaccent(string n, string param, string values)
1559 {
1560         stringstream s(n);
1561         string name;
1562         const char delim = '|';
1563         while (getline(s, name, delim)) {
1564                 size_t start = 0;
1565                 for (char c : param) {
1566                         string key = name + "{" + c + "}";
1567                         // get the corresponding utf8-value
1568                         if ((values[start] & 0xc0) != 0xc0) {
1569                                 // should not happen, utf8 encoding starts at least with 11xxxxxx
1570                                 // but value for '\dot{i}' is 'i', which is ascii
1571                                 if ((values[start] & 0x80) == 0) {
1572                                         // is ascii
1573                                         accents[key] = values.substr(start, 1);
1574                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1575                                 }
1576                                 start++;
1577                                 continue;
1578                         }
1579                         for (int j = 1; ;j++) {
1580                                 if (start + j >= values.size()) {
1581                                         accents[key] = values.substr(start, j);
1582                                         start = values.size() - 1;
1583                                         break;
1584                                 }
1585                                 else if ((values[start+j] & 0xc0) != 0x80) {
1586                                         // This is the first byte of following utf8 char
1587                                         accents[key] = values.substr(start, j);
1588                                         start += j;
1589                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1590                                         break;
1591                                 }
1592                         }
1593                 }
1594         }
1595 }
1596
1597 // Helper function
1598 static string getutf8(unsigned uchar)
1599 {
1600 #define maxc 5
1601         string ret = string();
1602         char c[maxc] = {0};
1603         if (uchar <= 0x7f) {
1604                 c[maxc-1] = uchar & 0x7f;
1605         }
1606         else {
1607                 unsigned char rest = 0x40;
1608                 unsigned char first = 0x80;
1609                 int start = maxc-1;
1610                 for (int i = start; i >=0; --i) {
1611                         if (uchar < rest) {
1612                                 c[i] = first + uchar;
1613                                 break;
1614                         }
1615                         c[i] = 0x80 | (uchar &  0x3f);
1616                         uchar >>= 6;
1617                         rest >>= 1;
1618                         first >>= 1;
1619                         first |= 0x80;
1620                 }
1621         }
1622         for (int i = 0; i < maxc; i++) {
1623                 if (c[i] == 0) continue;
1624                 ret += c[i];
1625         }
1626         return(ret);
1627 }
1628
1629 static void addAccents(string latex_in, string unicode_out)
1630 {
1631         latex_in = latex_in.substr(1);
1632         AccentsIterator it_ac = accents.find(latex_in);
1633         if (it_ac == accents.end()) {
1634                 accents[latex_in] = unicode_out;
1635         }
1636         else {
1637                 LYXERR0("Key " << latex_in  << " already set");
1638         }
1639 }
1640
1641 void static fillMissingUnicodesymbols()
1642 {
1643         addAccents("\\pounds", getutf8(0x00a3));
1644         addAccents("\\textsterling", getutf8(0x00a3));
1645         addAccents("\\textyen", getutf8(0x00a5));
1646         addAccents("\\yen", getutf8(0x00a5));
1647         addAccents("\\textsection", getutf8(0x00a7));
1648         addAccents("\\mathsection", getutf8(0x00a7));
1649         addAccents("\\textcopyright", getutf8(0x00a9));
1650         addAccents("\\copyright", getutf8(0x00a9));
1651         addAccents("\\textlnot", getutf8(0x00ac));
1652         addAccents("\\neg", getutf8(0x00ac));
1653         addAccents("\\textregistered", getutf8(0x00ae));
1654         addAccents("\\circledR", getutf8(0x00ae));
1655         addAccents("\\textpm", getutf8(0x00b1));
1656         addAccents("\\pm", getutf8(0x00b1));
1657         addAccents("\\textparagraph", getutf8(0x00b6));
1658         addAccents("\\mathparagraph", getutf8(0x00b6));
1659         addAccents("\\textperiodcentered", getutf8(0x00b7));
1660         addAccents("\\texttimes", getutf8(0x00d7));
1661         addAccents("\\times", getutf8(0x00d7));
1662         addAccents("\\O", getutf8(0x00d8));
1663         addAccents("\\dh", getutf8(0x00f0));
1664         addAccents("\\eth", getutf8(0x00f0));
1665         addAccents("\\textdiv", getutf8(0x00f7));
1666         addAccents("\\div", getutf8(0x00f7));
1667         addAccents("\\o", getutf8(0x00f8));
1668         addAccents("\\textcrlambda", getutf8(0x019b));
1669         addAccents("\\j", getutf8(0x0237));
1670         addAccents("\\textrevepsilon", getutf8(0x025c));
1671         addAccents("\\textbaru", getutf8(0x0289));
1672         addAccents("\\textquoteleft", getutf8(0x02bb));
1673         addAccents("\\textGamma", getutf8(0x0393));
1674         addAccents("\\Gamma", getutf8(0x0393));
1675         addAccents("\\textDelta", getutf8(0x0394));
1676         addAccents("\\Delta", getutf8(0x0394));
1677         addAccents("\\textTheta", getutf8(0x0398));
1678         addAccents("\\Theta", getutf8(0x0398));
1679         addAccents("\\textLambda", getutf8(0x039b));
1680         addAccents("\\Lambda", getutf8(0x039b));
1681         addAccents("\\textXi", getutf8(0x039e));
1682         addAccents("\\Xi", getutf8(0x039e));
1683         addAccents("\\textPi", getutf8(0x03a0));
1684         addAccents("\\Pi", getutf8(0x03a0));
1685         addAccents("\\textSigma", getutf8(0x03a3));
1686         addAccents("\\Sigma", getutf8(0x03a3));
1687         addAccents("\\textUpsilon", getutf8(0x03a5));
1688         addAccents("\\Upsilon", getutf8(0x03a5));
1689         addAccents("\\textPhi", getutf8(0x03a6));
1690         addAccents("\\Phi", getutf8(0x03a6));
1691         addAccents("\\textPsi", getutf8(0x03a8));
1692         addAccents("\\Psi", getutf8(0x03a8));
1693         addAccents("\\textOmega", getutf8(0x03a9));
1694         addAccents("\\Omega", getutf8(0x03a9));
1695         addAccents("\\textalpha", getutf8(0x03b1));
1696         addAccents("\\alpha", getutf8(0x03b1));
1697         addAccents("\\textbeta", getutf8(0x03b2));
1698         addAccents("\\beta", getutf8(0x03b2));
1699         addAccents("\\textgamma", getutf8(0x03b3));
1700         addAccents("\\gamma", getutf8(0x03b3));
1701         addAccents("\\textdelta", getutf8(0x03b4));
1702         addAccents("\\delta", getutf8(0x03b4));
1703         addAccents("\\textepsilon", getutf8(0x03b5));
1704         addAccents("\\varepsilon", getutf8(0x03b5));
1705         addAccents("\\textzeta", getutf8(0x03b6));
1706         addAccents("\\zeta", getutf8(0x03b6));
1707         addAccents("\\texteta", getutf8(0x03b7));
1708         addAccents("\\eta", getutf8(0x03b7));
1709         addAccents("\\texttheta", getutf8(0x03b8));
1710         addAccents("\\theta", getutf8(0x03b8));
1711         addAccents("\\textiota", getutf8(0x03b9));
1712         addAccents("\\iota", getutf8(0x03b9));
1713         addAccents("\\textkappa", getutf8(0x03ba));
1714         addAccents("\\kappa", getutf8(0x03ba));
1715         addAccents("\\textlambda", getutf8(0x03bb));
1716         addAccents("\\lambda", getutf8(0x03bb));
1717         addAccents("\\textmu", getutf8(0x03bc));
1718         addAccents("\\mu", getutf8(0x03bc));
1719         addAccents("\\textnu", getutf8(0x03bd));
1720         addAccents("\\nu", getutf8(0x03bd));
1721         addAccents("\\textxi", getutf8(0x03be));
1722         addAccents("\\xi", getutf8(0x03be));
1723         addAccents("\\textpi", getutf8(0x03c0));
1724         addAccents("\\pi", getutf8(0x03c0));
1725         addAccents("\\textrho", getutf8(0x03c1));
1726         addAccents("\\rho", getutf8(0x03c1));
1727         addAccents("\\textfinalsigma", getutf8(0x03c2));
1728         addAccents("\\varsigma", getutf8(0x03c2));
1729         addAccents("\\textsigma", getutf8(0x03c3));
1730         addAccents("\\sigma", getutf8(0x03c3));
1731         addAccents("\\texttau", getutf8(0x03c4));
1732         addAccents("\\tau", getutf8(0x03c4));
1733         addAccents("\\textupsilon", getutf8(0x03c5));
1734         addAccents("\\upsilon", getutf8(0x03c5));
1735         addAccents("\\textphi", getutf8(0x03c6));
1736         addAccents("\\varphi", getutf8(0x03c6));
1737         addAccents("\\textchi", getutf8(0x03c7));
1738         addAccents("\\chi", getutf8(0x03c7));
1739         addAccents("\\textpsi", getutf8(0x03c8));
1740         addAccents("\\psi", getutf8(0x03c8));
1741         addAccents("\\textomega", getutf8(0x03c9));
1742         addAccents("\\omega", getutf8(0x03c9));
1743         addAccents("\\textdigamma", getutf8(0x03dd));
1744         addAccents("\\digamma", getutf8(0x03dd));
1745         addAccents("\\hebalef", getutf8(0x05d0));
1746         addAccents("\\aleph", getutf8(0x05d0));
1747         addAccents("\\hebbet", getutf8(0x05d1));
1748         addAccents("\\beth", getutf8(0x05d1));
1749         addAccents("\\hebgimel", getutf8(0x05d2));
1750         addAccents("\\gimel", getutf8(0x05d2));
1751         addAccents("\\hebdalet", getutf8(0x05d3));
1752         addAccents("\\daleth", getutf8(0x05d3));
1753         addAccents("\\hebhe", getutf8(0x05d4));
1754         addAccents("\\hebvav", getutf8(0x05d5));
1755         addAccents("\\hebzayin", getutf8(0x05d6));
1756         addAccents("\\hebhet", getutf8(0x05d7));
1757         addAccents("\\hebtet", getutf8(0x05d8));
1758         addAccents("\\hebyod", getutf8(0x05d9));
1759         addAccents("\\hebfinalkaf", getutf8(0x05da));
1760         addAccents("\\hebkaf", getutf8(0x05db));
1761         addAccents("\\heblamed", getutf8(0x05dc));
1762         addAccents("\\hebfinalmem", getutf8(0x05dd));
1763         addAccents("\\hebmem", getutf8(0x05de));
1764         addAccents("\\hebfinalnun", getutf8(0x05df));
1765         addAccents("\\hebnun", getutf8(0x05e0));
1766         addAccents("\\hebsamekh", getutf8(0x05e1));
1767         addAccents("\\hebayin", getutf8(0x05e2));
1768         addAccents("\\hebfinalpe", getutf8(0x05e3));
1769         addAccents("\\hebpe", getutf8(0x05e4));
1770         addAccents("\\hebfinaltsadi", getutf8(0x05e5));
1771         addAccents("\\hebtsadi", getutf8(0x05e6));
1772         addAccents("\\hebqof", getutf8(0x05e7));
1773         addAccents("\\hebresh", getutf8(0x05e8));
1774         addAccents("\\hebshin", getutf8(0x05e9));
1775         addAccents("\\hebtav", getutf8(0x05ea));
1776
1777         // Thai characters
1778         addAccents("\\thaiKoKai", getutf8(0x0e01));
1779         addAccents("\\thaiKhoKhai", getutf8(0x0e02));
1780         addAccents("\\thaiKhoKhuat", getutf8(0x0e03));
1781         addAccents("\\thaiKhoKhwai", getutf8(0x0e04));
1782         addAccents("\\thaiKhoKhon", getutf8(0x0e05));
1783         addAccents("\\thaiKhoRakhang", getutf8(0x0e06));
1784         addAccents("\\thaiNgoNgu", getutf8(0x0e07));
1785         addAccents("\\thaiChoChan", getutf8(0x0e08));
1786         addAccents("\\thaiChoChing", getutf8(0x0e09));
1787         addAccents("\\thaiChoChang", getutf8(0x0e0a));
1788         addAccents("\\thaiSoSo", getutf8(0x0e0b));
1789         addAccents("\\thaiChoChoe", getutf8(0x0e0c));
1790         addAccents("\\thaiYoYing", getutf8(0x0e0d));
1791         addAccents("\\thaiDoChada", getutf8(0x0e0e));
1792         addAccents("\\thaiToPatak", getutf8(0x0e0f));
1793         addAccents("\\thaiThoThan", getutf8(0x0e10));
1794         addAccents("\\thaiThoNangmontho", getutf8(0x0e11));
1795         addAccents("\\thaiThoPhuthao", getutf8(0x0e12));
1796         addAccents("\\thaiNoNen", getutf8(0x0e13));
1797         addAccents("\\thaiDoDek", getutf8(0x0e14));
1798         addAccents("\\thaiToTao", getutf8(0x0e15));
1799         addAccents("\\thaiThoThung", getutf8(0x0e16));
1800         addAccents("\\thaiThoThahan", getutf8(0x0e17));
1801         addAccents("\\thaiThoThong", getutf8(0x0e18));
1802         addAccents("\\thaiNoNu", getutf8(0x0e19));
1803         addAccents("\\thaiBoBaimai", getutf8(0x0e1a));
1804         addAccents("\\thaiPoPla", getutf8(0x0e1b));
1805         addAccents("\\thaiPhoPhung", getutf8(0x0e1c));
1806         addAccents("\\thaiFoFa", getutf8(0x0e1d));
1807         addAccents("\\thaiPhoPhan", getutf8(0x0e1e));
1808         addAccents("\\thaiFoFan", getutf8(0x0e1f));
1809         addAccents("\\thaiPhoSamphao", getutf8(0x0e20));
1810         addAccents("\\thaiMoMa", getutf8(0x0e21));
1811         addAccents("\\thaiYoYak", getutf8(0x0e22));
1812         addAccents("\\thaiRoRua", getutf8(0x0e23));
1813         addAccents("\\thaiRu", getutf8(0x0e24));
1814         addAccents("\\thaiLoLing", getutf8(0x0e25));
1815         addAccents("\\thaiLu", getutf8(0x0e26));
1816         addAccents("\\thaiWoWaen", getutf8(0x0e27));
1817         addAccents("\\thaiSoSala", getutf8(0x0e28));
1818         addAccents("\\thaiSoRusi", getutf8(0x0e29));
1819         addAccents("\\thaiSoSua", getutf8(0x0e2a));
1820         addAccents("\\thaiHoHip", getutf8(0x0e2b));
1821         addAccents("\\thaiLoChula", getutf8(0x0e2c));
1822         addAccents("\\thaiOAng", getutf8(0x0e2d));
1823         addAccents("\\thaiHoNokhuk", getutf8(0x0e2e));
1824         addAccents("\\thaiPaiyannoi", getutf8(0x0e2f));
1825         addAccents("\\thaiSaraA", getutf8(0x0e30));
1826         addAccents("\\thaiMaiHanakat", getutf8(0x0e31));
1827         addAccents("\\thaiSaraAa", getutf8(0x0e32));
1828         addAccents("\\thaiSaraAm", getutf8(0x0e33));
1829         addAccents("\\thaiSaraI", getutf8(0x0e34));
1830         addAccents("\\thaiSaraIi", getutf8(0x0e35));
1831         addAccents("\\thaiSaraUe", getutf8(0x0e36));
1832         addAccents("\\thaiSaraUee", getutf8(0x0e37));
1833         addAccents("\\thaiSaraU", getutf8(0x0e38));
1834         addAccents("\\thaiSaraUu", getutf8(0x0e39));
1835         addAccents("\\thaiPhinthu", getutf8(0x0e3a));
1836         addAccents("\\thaiSaraE", getutf8(0x0e40));
1837         addAccents("\\thaiSaraAe", getutf8(0x0e41));
1838         addAccents("\\thaiSaraO", getutf8(0x0e42));
1839         addAccents("\\thaiSaraAiMaimuan", getutf8(0x0e43));
1840         addAccents("\\thaiSaraAiMaimalai", getutf8(0x0e44));
1841         addAccents("\\thaiLakkhangyao", getutf8(0x0e45));
1842         addAccents("\\thaiMaiyamok", getutf8(0x0e46));
1843         addAccents("\\thaiMaitaikhu", getutf8(0x0e47));
1844         addAccents("\\thaiMaiEk", getutf8(0x0e48));
1845         addAccents("\\thaiMaiTho", getutf8(0x0e49));
1846         addAccents("\\thaiMaiTri", getutf8(0x0e4a));
1847         addAccents("\\thaiMaiChattawa", getutf8(0x0e4b));
1848         addAccents("\\thaiThanthakhat", getutf8(0x0e4c));
1849         addAccents("\\thaiNikhahit", getutf8(0x0e4d));
1850         addAccents("\\thaiYamakkan", getutf8(0x0e4e));
1851         addAccents("\\thaiFongman", getutf8(0x0e4f));
1852         addAccents("\\thaizero", getutf8(0x0e50));
1853         addAccents("\\thaione", getutf8(0x0e51));
1854         addAccents("\\thaitwo", getutf8(0x0e52));
1855         addAccents("\\thaithree", getutf8(0x0e53));
1856         addAccents("\\thaifour", getutf8(0x0e54));
1857         addAccents("\\thaifive", getutf8(0x0e55));
1858         addAccents("\\thaisix", getutf8(0x0e56));
1859         addAccents("\\thaiseven", getutf8(0x0e57));
1860         addAccents("\\thaieight", getutf8(0x0e58));
1861         addAccents("\\thainine", getutf8(0x0e59));
1862         addAccents("\\thaiAngkhankhu", getutf8(0x0e5a));
1863         addAccents("\\thaiKhomut", getutf8(0x0e5b));
1864         addAccents("\\dag", getutf8(0x2020));
1865         addAccents("\\dagger", getutf8(0x2020));
1866         addAccents("\\textdagger", getutf8(0x2020));
1867         addAccents("\\ddag", getutf8(0x2021));
1868         addAccents("\\ddagger", getutf8(0x2021));
1869         addAccents("\\textdaggerdbl", getutf8(0x2021));
1870         addAccents("\\textbullet", getutf8(0x2022));
1871         addAccents("\\bullet", getutf8(0x2022));
1872         addAccents("\\dots", getutf8(0x2026));
1873         addAccents("\\ldots", getutf8(0x2026));
1874         addAccents("\\textellipsis", getutf8(0x2026));
1875         addAccents("\\textasciiacute", getutf8(0x2032));
1876         addAccents("\\prime", getutf8(0x2032));
1877         addAccents("\\textacutedbl", getutf8(0x2033));
1878         addAccents("\\dprime", getutf8(0x2033));
1879         addAccents("\\textasciigrave", getutf8(0x2035));
1880         addAccents("\\backprime", getutf8(0x2035));
1881         addAccents("\\textsubcircum{ }", getutf8(0x2038));
1882         addAccents("\\caretinsert", getutf8(0x2038));
1883         addAccents("\\textasteriskcentered", getutf8(0x204e));
1884         addAccents("\\ast", getutf8(0x204e));
1885         addAccents("\\textmho", getutf8(0x2127));
1886         addAccents("\\mho", getutf8(0x2127));
1887         addAccents("\\textleftarrow", getutf8(0x2190));
1888         addAccents("\\leftarrow", getutf8(0x2190));
1889         addAccents("\\textuparrow", getutf8(0x2191));
1890         addAccents("\\uparrow", getutf8(0x2191));
1891         addAccents("\\textrightarrow", getutf8(0x2192));
1892         addAccents("\\rightarrow", getutf8(0x2192));
1893         addAccents("\\textdownarrow", getutf8(0x2193));
1894         addAccents("\\downarrow", getutf8(0x2193));
1895         addAccents("\\textglobrise", getutf8(0x2197));
1896         addAccents("\\nearrow", getutf8(0x2197));
1897         addAccents("\\textglobfall", getutf8(0x2198));
1898         addAccents("\\searrow", getutf8(0x2198));
1899         addAccents("\\textsurd", getutf8(0x221a));
1900         addAccents("\\surd", getutf8(0x221a));
1901         addAccents("\\textbigcircle", getutf8(0x25ef));
1902         addAccents("\\bigcirc", getutf8(0x25ef));
1903         addAccents("\\FiveStar", getutf8(0x2605));
1904         addAccents("\\bigstar", getutf8(0x2605));
1905         addAccents("\\FiveStarOpen", getutf8(0x2606));
1906         addAccents("\\bigwhitestar", getutf8(0x2606));
1907         addAccents("\\Checkmark", getutf8(0x2713));
1908         addAccents("\\checkmark", getutf8(0x2713));
1909         addAccents("\\CrossMaltese", getutf8(0x2720));
1910         addAccents("\\maltese", getutf8(0x2720));
1911         addAccents("\\textlangle", getutf8(0x27e8));
1912         addAccents("\\langle", getutf8(0x27e8));
1913         addAccents("\\textrangle", getutf8(0x27e9));
1914         addAccents("\\rangle", getutf8(0x27e9));
1915 }
1916
1917 static void buildAccentsMap()
1918 {
1919         accents["imath"] = "ı";
1920         accents["i"] = "ı";
1921         accents["jmath"] = "ȷ";
1922         accents["cdot"] = "·";
1923         accents["textasciicircum"] = "^";
1924         accents["mathcircumflex"] = "^";
1925         accents["sim"] = "~";
1926         accents["guillemotright"] = "»";
1927         accents["guillemotleft"] = "«";
1928         accents["hairspace"]     = getutf8(0xf0000);    // select from free unicode plane 15
1929         accents["thinspace"]     = getutf8(0xf0002);    // and used _only_ by findadv
1930         accents["negthinspace"]  = getutf8(0xf0003);    // to omit backslashed latex macros
1931         accents["medspace"]      = getutf8(0xf0004);    // See https://en.wikipedia.org/wiki/Private_Use_Areas
1932         accents["negmedspace"]   = getutf8(0xf0005);
1933         accents["thickspace"]    = getutf8(0xf0006);
1934         accents["negthickspace"] = getutf8(0xf0007);
1935         accents["lyx"]           = getutf8(0xf0010);    // Used logos
1936         accents["LyX"]           = getutf8(0xf0010);
1937         accents["tex"]           = getutf8(0xf0011);
1938         accents["TeX"]           = getutf8(0xf0011);
1939         accents["latex"]         = getutf8(0xf0012);
1940         accents["LaTeX"]         = getutf8(0xf0012);
1941         accents["latexe"]        = getutf8(0xf0013);
1942         accents["LaTeXe"]        = getutf8(0xf0013);
1943         accents["lyxarrow"]      = getutf8(0xf0020);
1944         accents["braceleft"]     = getutf8(0xf0030);
1945         accents["braceright"]    = getutf8(0xf0031);
1946         accents["backslash lyx"]           = getutf8(0xf0010);  // Used logos inserted with starting \backslash
1947         accents["backslash LyX"]           = getutf8(0xf0010);
1948         accents["backslash tex"]           = getutf8(0xf0011);
1949         accents["backslash TeX"]           = getutf8(0xf0011);
1950         accents["backslash latex"]         = getutf8(0xf0012);
1951         accents["backslash LaTeX"]         = getutf8(0xf0012);
1952         accents["backslash latexe"]        = getutf8(0xf0013);
1953         accents["backslash LaTeXe"]        = getutf8(0xf0013);
1954         accents["backslash lyxarrow"]      = getutf8(0xf0020);
1955         accents["ddot{\\imath}"] = "ï";
1956         buildaccent("ddot", "aAeEhHiIoOtuUwWxXyY",
1957                     "äÄëËḧḦïÏöÖẗüÜẅẄẍẌÿŸ");   // umlaut
1958         buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1959                     "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ");      // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1960         accents["acute{\\imath}"] = "í";
1961         buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1962                     "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1963         buildaccent("dacute|H|h", "oOuU", "őŐűŰ");  // double acute
1964         buildaccent("mathring|r", "aAuUwy",
1965                     "åÅůŮẘẙ");  // ring
1966         accents["check{\\imath}"] = "ǐ";
1967         accents["check{\\jmath}"] = "ǰ";
1968         buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1969                     "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ");        // caron
1970         accents["hat{\\imath}"] = "î";
1971         accents["hat{\\jmath}"] = "ĵ";
1972         buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1973                     "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ");  // circ
1974         accents["bar{\\imath}"] = "ī";
1975         buildaccent("bar|=", "aAeEiIoOuUyY",
1976                     "āĀēĒīĪōŌūŪȳȲ");        // macron
1977         accents["tilde{\\imath}"] = "ĩ";
1978         buildaccent("tilde", "aAeEiInNoOuUvVyY",
1979                     "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ");  // tilde
1980         accents["breve{\\imath}"] = "ĭ";
1981         buildaccent("breve|u", "aAeEgGiIoOuU",
1982                     "ăĂĕĔğĞĭĬŏŎŭŬ");        // breve
1983         accents["grave{\\imath}"] = "ì";
1984         buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1985                     "àÀèÈìÌòÒùÙǹǸẁẀỳỲ");    // grave
1986         buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1987                     "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ");      // dot below
1988         buildaccent("ogonek|k", "AaEeIiUuOo",
1989                     "ĄąĘęĮįŲųǪǫ");    // ogonek
1990         buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1991                     "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ");        // cedilla
1992         buildaccent("subring|textsubring", "Aa",
1993                     "Ḁḁ");  // subring
1994         buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1995                     "ḒḓḘḙḼḽṊṋṰṱṶṷ");    // subcircum
1996         buildaccent("subtilde|textsubtilde", "EeIiUu",
1997                     "ḚḛḬḭṴṵ");      // subtilde
1998         accents["dgrave{\\imath}"] = "ȉ";
1999         accents["textdoublegrave{\\i}"] = "ȉ";
2000         buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
2001                     "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
2002         accents["rcap{\\imath}"] = "ȋ";
2003         accents["textroundcap{\\i}"] = "ȋ";
2004         buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
2005                     "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
2006         buildaccent("slashed", "oO",
2007                     "øØ"); // slashed
2008         fillMissingUnicodesymbols(); // Add some still not handled entries contained in 'unicodesynbols'
2009         // LYXERR0("Number of accents " << accents.size());
2010 }
2011
2012 /*
2013  * Created accents in math or regexp environment
2014  * are macros, but we need the utf8 equivalent
2015  */
2016 void Intervall::removeAccents()
2017 {
2018         if (accents.empty())
2019                 buildAccentsMap();
2020         static regex const accre("\\\\("
2021                                  "([\\S]|[A-Za-z]+)\\{[^\\{\\}]+\\}"
2022                                  "|("
2023                                  "(backslash ([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))"
2024                                  "|[A-Za-z]+"
2025                                  ")"
2026                                  "(?![a-zA-Z]))");
2027         smatch sub;
2028         for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
2029                 sub = *itacc;
2030                 string key = sub.str(1);
2031                 AccentsIterator it_ac = accents.find(key);
2032                 if (it_ac != accents.end()) {
2033                         string val = it_ac->second;
2034                         size_t pos = sub.position(size_t(0));
2035                         for (size_t i = 0; i < val.size(); i++) {
2036                                 par[pos+i] = val[i];
2037                         }
2038                         // Remove possibly following space too
2039                         if (par[pos+sub.str(0).size()] == ' ')
2040                                 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
2041                         else
2042                                 addIntervall(pos+val.size(), pos + sub.str(0).size());
2043                         for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
2044                                 // remove traces of any remaining chars
2045                                 par[i] = ' ';
2046                         }
2047                 }
2048                 else {
2049                         LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
2050                 }
2051         }
2052 }
2053
2054 void Intervall::handleOpenP(int i)
2055 {
2056         actualdeptindex++;
2057         depts[actualdeptindex] = i+1;
2058         closes[actualdeptindex] = -1;
2059         checkDepthIndex(actualdeptindex);
2060 }
2061
2062 void Intervall::handleCloseP(int i, bool closingAllowed)
2063 {
2064         if (actualdeptindex <= 0) {
2065                 if (! closingAllowed)
2066                         LYXERR(Debug::FINDVERBOSE, "Bad closing parenthesis in latex");  /* should not happen, but the latex input may be wrong */
2067                 // if we are at the very end
2068                 addIntervall(i, i+1);
2069         }
2070         else {
2071                 closes[actualdeptindex] = i+1;
2072                 actualdeptindex--;
2073         }
2074 }
2075
2076 void Intervall::resetOpenedP(int openPos)
2077 {
2078         // Used as initializer for foreignlanguage entry
2079         actualdeptindex = 1;
2080         depts[1] = openPos+1;
2081         closes[1] = -1;
2082 }
2083
2084 int Intervall::previousNotIgnored(int start) const
2085 {
2086         int idx = 0;                          /* int intervalls */
2087         for (idx = ignoreidx; idx >= 0; --idx) {
2088                 if (start > borders[idx].upper)
2089                         return start;
2090                 if (start >= borders[idx].low)
2091                         start = borders[idx].low-1;
2092         }
2093         return start;
2094 }
2095
2096 int Intervall::nextNotIgnored(int start) const
2097 {
2098         int idx = 0;                          /* int intervalls */
2099         for (idx = 0; idx <= ignoreidx; idx++) {
2100                 if (start < borders[idx].low)
2101                         return start;
2102                 if (start < borders[idx].upper)
2103                         start = borders[idx].upper;
2104         }
2105         return start;
2106 }
2107
2108 typedef unordered_map<string, KeyInfo> KeysMap;
2109 typedef unordered_map<string, KeyInfo>::const_iterator KeysIterator;
2110 typedef vector< KeyInfo> Entries;
2111 static KeysMap keys = unordered_map<string, KeyInfo>();
2112
2113 class LatexInfo {
2114 private:
2115         int entidx_;
2116         Entries entries_;
2117         Intervall interval_;
2118         void buildKeys(bool);
2119         void buildEntries(bool);
2120         void makeKey(const string &, KeyInfo, bool isPatternString);
2121         void processRegion(int start, int region_end); /*  remove {} parts */
2122         void removeHead(KeyInfo const &, int count=0);
2123
2124 public:
2125         LatexInfo(string const & par, bool isPatternString)
2126                 : entidx_(-1), interval_(isPatternString, par)
2127         {
2128                 buildKeys(isPatternString);
2129                 entries_ = vector<KeyInfo>();
2130                 buildEntries(isPatternString);
2131         }
2132         int getFirstKey() {
2133                 entidx_ = 0;
2134                 if (entries_.empty()) {
2135                         return -1;
2136                 }
2137                 if (entries_[0].keytype == KeyInfo::isTitle) {
2138                         interval_.hasTitle = true;
2139                         if (! entries_[0].disabled) {
2140                                 interval_.titleValue = entries_[0].head;
2141                         }
2142                         else {
2143                                 interval_.titleValue = "";
2144                         }
2145                         removeHead(entries_[0]);
2146                         if (entries_.size() > 1)
2147                                 return 1;
2148                         else
2149                                 return -1;
2150                 }
2151                 return 0;
2152         }
2153         int getNextKey() {
2154                 entidx_++;
2155                 if (int(entries_.size()) > entidx_) {
2156                         return entidx_;
2157                 }
2158                 else {
2159                         return -1;
2160                 }
2161         }
2162         bool setNextKey(int idx) {
2163                 if ((idx == entidx_) && (entidx_ >= 0)) {
2164                         entidx_--;
2165                         return true;
2166                 }
2167                 else
2168                         return false;
2169         }
2170         int find(int start, KeyInfo::KeyType keytype) const {
2171                 if (start < 0)
2172                         return -1;
2173                 int tmpIdx = start;
2174                 while (tmpIdx < int(entries_.size())) {
2175                         if (entries_[tmpIdx].keytype == keytype)
2176                                 return tmpIdx;
2177                         tmpIdx++;
2178                 }
2179                 return -1;
2180         }
2181         int process(ostringstream & os, KeyInfo const & actual);
2182         int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
2183         // string show(int lastpos) { return interval.show(lastpos);}
2184         int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);}
2185         KeyInfo &getKeyInfo(int keyinfo) {
2186                 static KeyInfo invalidInfo = KeyInfo();
2187                 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
2188                         return invalidInfo;
2189                 else
2190                         return entries_[keyinfo];
2191         }
2192         void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);}
2193         void addIntervall(int low, int up) { interval_.addIntervall(low, up); }
2194 };
2195
2196
2197 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
2198 {
2199         int skip = 0;
2200         int depth = 0;
2201         for (int i = start; i < end; i += 1 + skip) {
2202                 char c;
2203                 c = par[i];
2204                 skip = 0;
2205                 if (c == '\\') skip = 1;
2206                 else if (c == up) {
2207                         depth++;
2208                 }
2209                 else if (c == down) {
2210                         if (depth == 0) {
2211                                 repeat--;
2212                                 if ((repeat <= 0) || (par[i+1] != up))
2213                                         return i;
2214                         }
2215                         --depth;
2216                 }
2217         }
2218         return end;
2219 }
2220
2221 class MathInfo {
2222         class MathEntry {
2223         public:
2224                 string wait;
2225                 size_t mathEnd;
2226                 size_t mathpostfixsize;
2227                 size_t mathStart;
2228                 size_t mathprefixsize;
2229                 size_t mathSize;
2230         };
2231         size_t actualIdx_;
2232         vector<MathEntry> entries_;
2233 public:
2234         MathInfo() {
2235                 actualIdx_ = 0;
2236         }
2237         void insert(string const & wait, size_t start, size_t prefixsize, size_t end, size_t postfixsize) {
2238                 MathEntry m = MathEntry();
2239                 m.wait = wait;
2240                 m.mathStart = start;
2241                 m.mathprefixsize = prefixsize;
2242                 m.mathEnd = end + postfixsize;
2243                 m.mathpostfixsize = postfixsize;
2244                 m.mathSize = m.mathEnd - m.mathStart;
2245                 entries_.push_back(m);
2246         }
2247         bool empty() const { return entries_.empty(); }
2248         size_t getEndPos() const {
2249                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2250                         return 0;
2251                 }
2252                 return entries_[actualIdx_].mathEnd;
2253         }
2254         size_t getStartPos() const {
2255                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2256                         return 100000;                    /*  definitely enough? */
2257                 }
2258                 return entries_[actualIdx_].mathStart;
2259         }
2260         size_t getPrefixSize() const {
2261                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2262                         return 0;
2263                 }
2264                 return entries_[actualIdx_].mathprefixsize;
2265         }
2266         size_t getPostfixSize() const {
2267                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2268                         return 0;
2269                 }
2270                 return entries_[actualIdx_].mathpostfixsize;
2271         }
2272         size_t getFirstPos() {
2273                 actualIdx_ = 0;
2274                 return getStartPos();
2275         }
2276         size_t getSize() const {
2277                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2278                         return size_t(0);
2279                 }
2280                 return entries_[actualIdx_].mathSize;
2281         }
2282         void incrEntry() { actualIdx_++; }
2283 };
2284
2285 void LatexInfo::buildEntries(bool isPatternString)
2286 {
2287         static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|x?x?alignat)\\*?\\})(\\{[0-9]+\\})?)");
2288         static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
2289         static bool disableLanguageOverride = false;
2290         smatch sub, submath;
2291         bool evaluatingRegexp = false;
2292         MathInfo mi;
2293         bool evaluatingMath = false;
2294         bool evaluatingCode = false;
2295         size_t codeEnd = 0;
2296         bool evaluatingOptional = false;
2297         size_t optionalEnd = 0;
2298         int codeStart = -1;
2299         KeyInfo found;
2300         bool math_end_waiting = false;
2301         size_t math_pos = 10000;
2302         size_t math_prefix_size = 1;
2303         string math_end;
2304         static vector<string> usedText = vector<string>();
2305         static bool removeMathHull = false;
2306
2307         interval_.removeAccents();
2308
2309         for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
2310                 submath = *itmath;
2311                 if ((submath.position(2) - submath.position(0)) %2 == 1) {
2312                         // prefixed by odd count of '\\'
2313                         continue;
2314                 }
2315                 if (math_end_waiting) {
2316                         size_t pos = submath.position(size_t(2));
2317                         if ((math_end == "$") &&
2318                                         (submath.str(2) == "$")) {
2319                                 mi.insert("$", math_pos, 1, pos, 1);
2320                                 math_end_waiting = false;
2321                         }
2322                         else if ((math_end == "\\]") &&
2323                                  (submath.str(2) == "\\]")) {
2324                                 mi.insert("\\]", math_pos, 2, pos, 2);
2325                                 math_end_waiting = false;
2326                         }
2327                         else if ((submath.str(3).compare("end") == 0) &&
2328                                  (submath.str(5).compare(math_end) == 0)) {
2329                                 mi.insert(math_end, math_pos, math_prefix_size, pos, submath.str(2).length());
2330                                 math_end_waiting = false;
2331                         }
2332                         else
2333                                 continue;
2334                 }
2335                 else {
2336                         if (submath.str(3).compare("begin") == 0) {
2337                                 math_end_waiting = true;
2338                                 math_end = submath.str(5);
2339                                 math_pos = submath.position(size_t(2));
2340                                 math_prefix_size = submath.str(2).length();
2341                         }
2342                         else if (submath.str(2).compare("\\[") == 0) {
2343                                 math_end_waiting = true;
2344                                 math_end = "\\]";
2345                                 math_pos = submath.position(size_t(2));
2346                         }
2347                         else if (submath.str(2) == "$") {
2348                                 size_t pos = submath.position(size_t(2));
2349                                 math_end_waiting = true;
2350                                 math_end = "$";
2351                                 math_pos = pos;
2352                         }
2353                 }
2354         }
2355         // Ignore language if there is math somewhere in pattern-string
2356         if (isPatternString) {
2357                 for (auto s: usedText) {
2358                         // Remove entries created in previous search runs
2359                         keys.erase(s);
2360                 }
2361                 usedText = vector<string>();
2362                 if (! mi.empty()) {
2363                         // Disable language
2364                         keys["foreignlanguage"].disabled = true;
2365                         disableLanguageOverride = true;
2366                         removeMathHull = false;
2367                 }
2368                 else {
2369                         removeMathHull = true;  // used later if not isPatternString
2370                         disableLanguageOverride = false;
2371                 }
2372         }
2373         else {
2374                 if (disableLanguageOverride) {
2375                         keys["foreignlanguage"].disabled = true;
2376                 }
2377         }
2378         math_pos = mi.getFirstPos();
2379         for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
2380                 sub = *it;
2381                 if ((sub.position(2) - sub.position(0)) %2 == 1) {
2382                         // prefixed by odd count of '\\'
2383                         continue;
2384                 }
2385                 string key = sub.str(5);
2386                 if (key == "") {
2387                         if (sub.str(2)[0] == '\\')
2388                                 key = sub.str(2)[1];
2389                         else {
2390                                 key = sub.str(2);
2391                         }
2392                 }
2393                 KeysIterator it_key = keys.find(key);
2394                 if (it_key != keys.end()) {
2395                         if (it_key->second.keytype == KeyInfo::headRemove) {
2396                                 KeyInfo found1 = it_key->second;
2397                                 found1.disabled = true;
2398                                 found1.head = "\\" + key + "{";
2399                                 found1._tokenstart = sub.position(size_t(2));
2400                                 found1._tokensize = found1.head.length();
2401                                 found1._dataStart = found1._tokenstart + found1.head.length();
2402                                 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
2403                                 found1._dataEnd = endpos;
2404                                 removeHead(found1);
2405                                 continue;
2406                         }
2407                 }
2408                 if (evaluatingRegexp) {
2409                         if (sub.str(3).compare("endregexp") == 0) {
2410                                 evaluatingRegexp = false;
2411                                 // found._tokenstart already set
2412                                 found._dataEnd = sub.position(size_t(2)) + 13;
2413                                 found._dataStart = found._dataEnd;
2414                                 found._tokensize = found._dataEnd - found._tokenstart;
2415                                 found.parenthesiscount = 0;
2416                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2417                         }
2418                         else {
2419                                 continue;
2420                         }
2421                 }
2422                 else {
2423                         if (evaluatingMath) {
2424                                 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
2425                                         continue;
2426                                 evaluatingMath = false;
2427                                 mi.incrEntry();
2428                                 math_pos = mi.getStartPos();
2429                         }
2430                         if (it_key == keys.end()) {
2431                                 found = KeyInfo(KeyInfo::isStandard, 0, true);
2432                                 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
2433                                 found = KeyInfo(KeyInfo::isText, 0, false);
2434                                 if (isPatternString) {
2435                                         found.keytype = KeyInfo::isChar;
2436                                         found.disabled = false;
2437                                         found.used = true;
2438                                 }
2439                                 keys[key] = found;
2440                                 usedText.push_back(key);
2441                         }
2442                         else
2443                                 found = keys[key];
2444                         if (key.compare("regexp") == 0) {
2445                                 evaluatingRegexp = true;
2446                                 found._tokenstart = sub.position(size_t(2));
2447                                 found._tokensize = 0;
2448                                 continue;
2449                         }
2450                 }
2451                 // Handle the other params of key
2452                 if (found.keytype == KeyInfo::isIgnored)
2453                         continue;
2454                 else if (found.keytype == KeyInfo::isMath) {
2455                         if (size_t(sub.position(size_t(2))) == math_pos) {
2456                                 found = keys[key];
2457                                 found._tokenstart = sub.position(size_t(2));
2458                                 found._tokensize = mi.getSize();
2459                                 found._dataEnd = found._tokenstart + found._tokensize;
2460                                 found._dataStart = found._dataEnd;
2461                                 found.parenthesiscount = 0;
2462                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2463                                 if (removeMathHull) {
2464                                         interval_.addIntervall(found._tokenstart, found._tokenstart + mi.getPrefixSize());
2465                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize(), found._dataEnd);
2466                                 }
2467                                 else {
2468                                         // Treate all math constructs as simple math
2469                                         interval_.par[found._tokenstart] = '$';
2470                                         interval_.par[found._dataEnd - mi.getPostfixSize()] = '$';
2471                                         interval_.addIntervall(found._tokenstart + 1, found._tokenstart + mi.getPrefixSize());
2472                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize() + 1, found._dataEnd);
2473                                 }
2474                                 evaluatingMath = true;
2475                         }
2476                         else {
2477                                 // begin|end of unknown env, discard
2478                                 // First handle tables
2479                                 // longtable|tabular
2480                                 bool discardComment;
2481                                 found = keys[key];
2482                                 found.keytype = KeyInfo::doRemove;
2483                                 if ((sub.str(7).compare("longtable") == 0) ||
2484                                                 (sub.str(7).compare("tabular") == 0)) {
2485                                         discardComment = true;        /* '%' */
2486                                 }
2487                                 else {
2488                                         discardComment = false;
2489                                         static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
2490                                         smatch sub2;
2491                                         string token = sub.str(7);
2492                                         if (regex_match(token, sub2, removeArgs)) {
2493                                                 found.keytype = KeyInfo::removeWithArg;
2494                                         }
2495                                 }
2496                                 // discard spaces before pos(2)
2497                                 int pos = sub.position(size_t(2));
2498                                 int count;
2499                                 for (count = 0; pos - count > 0; count++) {
2500                                         char c = interval_.par[pos-count-1];
2501                                         if (discardComment) {
2502                                                 if ((c != ' ') && (c != '%'))
2503                                                         break;
2504                                         }
2505                                         else if (c != ' ')
2506                                                 break;
2507                                 }
2508                                 found._tokenstart = pos - count;
2509                                 if (sub.str(3).compare(0, 5, "begin") == 0) {
2510                                         size_t pos1 = pos + sub.str(2).length();
2511                                         if (sub.str(7).compare("cjk") == 0) {
2512                                                 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2513                                                 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2514                                                         pos1 += 2;
2515                                                 found.keytype = KeyInfo::isMain;
2516                                                 found._dataStart = pos1;
2517                                                 found._dataEnd = interval_.par.length();
2518                                                 found.disabled = keys["foreignlanguage"].disabled;
2519                                                 found.used = keys["foreignlanguage"].used;
2520                                                 found._tokensize = pos1 - found._tokenstart;
2521                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2522                                         }
2523                                         else {
2524                                                 // Swallow possible optional params
2525                                                 while (interval_.par[pos1] == '[') {
2526                                                         pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2527                                                 }
2528                                                 // Swallow also the eventual parameter
2529                                                 if (interval_.par[pos1] == '{') {
2530                                                         found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2531                                                 }
2532                                                 else {
2533                                                         found._dataEnd = pos1;
2534                                                 }
2535                                                 found._dataStart = found._dataEnd;
2536                                                 found._tokensize = count + found._dataEnd - pos;
2537                                                 found.parenthesiscount = 0;
2538                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2539                                                 found.disabled = true;
2540                                         }
2541                                 }
2542                                 else {
2543                                         // Handle "\end{...}"
2544                                         found._dataStart = pos + sub.str(2).length();
2545                                         found._dataEnd = found._dataStart;
2546                                         found._tokensize = count + found._dataEnd - pos;
2547                                         found.parenthesiscount = 0;
2548                                         found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2549                                         found.disabled = true;
2550                                 }
2551                         }
2552                 }
2553                 else if (found.keytype != KeyInfo::isRegex) {
2554                         found._tokenstart = sub.position(size_t(2));
2555                         if (found.parenthesiscount == 0) {
2556                                 // Probably to be discarded
2557                                 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2558                                 char following = interval_.par[following_pos];
2559                                 if (following == ' ')
2560                                         found.head = "\\" + sub.str(5) + " ";
2561                                 else if (following == '=') {
2562                                         // like \uldepth=1000pt
2563                                         found.head = sub.str(2);
2564                                 }
2565                                 else
2566                                         found.head = "\\" + key;
2567                                 found._tokensize = found.head.length();
2568                                 found._dataEnd = found._tokenstart + found._tokensize;
2569                                 found._dataStart = found._dataEnd;
2570                         }
2571                         else {
2572                                 int params = found._tokenstart + key.length() + 1;
2573                                 if (evaluatingOptional) {
2574                                         if (size_t(found._tokenstart) > optionalEnd) {
2575                                                 evaluatingOptional = false;
2576                                         }
2577                                         else {
2578                                                 found.disabled = true;
2579                                         }
2580                                 }
2581                                 int optend = params;
2582                                 while (interval_.par[optend] == '[') {
2583                                         // discard optional parameters
2584                                         optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2585                                 }
2586                                 if (optend > params) {
2587                                         key += interval_.par.substr(params, optend-params);
2588                                         evaluatingOptional = true;
2589                                         optionalEnd = optend;
2590                                         if (found.keytype == KeyInfo::isSectioning) {
2591                                                 // Remove optional values (but still keep in header)
2592                                                 interval_.addIntervall(params, optend);
2593                                         }
2594                                 }
2595                                 string token = sub.str(7);
2596                                 int closings;
2597                                 if (interval_.par[optend] != '{') {
2598                                         closings = 0;
2599                                         found.parenthesiscount = 0;
2600                                         found.head = "\\" + key;
2601                                 }
2602                                 else
2603                                         closings = found.parenthesiscount;
2604                                 if (found.parenthesiscount == 1) {
2605                                         found.head = "\\" + key + "{";
2606                                 }
2607                                 else if (found.parenthesiscount > 1) {
2608                                         if (token != "") {
2609                                                 found.head = sub.str(2) + "{";
2610                                                 closings = found.parenthesiscount - 1;
2611                                         }
2612                                         else {
2613                                                 found.head = "\\" + key + "{";
2614                                         }
2615                                 }
2616                                 found._tokensize = found.head.length();
2617                                 found._dataStart = found._tokenstart + found.head.length();
2618                                 if (found.keytype == KeyInfo::doRemove) {
2619                                         if (closings > 0) {
2620                                                 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2621                                                 if (endpar >= interval_.par.length())
2622                                                         found._dataStart = interval_.par.length();
2623                                                 else
2624                                                         found._dataStart = endpar;
2625                                                 found._tokensize = found._dataStart - found._tokenstart;
2626                                         }
2627                                         else {
2628                                                 found._dataStart = found._tokenstart + found._tokensize;
2629                                         }
2630                                         closings = 0;
2631                                 }
2632                                 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2633                                         found._dataStart += 15;
2634                                 }
2635                                 size_t endpos;
2636                                 if (closings < 1)
2637                                         endpos = found._dataStart - 1;
2638                                 else
2639                                         endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2640                                 if (found.keytype == KeyInfo::isList) {
2641                                         // Check if it really is list env
2642                                         static regex const listre("^([a-z]+)$");
2643                                         smatch sub2;
2644                                         if (!regex_match(token, sub2, listre)) {
2645                                                 // Change the key of this entry. It is not in a list/item environment
2646                                                 found.keytype = KeyInfo::endArguments;
2647                                         }
2648                                 }
2649                                 if (found.keytype == KeyInfo::noMain) {
2650                                         evaluatingCode = true;
2651                                         codeEnd = endpos;
2652                                         codeStart = found._dataStart;
2653                                 }
2654                                 else if (evaluatingCode) {
2655                                         if (size_t(found._dataStart) > codeEnd)
2656                                                 evaluatingCode = false;
2657                                         else if (found.keytype == KeyInfo::isMain) {
2658                                                 // Disable this key, treate it as standard
2659                                                 found.keytype = KeyInfo::isStandard;
2660                                                 found.disabled = true;
2661                                                 if ((codeEnd +1 >= interval_.par.length()) &&
2662                                                                 (found._tokenstart == codeStart)) {
2663                                                         // trickery, because the code inset starts
2664                                                         // with \selectlanguage ...
2665                                                         codeEnd = endpos;
2666                                                         if (entries_.size() > 1) {
2667                                                                 entries_[entries_.size()-1]._dataEnd = codeEnd;
2668                                                         }
2669                                                 }
2670                                         }
2671                                 }
2672                                 if ((endpos == interval_.par.length()) &&
2673                                                 (found.keytype == KeyInfo::doRemove)) {
2674                                         // Missing closing => error in latex-input?
2675                                         // therefore do not delete remaining data
2676                                         found._dataStart -= 1;
2677                                         found._dataEnd = found._dataStart;
2678                                 }
2679                                 else
2680                                         found._dataEnd = endpos;
2681                         }
2682                         if (isPatternString) {
2683                                 keys[key].used = true;
2684                         }
2685                 }
2686                 entries_.push_back(found);
2687         }
2688 }
2689
2690 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2691 {
2692         stringstream s(keysstring);
2693         string key;
2694         const char delim = '|';
2695         while (getline(s, key, delim)) {
2696                 KeyInfo keyII(keyI);
2697                 if (isPatternString) {
2698                         keyII.used = false;
2699                 }
2700                 else if ( !keys[key].used)
2701                         keyII.disabled = true;
2702                 keys[key] = keyII;
2703         }
2704 }
2705
2706 void LatexInfo::buildKeys(bool isPatternString)
2707 {
2708
2709         static bool keysBuilt = false;
2710         if (keysBuilt && !isPatternString) return;
2711
2712         // Keys to ignore in any case
2713         makeKey("text|textcyrillic|lyxmathsym|ensuremath", KeyInfo(KeyInfo::headRemove, 1, true), true);
2714         // Known standard keys with 1 parameter.
2715         // Split is done, if not at start of region
2716         makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2717         makeKey("textbf",               KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2718         makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2719         makeKey("uuline|uline|uwave",   KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2720         makeKey("emph|noun",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2721         makeKey("sout|xout",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2722
2723         makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2724                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2725         makeKey("section*|subsection*|subsubsection*|paragraph*",
2726                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2727         makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2728         makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2729         // Regex
2730         makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2731
2732         // Split is done, if not at start of region
2733         makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2734         makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2735
2736         // Split is done always.
2737         makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2738
2739         // Known charaters
2740         // No split
2741         makeKey("backslash|textbackslash|slash",  KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2742         makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2743         makeKey("textasciiacute|texemdash",       KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2744         makeKey("dots|ldots",                     KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2745         // Spaces
2746         makeKey("quad|qquad|hfill|dotfill",               KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2747         makeKey("textvisiblespace|nobreakspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2748         makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2749         makeKey("thickspace|medspace|thinspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2750         // Skip
2751         // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2752         // Custom space/skip, remove the content (== length value)
2753         makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2754         // Found in fr/UserGuide.lyx
2755         makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2756         // quotes
2757         makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2758         makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2759         // Known macros to remove (including their parameter)
2760         // No split
2761         makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2762         makeKey("addtocounter|setlength",                 KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2763         // handle like standard keys with 1 parameter.
2764         makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2765
2766         if (ignoreFormats.getDeleted()) {
2767                 // Ignore deleted text
2768                 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2769         }
2770         else {
2771                 // but preserve added text
2772                 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2773         }
2774         makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2775
2776         // Macros to remove, but let the parameter survive
2777         // No split
2778         makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2779
2780         // Remove language spec from content of these insets
2781         makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2782
2783         // Same effect as previous, parameter will survive (because there is no one anyway)
2784         // No split
2785         makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2786         // Remove table decorations
2787         makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2788         // Discard shape-header.
2789         // For footnote or shortcut too, because of lang settings
2790         // and wrong handling if used 'KeyInfo::noMain'
2791         makeKey("circlepar|diamondpar|heartpar|nutpar",  KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2792         makeKey("trianglerightpar|hexagonpar|starpar",   KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2793         makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2794         makeKey("triangleleftpar|shapepar|dropuppar",    KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2795         makeKey("hphantom|vphantom|note|footnote|shortcut|include|includegraphics",     KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2796         makeKey("textgreek|textcyrillic", KeyInfo(KeyInfo::isStandard, 1, true), false);
2797         makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2798         // like ('tiny{}' or '\tiny ' ... )
2799         makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, ignoreFormats.getSize()), isPatternString);
2800
2801         // Survives, like known character
2802         // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2803         makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2804         makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2805
2806         makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2807         makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2808         makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2809
2810         makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2811         // Remove RTL/LTR marker
2812         makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2813         makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2814         makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2815         makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2816         makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2817         makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2818         makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2819         if (isPatternString) {
2820                 // Allow the first searched string to rebuild the keys too
2821                 keysBuilt = false;
2822         }
2823         else {
2824                 // no need to rebuild again
2825                 keysBuilt = true;
2826         }
2827 }
2828
2829 /*
2830  * Keep the list of actual opened parentheses actual
2831  * (e.g. depth == 4 means there are 4 '{' not processed yet)
2832  */
2833 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2834 {
2835         int skip = 0;
2836         for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2837                 char c;
2838                 c = par[i];
2839                 skip = 0;
2840                 if (c == '\\') skip = 1;
2841                 else if (c == '{') {
2842                         handleOpenP(i);
2843                 }
2844                 else if (c == '}') {
2845                         handleCloseP(i, closingAllowed);
2846                 }
2847         }
2848 }
2849
2850 #if (0)
2851 string Intervall::show(int lastpos)
2852 {
2853         int idx = 0;                          /* int intervalls */
2854         string s;
2855         int i = 0;
2856         if ((unsigned) lastpos > par.size())
2857                 lastpos = par.size();
2858         for (idx = 0; idx <= ignoreidx; idx++) {
2859                 while (i < lastpos) {
2860                         int printsize;
2861                         if (i <= borders[idx].low) {
2862                                 if (borders[idx].low > lastpos)
2863                                         printsize = lastpos - i;
2864                                 else
2865                                         printsize = borders[idx].low - i;
2866                                 s += par.substr(i, printsize);
2867                                 i += printsize;
2868                                 if (i >= borders[idx].low)
2869                                         i = borders[idx].upper;
2870                         }
2871                         else {
2872                                 i = borders[idx].upper;
2873                                 break;
2874                         }
2875                 }
2876         }
2877         if (lastpos > i) {
2878                 s += par.substr(i, lastpos-i);
2879         }
2880         return s;
2881 }
2882 #endif
2883
2884 void Intervall::output(ostringstream &os, int lastpos)
2885 {
2886         // get number of chars to output
2887         int idx = 0;                          /* int intervalls */
2888         int i = 0;
2889         int printed = 0;
2890         string startTitle = titleValue;
2891         for (idx = 0; idx <= ignoreidx; idx++) {
2892                 if (i < lastpos) {
2893                         if (i <= borders[idx].low) {
2894                                 int printsize;
2895                                 if (borders[idx].low > lastpos)
2896                                         printsize = lastpos - i;
2897                                 else
2898                                         printsize = borders[idx].low - i;
2899                                 if (printsize > 0) {
2900                                         os << startTitle << par.substr(i, printsize);
2901                                         i += printsize;
2902                                         printed += printsize;
2903                                         startTitle = "";
2904                                 }
2905                                 handleParentheses(i, false);
2906                                 if (i >= borders[idx].low)
2907                                         i = borders[idx].upper;
2908                         }
2909                         else {
2910                                 i = borders[idx].upper;
2911                         }
2912                 }
2913                 else
2914                         break;
2915         }
2916         if (lastpos > i) {
2917                 os << startTitle << par.substr(i, lastpos-i);
2918                 printed += lastpos-i;
2919         }
2920         handleParentheses(lastpos, false);
2921         int startindex;
2922         if (keys["foreignlanguage"].disabled)
2923                 startindex = actualdeptindex-langcount;
2924         else
2925                 startindex = actualdeptindex;
2926         for (int i = startindex; i > 0; --i) {
2927                 os << "}";
2928         }
2929         if (hasTitle && (printed > 0))
2930                 os << "}";
2931         if (! isPatternString_)
2932                 os << "\n";
2933         handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2934 }
2935
2936 void LatexInfo::processRegion(int start, int region_end)
2937 {
2938         while (start < region_end) {          /* Let {[} and {]} survive */
2939                 int cnt = interval_.isOpeningPar(start);
2940                 if (cnt == 1) {
2941                         // Closing is allowed past the region
2942                         int closing = interval_.findclosing(start+1, interval_.par.length());
2943                         interval_.addIntervall(start, start+1);
2944                         interval_.addIntervall(closing, closing+1);
2945                 }
2946                 else if (cnt == 3)
2947                         start += 2;
2948                 start = interval_.nextNotIgnored(start+1);
2949         }
2950 }
2951
2952 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2953 {
2954         if (actual.parenthesiscount == 0) {
2955                 // "{\tiny{} ...}" ==> "{{} ...}"
2956                 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2957         }
2958         else {
2959                 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2960                 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2961                 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2962         }
2963 }
2964
2965 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2966 {
2967         int nextKeyIdx = 0;
2968         switch (actual.keytype)
2969         {
2970         case KeyInfo::isTitle: {
2971                 removeHead(actual);
2972                 nextKeyIdx = getNextKey();
2973                 break;
2974         }
2975         case KeyInfo::cleanToStart: {
2976                 actual._dataEnd = actual._dataStart;
2977                 nextKeyIdx = getNextKey();
2978                 // Search for end of arguments
2979                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2980                 if (tmpIdx > 0) {
2981                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2982                                 entries_[i].disabled = true;
2983                         }
2984                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
2985                 }
2986                 while (interval_.par[actual._dataEnd] == ' ')
2987                         actual._dataEnd++;
2988                 interval_.addIntervall(0, actual._dataEnd+1);
2989                 interval_.actualdeptindex = 0;
2990                 interval_.depts[0] = actual._dataEnd+1;
2991                 interval_.closes[0] = -1;
2992                 break;
2993         }
2994         case KeyInfo::isText:
2995                 interval_.par[actual._tokenstart] = '#';
2996                 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2997                 nextKeyIdx = getNextKey();
2998                 break;
2999         case KeyInfo::noContent: {          /* char like "\hspace{2cm}" */
3000                 if (actual.disabled)
3001                         interval_.addIntervall(actual._tokenstart, actual._dataEnd);
3002                 else
3003                         interval_.addIntervall(actual._dataStart, actual._dataEnd);
3004         }
3005                 // fall through
3006         case KeyInfo::isChar: {
3007                 nextKeyIdx = getNextKey();
3008                 break;
3009         }
3010         case KeyInfo::isSize: {
3011                 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
3012                         if (actual.parenthesiscount == 0)
3013                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
3014                         else {
3015                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3016                         }
3017                         nextKeyIdx = getNextKey();
3018                 } else {
3019                         // Here _dataStart points to '{', so correct it
3020                         actual._dataStart += 1;
3021                         actual._tokensize += 1;
3022                         actual.parenthesiscount = 1;
3023                         if (interval_.par[actual._dataStart] == '}') {
3024                                 // Determine the end if used like '{\tiny{}...}'
3025                                 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
3026                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3027                         }
3028                         else {
3029                                 // Determine the end if used like '\tiny{...}'
3030                                 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
3031                         }
3032                         // Split on this key if not at start
3033                         int start = interval_.nextNotIgnored(previousStart);
3034                         if (start < actual._tokenstart) {
3035                                 interval_.output(os, actual._tokenstart);
3036                                 interval_.addIntervall(start, actual._tokenstart);
3037                         }
3038                         // discard entry if at end of actual
3039                         nextKeyIdx = process(os, actual);
3040                 }
3041                 break;
3042         }
3043         case KeyInfo::endArguments: {
3044                 // Remove trailing '{}' too
3045                 actual._dataStart += 1;
3046                 actual._dataEnd += 1;
3047                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3048                 nextKeyIdx = getNextKey();
3049                 break;
3050         }
3051         case KeyInfo::noMain:
3052                 // fall through
3053         case KeyInfo::isStandard: {
3054                 if (actual.disabled) {
3055                         removeHead(actual);
3056                         processRegion(actual._dataStart, actual._dataStart+1);
3057                         nextKeyIdx = getNextKey();
3058                 } else {
3059                         // Split on this key if not at datastart of calling entry
3060                         int start = interval_.nextNotIgnored(previousStart);
3061                         if (start < actual._tokenstart) {
3062                                 interval_.output(os, actual._tokenstart);
3063                                 interval_.addIntervall(start, actual._tokenstart);
3064                         }
3065                         // discard entry if at end of actual
3066                         nextKeyIdx = process(os, actual);
3067                 }
3068                 break;
3069         }
3070         case KeyInfo::removeWithArg: {
3071                 nextKeyIdx = getNextKey();
3072                 // Search for end of arguments
3073                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3074                 if (tmpIdx > 0) {
3075                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3076                                 entries_[i].disabled = true;
3077                         }
3078                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3079                 }
3080                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3081                 break;
3082         }
3083         case KeyInfo::doRemove: {
3084                 // Remove the key with all parameters and following spaces
3085                 size_t pos;
3086                 size_t start;
3087                 if (interval_.par[actual._dataEnd-1] == ' ' || interval_.par[actual._dataEnd-1] == '}')
3088                         start = actual._dataEnd;
3089                 else
3090                         start = actual._dataEnd+1;
3091                 for (pos = start; pos < interval_.par.length(); pos++) {
3092                         if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
3093                                 break;
3094                 }
3095                 // Remove also enclosing parentheses [] and {}
3096                 int numpars = 0;
3097                 int spaces = 0;
3098                 while (actual._tokenstart > numpars) {
3099                         if (pos+numpars >= interval_.par.size())
3100                                 break;
3101                         else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
3102                                 numpars++;
3103                         else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
3104                                 numpars++;
3105                         else
3106                                 break;
3107                 }
3108                 if (numpars > 0) {
3109                         if (interval_.par[pos+numpars] == ' ')
3110                                 spaces++;
3111                 }
3112
3113                 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
3114                 nextKeyIdx = getNextKey();
3115                 break;
3116         }
3117         case KeyInfo::isList: {
3118                 // Discard space before _tokenstart
3119                 int count;
3120                 for (count = 0; count < actual._tokenstart; count++) {
3121                         if (interval_.par[actual._tokenstart-count-1] != ' ')
3122                                 break;
3123                 }
3124                 nextKeyIdx = getNextKey();
3125                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3126                 if (tmpIdx > 0) {
3127                         // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
3128                         // with arguments
3129                         // How else can we catch this one?
3130                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3131                                 entries_[i].disabled = true;
3132                         }
3133                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3134                 }
3135                 else if (nextKeyIdx > 0) {
3136                         // Ignore any lang entries inside data region
3137                         for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
3138                                 if (entries_[i].keytype == KeyInfo::isMain)
3139                                         entries_[i].disabled = true;
3140                         }
3141                 }
3142                 if (actual.disabled) {
3143                         interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
3144                 }
3145                 else {
3146                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3147                 }
3148                 if (interval_.par[actual._dataEnd+1] == '[') {
3149                         int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
3150                         if ((interval_.par[actual._dataEnd+2] == '{') &&
3151                                         (interval_.par[posdown-1] == '}')) {
3152                                 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
3153                                 interval_.addIntervall(posdown-1, posdown+1);
3154                         }
3155                         else {
3156                                 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
3157                                 interval_.addIntervall(posdown, posdown+1);
3158                         }
3159                         int blk = interval_.nextNotIgnored(actual._dataEnd+1);
3160                         if (blk > posdown) {
3161                                 // Discard at most 1 space after empty item
3162                                 int count;
3163                                 for (count = 0; count < 1; count++) {
3164                                         if (interval_.par[blk+count] != ' ')
3165                                                 break;
3166                                 }
3167                                 if (count > 0)
3168                                         interval_.addIntervall(blk, blk+count);
3169                         }
3170                 }
3171                 break;
3172         }
3173         case KeyInfo::isSectioning: {
3174                 // Discard spaces before _tokenstart
3175                 int count;
3176                 int val = actual._tokenstart;
3177                 for (count = 0; count < actual._tokenstart;) {
3178                         val = interval_.previousNotIgnored(val-1);
3179                         if (val < 0 || interval_.par[val] != ' ')
3180                                 break;
3181                         else {
3182                                 count = actual._tokenstart - val;
3183                         }
3184                 }
3185                 if (actual.disabled) {
3186                         removeHead(actual, count);
3187                         nextKeyIdx = getNextKey();
3188                 } else {
3189                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3190                         nextKeyIdx = process(os, actual);
3191                 }
3192                 break;
3193         }
3194         case KeyInfo::isMath: {
3195                 // Same as regex, use the content unchanged
3196                 nextKeyIdx = getNextKey();
3197                 break;
3198         }
3199         case KeyInfo::isRegex: {
3200                 // DO NOT SPLIT ON REGEX
3201                 // Do not disable
3202                 nextKeyIdx = getNextKey();
3203                 break;
3204         }
3205         case KeyInfo::isIgnored: {
3206                 // Treat like a character for now
3207                 nextKeyIdx = getNextKey();
3208                 break;
3209         }
3210         case KeyInfo::isMain: {
3211                 if (interval_.par.substr(actual._dataStart, 2) == "% ")
3212                         interval_.addIntervall(actual._dataStart, actual._dataStart+2);
3213                 if (actual._tokenstart > 0) {
3214                         int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
3215                         if ((prev >= 0) && interval_.par[prev] == '%')
3216                                 interval_.addIntervall(prev, prev+1);
3217                 }
3218                 if (actual.disabled) {
3219                         removeHead(actual);
3220                         interval_.langcount++;
3221                         if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
3222                                         (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
3223                                 // Discard also the space before math-equation
3224                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3225                         }
3226                         nextKeyIdx = getNextKey();
3227                         // interval.resetOpenedP(actual._dataStart-1);
3228                 }
3229                 else {
3230                         if (actual._tokenstart < 26) {
3231                                 // for the first (and maybe dummy) language
3232                                 interval_.setForDefaultLang(actual);
3233                         }
3234                         interval_.resetOpenedP(actual._dataStart-1);
3235                 }
3236                 break;
3237         }
3238         case KeyInfo::invalid:
3239         case KeyInfo::headRemove:
3240                 // These two cases cannot happen, already handled
3241                 // fall through
3242         default: {
3243                 // LYXERR(Debug::INFO, "Unhandled keytype");
3244                 nextKeyIdx = getNextKey();
3245                 break;
3246         }
3247         }
3248         return nextKeyIdx;
3249 }
3250
3251 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
3252 {
3253         int end = interval_.nextNotIgnored(actual._dataEnd);
3254         int oldStart = actual._dataStart;
3255         int nextKeyIdx = getNextKey();
3256         while (true) {
3257                 if ((nextKeyIdx < 0) ||
3258                                 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
3259                                 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
3260                         if (oldStart <= end) {
3261                                 processRegion(oldStart, end);
3262                                 oldStart = end+1;
3263                         }
3264                         break;
3265                 }
3266                 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
3267
3268                 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
3269                         (void) dispatch(os, actual._dataStart, nextKey);
3270                         end = nextKey._tokenstart;
3271                         break;
3272                 }
3273                 processRegion(oldStart, nextKey._tokenstart);
3274                 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
3275
3276                 oldStart = nextKey._dataEnd+1;
3277         }
3278         // now nextKey is either invalid or is outside of actual._dataEnd
3279         // output the remaining and discard myself
3280         if (oldStart <= end) {
3281                 processRegion(oldStart, end);
3282         }
3283         if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
3284                 end += 1;
3285                 // This is the normal case.
3286                 // But if using the firstlanguage, the closing may be missing
3287         }
3288         // get minimum of 'end' and  'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
3289         int output_end;
3290         if (actual._dataEnd < end)
3291                 output_end = interval_.nextNotIgnored(actual._dataEnd);
3292         else if (interval_.par.size() > (size_t) end)
3293                 output_end = interval_.nextNotIgnored(end);
3294         else
3295                 output_end = interval_.par.size();
3296         if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
3297                 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
3298         }
3299         // Remove possible empty data
3300         int dstart = interval_.nextNotIgnored(actual._dataStart);
3301         while (interval_.isOpeningPar(dstart) == 1) {
3302                 interval_.addIntervall(dstart, dstart+1);
3303                 int dend = interval_.findclosing(dstart+1, output_end);
3304                 interval_.addIntervall(dend, dend+1);
3305                 dstart = interval_.nextNotIgnored(dstart+1);
3306         }
3307         if (dstart < output_end)
3308                 interval_.output(os, output_end);
3309         if (nextKeyIdx < 0)
3310                 interval_.addIntervall(0, end);
3311         else
3312                 interval_.addIntervall(actual._tokenstart, end);
3313         return nextKeyIdx;
3314 }
3315
3316 string splitOnKnownMacros(string par, bool isPatternString)
3317 {
3318         ostringstream os;
3319         LatexInfo li(par, isPatternString);
3320         // LYXERR(Debug::INFO, "Berfore split: " << par);
3321         KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
3322         DummyKey.head = "";
3323         DummyKey._tokensize = 0;
3324         DummyKey._dataStart = 0;
3325         DummyKey._dataEnd = par.length();
3326         DummyKey.disabled = true;
3327         int firstkeyIdx = li.getFirstKey();
3328         string s;
3329         if (firstkeyIdx >= 0) {
3330                 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
3331                 DummyKey._tokenstart = firstKey._tokenstart;
3332                 int nextkeyIdx;
3333                 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
3334                         // Use dummy firstKey
3335                         firstKey = DummyKey;
3336                         (void) li.setNextKey(firstkeyIdx);
3337                 }
3338                 else {
3339                         if (par.substr(firstKey._dataStart, 2) == "% ")
3340                                 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
3341                 }
3342                 nextkeyIdx = li.process(os, firstKey);
3343                 while (nextkeyIdx >= 0) {
3344                         // Check for a possible gap between the last
3345                         // entry and this one
3346                         int datastart = li.nextNotIgnored(firstKey._dataStart);
3347                         KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
3348                         if ((nextKey._tokenstart > datastart)) {
3349                                 // Handle the gap
3350                                 firstKey._dataStart = datastart;
3351                                 firstKey._dataEnd = par.length();
3352                                 (void) li.setNextKey(nextkeyIdx);
3353                                 // Fake the last opened parenthesis
3354                                 li.setForDefaultLang(firstKey);
3355                                 nextkeyIdx = li.process(os, firstKey);
3356                         }
3357                         else {
3358                                 if (nextKey.keytype != KeyInfo::isMain) {
3359                                         firstKey._dataStart = datastart;
3360                                         firstKey._dataEnd = nextKey._dataEnd+1;
3361                                         (void) li.setNextKey(nextkeyIdx);
3362                                         li.setForDefaultLang(firstKey);
3363                                         nextkeyIdx = li.process(os, firstKey);
3364                                 }
3365                                 else {
3366                                         nextkeyIdx = li.process(os, nextKey);
3367                                 }
3368                         }
3369                 }
3370                 // Handle the remaining
3371                 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
3372                 firstKey._dataEnd = par.length();
3373                 // Check if ! empty
3374                 if ((firstKey._dataStart < firstKey._dataEnd) &&
3375                                 (par[firstKey._dataStart] != '}')) {
3376                         li.setForDefaultLang(firstKey);
3377                         (void) li.process(os, firstKey);
3378                 }
3379                 s = os.str();
3380                 // return string definitelly impossible to match, but should be known
3381         }
3382         else
3383                 s = par;                            /* no known macros found */
3384         // LYXERR(Debug::INFO, "After split: " << s);
3385         return s;
3386 }
3387
3388 /*
3389  * Try to unify the language specs in the latexified text.
3390  * Resulting modified string is set to "", if
3391  * the searched tex does not contain all the features in the search pattern
3392  */
3393 static string correctlanguagesetting(string par, bool isPatternString, bool withformat, lyx::Buffer *pbuf = nullptr)
3394 {
3395         static Features regex_f;
3396         static int missed = 0;
3397         static bool regex_with_format = false;
3398
3399         int parlen = par.length();
3400
3401         while ((parlen > 0) && (par[parlen-1] == '\n')) {
3402                 parlen--;
3403         }
3404 #if 0
3405         if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
3406                 // Happens to be there in case of description or labeling environment
3407                 parlen--;
3408         }
3409 #endif
3410         string result;
3411         if (withformat) {
3412                 // Split the latex input into pieces which
3413                 // can be digested by our search engine
3414                 LYXERR(Debug::FINDVERBOSE, "input: \"" << par << "\"");
3415                 if (isPatternString && (pbuf != nullptr)) { // Check if we should disable/enable test for language
3416                         // We check for polyglossia, because in runparams.flavor we use Flavor::XeTeX
3417                         string doclang = pbuf->params().language->polyglossia();
3418                         static regex langre("\\\\(foreignlanguage)\\{([^\\}]+)\\}");
3419                         smatch sub;
3420                         bool toIgnoreLang = true;
3421                         for (sregex_iterator it(par.begin(), par.end(), langre), end; it != end; ++it) {
3422                                 sub = *it;
3423                                 if (sub.str(2) != doclang) {
3424                                         toIgnoreLang = false;
3425                                         break;
3426                                 }
3427                         }
3428                         setIgnoreFormat("language", toIgnoreLang, false);
3429
3430                 }
3431                 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
3432                 LYXERR(Debug::FINDVERBOSE, "After splitOnKnownMacros:\n\"" << result << "\"");
3433         }
3434         else
3435                 result = par.substr(0, parlen);
3436         if (isPatternString) {
3437                 missed = 0;
3438                 if (withformat) {
3439                         regex_f = identifyFeatures(result);
3440                         string features = "";
3441                         for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3442                                 string a = it->first;
3443                                 regex_with_format = true;
3444                                 features += " " + a;
3445                                 // LYXERR(Debug::INFO, "Identified regex format:" << a);
3446                         }
3447                         LYXERR(Debug::FINDVERBOSE, "Identified Features" << features);
3448
3449                 }
3450         } else if (regex_with_format) {
3451                 Features info = identifyFeatures(result);
3452                 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3453                         string a = it->first;
3454                         bool b = it->second;
3455                         if (b && ! info[a]) {
3456                                 missed++;
3457                                 LYXERR(Debug::FINDVERBOSE, "Missed(" << missed << " " << a <<", srclen = " << parlen );
3458                                 return "";
3459                         }
3460                 }
3461
3462         }
3463         else {
3464                 // LYXERR(Debug::INFO, "No regex formats");
3465         }
3466         return result;
3467 }
3468
3469
3470 // Remove trailing closure of math, macros and environments, so to catch parts of them.
3471 static void identifyClosing(string & t, bool ignoreformat)
3472 {
3473         do {
3474                 LYXERR(Debug::FINDVERBOSE, "identifyClosing(): t now is '" << t << "'");
3475                 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
3476                         continue;
3477                 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
3478                         continue;
3479                 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]+\\*?\\}$", "$1"))
3480                         continue;
3481                 if (! ignoreformat) {
3482                         if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1"))
3483                                 continue;
3484                 }
3485                 break;
3486         } while (true);
3487         return;
3488 }
3489
3490 static int num_replaced = 0;
3491 static bool previous_single_replace = true;
3492
3493 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
3494 {
3495 #if QTSEARCH
3496         if (regexp_str.empty() || regexp2_str.empty()) {
3497                 regexIsValid = false;
3498                 regexError = "Invalid empty regex";
3499                 return;
3500         }
3501         // Handle \w properly
3502         QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3503         if (! opt.casesensitive) {
3504                 popts |= QRegularExpression::CaseInsensitiveOption;
3505         }
3506         regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3507         regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3508         regexError = "";
3509         if (regexp.isValid() && regexp2.isValid()) {
3510                 regexIsValid = true;
3511                 // Check '{', '}' pairs inside the regex
3512                 int balanced = 0;
3513                 int skip = 1;
3514                 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3515                         char c = par_as_string[i];
3516                         if (c == '\\') {
3517                                 skip = 2;
3518                                 continue;
3519                         }
3520                         if (c == '{')
3521                                 balanced++;
3522                         else if (c == '}') {
3523                                 balanced--;
3524                                 if (balanced < 0)
3525                                         break;
3526                         }
3527                         skip = 1;
3528                 }
3529                 if (balanced != 0) {
3530                         regexIsValid = false;
3531                         regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3532                 }
3533         }
3534         else {
3535                 regexIsValid = false;
3536                 if (!regexp.isValid())
3537                         regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3538                 else
3539                         regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3540         }
3541 #else
3542         (void)par_as_string;
3543         if (opt.casesensitive) {
3544                 regexp = regex(regexp_str);
3545                 regexp2 = regex(regexp2_str);
3546         }
3547         else {
3548                 regexp = regex(regexp_str, std::regex_constants::icase);
3549                 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3550         }
3551 #endif
3552 }
3553
3554 static void modifyRegexForMatchWord(string &t)
3555 {
3556         string s("");
3557         regex wordre("(\\\\)*((\\.|\\\\b))");
3558         size_t lastpos = 0;
3559         smatch sub;
3560         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3561                 sub = *it;
3562                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3563                         continue;
3564                 }
3565                 else if (sub.str(2) == "\\\\b")
3566                         return;
3567                 if (lastpos < (size_t) sub.position(2))
3568                         s += t.substr(lastpos, sub.position(2) - lastpos);
3569                 s += "\\S";
3570                 lastpos = sub.position(2) + sub.length(2);
3571         }
3572         if (lastpos == 0) {
3573                 s = "\\b" + t + "\\b";
3574                 t = s;
3575                 return;
3576         }
3577         else if (lastpos < t.length())
3578                 s += t.substr(lastpos, t.length() - lastpos);
3579         t = "\\b" + s + "\\b";
3580 }
3581
3582 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3583         : p_buf(&buf), p_first_buf(&buf), opt(opt)
3584 {
3585         Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3586         docstring const & ds = stringifySearchBuffer(find_buf, opt);
3587         if (ds.empty() ) {
3588                 CreateRegexp(opt, "", "", "");
3589                 return;
3590         }
3591         use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3592         if (opt.replace_all && previous_single_replace) {
3593                 previous_single_replace = false;
3594                 num_replaced = 0;
3595         }
3596         else if (!opt.replace_all) {
3597                 num_replaced = 0;       // count number of replaced strings
3598                 previous_single_replace = true;
3599         }
3600         // When using regexp, braces are hacked already by escape_for_regex()
3601         par_as_string = normalize(ds, opt.ignoreformat);
3602         open_braces = 0;
3603         close_wildcards = 0;
3604
3605         size_t lead_size = 0;
3606         // correct the language settings
3607         par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat, &buf);
3608         if (par_as_string.empty()) {
3609                 CreateRegexp(opt, "", "", "");
3610                 return;
3611         }
3612         opt.matchAtStart = false;
3613         if (!use_regexp) {
3614                 identifyClosing(par_as_string, opt.ignoreformat); // Removes math closings ($, ], ...) at end of string
3615                 if (opt.ignoreformat) {
3616                         lead_size = 0;
3617                 }
3618                 else {
3619                         lead_size = identifyLeading(par_as_string);
3620                 }
3621                 lead_as_string = par_as_string.substr(0, lead_size);
3622                 string lead_as_regex_string = string2regex(lead_as_string);
3623                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3624                 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3625                 /* Handle whole words too in this case
3626                 */
3627                 if (opt.matchword) {
3628                         par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3629                         opt.matchword = false;
3630                 }
3631                 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3632                 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3633                 CreateRegexp(opt, regexp_str, regexp2_str);
3634                 use_regexp = true;
3635                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3636                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3637                 return;
3638         }
3639
3640         if (!opt.ignoreformat) {
3641                 lead_size = identifyLeading(par_as_string);
3642                 LYXERR(Debug::FINDVERBOSE, "Lead_size: " << lead_size);
3643                 lead_as_string = par_as_string.substr(0, lead_size);
3644                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3645         }
3646
3647         // Here we are using regexp
3648         LASSERT(use_regexp, /**/);
3649         {
3650                 string lead_as_regexp;
3651                 if (lead_size > 0) {
3652                         lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3653                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "\\$$", "");
3654                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3655                         par_as_string = par_as_string_nolead;
3656                         LYXERR(Debug::FINDVERBOSE, "lead_as_regexp is '" << lead_as_regexp << "'");
3657                         LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3658                 }
3659                 // LYXERR(Debug::FINDVERBOSE, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3660                 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3661                 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3662                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3663                 ++close_wildcards;
3664                 size_t lng = par_as_string.size();
3665                 if (!opt.ignoreformat) {
3666                         // Remove extra '\}' at end if not part of \{\.\}
3667                         while(lng > 2) {
3668                                 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3669                                         if (lng >= 6) {
3670                                                 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3671                                                         break;
3672                                         }
3673                                         lng -= 2;
3674                                         open_braces++;
3675                                 }
3676                                 else
3677                                         break;
3678                         }
3679                         if (lng < par_as_string.size())
3680                                 par_as_string = par_as_string.substr(0,lng);
3681                 }
3682                 LYXERR(Debug::FINDVERBOSE, "par_as_string after correctRegex is '" << par_as_string << "'");
3683                 if ((lng > 0) && (par_as_string[0] == '^')) {
3684                         par_as_string = par_as_string.substr(1);
3685                         --lng;
3686                         opt.matchAtStart = true;
3687                 }
3688                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3689                 // LYXERR(Debug::FINDVERBOSE, "Open braces: " << open_braces);
3690                 // LYXERR(Debug::FINDVERBOSE, "Replaced text (to be used as regex): " << par_as_string);
3691
3692                 // If entered regexp must match at begin of searched string buffer
3693                 // Kornel: Added parentheses to use $1 for size of the leading string
3694                 string regexp_str;
3695                 string regexp2_str;
3696                 {
3697                         // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3698                         // Unfortunately is '\1', '\2', etc not working for strings with extra format
3699                         // so the convert has no effect in that case
3700                         for (int i = 7; i > 0; --i) {
3701                                 string orig = "\\\\" + std::to_string(i);
3702                                 string dest = "\\" + std::to_string(i+2);
3703                                 while (regex_replace(par_as_string, par_as_string, orig, dest));
3704                         }
3705                         if (opt.matchword) {
3706                                 modifyRegexForMatchWord(par_as_string);
3707                                 opt.matchword = false;
3708                         }
3709                         regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3710                         regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3711                 }
3712                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3713                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3714                 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3715         }
3716 }
3717
3718 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3719 {
3720         MatchResult mres;
3721
3722         mres.searched_size = len;
3723         if (at_begin &&
3724                         (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3725                 return mres;
3726
3727         docstring docstr = stringifyFromForSearch(opt, cur, len);
3728         string str;
3729         str = normalize(docstr, opt.ignoreformat);
3730         if (!opt.ignoreformat) {
3731                 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3732                 // remove closing '}' and '\n' to allow for use of '$' in regex
3733                 size_t lng = str.size();
3734                 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3735                         lng--;
3736                 if (lng != str.size()) {
3737                         str = str.substr(0, lng);
3738                 }
3739                 // Replace occurences of '~' to ' '
3740                 static std::regex specialChars { R"(~)" };
3741                 str = std::regex_replace(str, specialChars,  R"( )" );
3742         }
3743         if (str.empty()) {
3744                 mres.match_len = -1;
3745                 return mres;
3746         }
3747         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
3748
3749         LASSERT(use_regexp, /**/);
3750         {
3751                 // use_regexp always true
3752                 LYXERR(Debug::FINDVERBOSE, "Searching in regexp mode: at_begin=" << at_begin);
3753 #if QTSEARCH
3754                 QString qstr = QString::fromStdString(str);
3755                 QRegularExpression const *p_regexp;
3756                 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3757                 if (at_begin) {
3758                         p_regexp = &regexp;
3759                 } else {
3760                         p_regexp = &regexp2;
3761                 }
3762                 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3763                 if (!match.hasMatch())
3764                         return mres;
3765 #else
3766                 regex const *p_regexp;
3767                 regex_constants::match_flag_type flags;
3768                 if (at_begin) {
3769                         flags = regex_constants::match_continuous;
3770                         p_regexp = &regexp;
3771                 } else {
3772                         flags = regex_constants::match_default;
3773                         p_regexp = &regexp2;
3774                 }
3775                 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3776                 if (re_it == sregex_iterator())
3777                         return mres;
3778                 match_results<string::const_iterator> const & m = *re_it;
3779 #endif
3780                 // Whole found string, including the leading
3781                 // std: m[0].second - m[0].first
3782                 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3783                 //
3784                 // Size of the leading string
3785                 // std: m[1].second - m[1].first
3786                 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3787                 int leadingsize = 0;
3788 #if QTSEARCH
3789                 if (match.lastCapturedIndex() > 0) {
3790                         leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3791                 }
3792
3793 #else
3794                 if (m.size() > 2) {
3795                         leadingsize = m[1].second - m[1].first;
3796                 }
3797 #endif
3798 #if QTSEARCH
3799                 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3800                 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3801                 // because of different number of closing at end of string
3802                 // we have to 'unify' the length of the post-match.
3803                 // Done by ignoring closing parenthesis and linefeeds at string end
3804                 int matchend = match.capturedEnd(0);
3805                 size_t strsize = qstr.size();
3806                 if (!opt.ignoreformat) {
3807                         while (mres.match_len > 0) {
3808                                 QChar c = qstr.at(matchend - 1);
3809                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3810                                         // mres.match_len--;
3811                                         matchend--;
3812                                 }
3813                                 else
3814                                         break;
3815                         }
3816                         while (strsize > (size_t) match.capturedEnd(0)) {
3817                                 QChar c = qstr.at(strsize-1);
3818                                 if ((c == '\n') || (c == '}')) {
3819                                         --strsize;
3820                                 }
3821                                 else
3822                                         break;
3823                         }
3824                 }
3825                 // LYXERR0(qstr.toStdString());
3826                 mres.match2end = strsize - matchend;
3827                 mres.pos = match.capturedStart(2);
3828 #else
3829                 mres.match_prefix = m[2].second - m[2].first;
3830                 mres.match_len = m[0].second - m[2].second;
3831                 // ignore closing parenthesis and linefeeds at string end
3832                 size_t strend = m[0].second - m[0].first;
3833                 int matchend = strend;
3834                 size_t strsize = str.size();
3835                 if (!opt.ignoreformat) {
3836                         while (mres.match_len > 0) {
3837                                 char c = str.at(matchend - 1);
3838                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3839                                         // mres.match_len--;
3840                                         matchend--;
3841                                 }
3842                                 else
3843                                         break;
3844                         }
3845                         while (strsize > strend) {
3846                                 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3847                                         --strsize;
3848                                 }
3849                                 else
3850                                         break;
3851                         }
3852                 }
3853                 // LYXERR0(str);
3854                 mres.match2end = strsize - matchend;
3855                 mres.pos = m[2].first - m[0].first;;
3856 #endif
3857                 if (mres.match2end < 0)
3858                         mres.match_len = 0;
3859                 mres.leadsize = leadingsize;
3860 #if QTSEARCH
3861                 if (mres.match_len > 0) {
3862                         string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3863                         mres.result.push_back(a0);
3864                         for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3865                                 mres.result.push_back(match.captured(i).toStdString());
3866                         }
3867                 }
3868 #else
3869                 if (mres.match_len > 0) {
3870                         string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3871                         mres.result.push_back(a0);
3872                         for (size_t i = 3; i < m.size(); i++) {
3873                                 mres.result.push_back(m[i]);
3874                         }
3875                 }
3876 #endif
3877                 return mres;
3878         }
3879 }
3880
3881
3882 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3883 {
3884         MatchResult mres = findAux(cur, len, at_begin);
3885         int res = mres.match_len;
3886         LYXERR(Debug::FINDVERBOSE,
3887                "res=" << res << ", at_begin=" << at_begin
3888                << ", matchAtStart=" << opt.matchAtStart
3889                << ", inTexted=" << cur.inTexted());
3890         if (opt.matchAtStart) {
3891                 if (cur.pos() != 0)
3892                         mres.match_len = 0;
3893                 else if (mres.match_prefix > 0)
3894                         mres.match_len = 0;
3895                 return mres;
3896         }
3897         else
3898                 return mres;
3899 }
3900
3901 #if 0
3902 static bool simple_replace(string &t, string from, string to)
3903 {
3904         regex repl("(\\\\)*(" + from + ")");
3905         string s("");
3906         size_t lastpos = 0;
3907         smatch sub;
3908         for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3909                 sub = *it;
3910                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3911                         continue;
3912                 if (lastpos < (size_t) sub.position(2))
3913                         s += t.substr(lastpos, sub.position(2) - lastpos);
3914                 s += to;
3915                 lastpos = sub.position(2) + sub.length(2);
3916         }
3917         if (lastpos == 0)
3918                 return false;
3919         else if (lastpos < t.length())
3920                 s += t.substr(lastpos, t.length() - lastpos);
3921         t = s;
3922         return true;
3923 }
3924 #endif
3925
3926 string MatchStringAdv::normalize(docstring const & s, bool ignore_format) const
3927 {
3928         string t;
3929         t = lyx::to_utf8(s);
3930         // Remove \n at begin
3931         while (!t.empty() && t[0] == '\n')
3932                 t = t.substr(1);
3933         // Remove \n* at end
3934         while (!t.empty() && t[t.size() - 1] == '\n') {
3935                 t = t.substr(0, t.size() - 1);
3936         }
3937         size_t pos;
3938         // Handle all other '\n'
3939         while ((pos = t.find("\n")) != string::npos) {
3940                 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3941                         // Handle '\\\n'
3942                         if (isAlnumASCII(t[pos+1])) {
3943                                 t.replace(pos-2, 3, " ");
3944                         }
3945                         else {
3946                                 t.replace(pos-2, 3, "");
3947                         }
3948                 }
3949                 else {
3950                         if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3951                                 // '\n' adjacent to non-alpha-numerics, discard
3952                                 t.replace(pos, 1, "");
3953                         }
3954                         else {
3955                                 // Replace all other \n with spaces
3956                                 t.replace(pos, 1, " ");
3957                         }
3958                         if (!ignore_format) {
3959                                 size_t count = 0;
3960                                 while ((pos > count + 1) && (t[pos - 1 -count] == '%')) {
3961                                         count++;
3962                                 }
3963                                 if (count > 0) {
3964                                         t.replace(pos - count, count, "");
3965                                 }
3966                         }
3967                 }
3968         }
3969         // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3970         // Kornel: Added textsl, textsf, textit, texttt and noun
3971         // + allow to seach for colored text too
3972         LYXERR(Debug::FINDVERBOSE, "Removing stale empty macros from: " << t);
3973         while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3974                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3975         while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3976                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\section{}, \\part{}, \\paragraph{} macros from: " << t);
3977         while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3978
3979         return t;
3980 }
3981
3982
3983 docstring stringifyFromCursor(DocIterator const & cur, int len)
3984 {
3985         LYXERR(Debug::FINDVERBOSE, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3986         if (cur.inTexted()) {
3987                 Paragraph const & par = cur.paragraph();
3988                 // TODO what about searching beyond/across paragraph breaks ?
3989                 // TODO Try adding a AS_STR_INSERTS as last arg
3990                 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3991                                         int(par.size()) : cur.pos() + len;
3992                 // OutputParams runparams(&cur.buffer()->params().encoding());
3993                 OutputParams runparams(encodings.fromLyXName("utf8"));
3994                 runparams.nice = true;
3995                 setFindParams(runparams);
3996                 int option = AS_STR_INSETS | AS_STR_PLAINTEXT;
3997                 if (ignoreFormats.getDeleted()) {
3998                         option |= AS_STR_SKIPDELETE;
3999                         runparams.for_search = OutputParams::SearchWithoutDeleted;
4000                 }
4001                 else {
4002                         runparams.for_search = OutputParams::SearchWithDeleted;
4003                 }
4004                 if (ignoreFormats.getNonContent()) {
4005                         runparams.for_search |= OutputParams::SearchNonOutput;
4006                 }
4007                 LYXERR(Debug::FINDVERBOSE, "Stringifying with cur: "
4008                        << cur << ", from pos: " << cur.pos() << ", end: " << end);
4009                 docstring res = from_utf8(latexNamesToUtf8(par.asString(cur.pos(), end,
4010                                                                         option,
4011                                                                         &runparams)));
4012                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Stringified text from pos(" << cur.pos() << ") len(" << len << "): " << res);
4013                 return res;
4014         } else if (cur.inMathed()) {
4015                 CursorSlice cs = cur.top();
4016                 MathData md = cs.cell();
4017                 MathData::const_iterator it_end =
4018                                 (( len == -1 || cs.pos() + len > int(md.size()))
4019                                  ? md.end()
4020                                  : md.begin() + cs.pos() + len );
4021                 MathData md2;
4022                 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
4023                         md2.push_back(*it);
4024                 docstring res = from_utf8(latexNamesToUtf8(asString(md2)));
4025                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Stringified math from pos(" << cur.pos() << ") len(" << len << "): " << res);
4026                 return res;
4027         }
4028         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
4029         return docstring();
4030 }
4031
4032 /** Computes the LaTeX export of buf starting from cur and ending len positions
4033  * after cur, if len is positive, or at the paragraph or innermost inset end
4034  * if len is -1.
4035  */
4036 docstring latexifyFromCursor(DocIterator const & cur, int len)
4037 {
4038         /*
4039         LYXERR(Debug::FINDVERBOSE, "Latexifying with len=" << len << " from cursor at pos: " << cur);
4040         LYXERR(Debug::FINDVERBOSE, "  with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
4041                << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
4042         */
4043         Buffer const & buf = *cur.buffer();
4044
4045         odocstringstream ods;
4046         otexstream os(ods);
4047         //OutputParams runparams(&buf.params().encoding());
4048         OutputParams runparams(encodings.fromLyXName("utf8"));
4049         runparams.nice = false;
4050         setFindParams(runparams);
4051         if (ignoreFormats.getDeleted()) {
4052                 runparams.for_search = OutputParams::SearchWithoutDeleted;
4053         }
4054         else {
4055                 runparams.for_search = OutputParams::SearchWithDeleted;
4056         }
4057         if (ignoreFormats.getNonContent()) {
4058                 runparams.for_search |= OutputParams::SearchNonOutput;
4059         }
4060
4061         if (cur.inTexted()) {
4062                 // @TODO what about searching beyond/across paragraph breaks ?
4063                 pos_type endpos = cur.paragraph().size();
4064                 if (len != -1 && endpos > cur.pos() + len)
4065                         endpos = cur.pos() + len;
4066                 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
4067                           string(), cur.pos(), endpos, true);
4068                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Latexified text from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4069                 return(ods.str());
4070         } else if (cur.inMathed()) {
4071                 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
4072                 for (int s = cur.depth() - 1; s >= 0; --s) {
4073                         CursorSlice const & cs = cur[s];
4074                         if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
4075                                 TeXMathStream ws(os);
4076                                 cs.asInsetMath()->asHullInset()->header_write(ws);
4077                                 break;
4078                         }
4079                 }
4080
4081                 CursorSlice const & cs = cur.top();
4082                 MathData md = cs.cell();
4083                 MathData::const_iterator it_end =
4084                                 ((len == -1 || cs.pos() + len > int(md.size()))
4085                                  ? md.end()
4086                                  : md.begin() + cs.pos() + len);
4087                 MathData md2;
4088                 for (MathData::const_iterator it = md.begin() + cs.pos();
4089                      it != it_end; ++it)
4090                         md2.push_back(*it);
4091
4092                 ods << asString(md2);
4093                 // Retrieve the math environment type, and add '$' or '$]'
4094                 // or others (\end{equation}) accordingly
4095                 for (int s = cur.depth() - 1; s >= 0; --s) {
4096                         CursorSlice const & cs2 = cur[s];
4097                         InsetMath * inset = cs2.asInsetMath();
4098                         if (inset && inset->asHullInset()) {
4099                                 TeXMathStream ws(os);
4100                                 inset->asHullInset()->footer_write(ws);
4101                                 break;
4102                         }
4103                 }
4104                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Latexified math from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4105         } else {
4106                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
4107         }
4108         return ods.str();
4109 }
4110
4111 #if defined(ResultsDebug)
4112 // Debugging output
4113 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
4114 {
4115         LYXERR0( "from:\t\t\t" << from);
4116         string status;
4117         if (mres.pos_len > 0) {
4118                 // Set in finalize
4119                 status = "FINALSEARCH";
4120         }
4121         else {
4122                 if (mres.match_len > 0) {
4123                         if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
4124                                 status = "Good Match";
4125                         else
4126                                 status = "Matched in";
4127                 }
4128                 else
4129                         status = "MissedSearch";
4130         }
4131
4132         LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
4133         if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
4134                 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
4135         if ((mres.pos > 0) || (mres.match_prefix > 0))
4136                 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
4137         for (size_t i = 0; i < mres.result.size(); i++)
4138                 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
4139 }
4140 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
4141 #else
4142 #define displayMres(s, txt, cur)
4143 #endif
4144
4145 /** Finalize an advanced find operation, advancing the cursor to the innermost
4146  ** position that matches, plus computing the length of the matching text to
4147  ** be selected
4148  ** Return the cur.pos() difference between start and end of found match
4149  **/
4150 MatchResult findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
4151 {
4152         // Search the foremost position that matches (avoids find of entire math
4153         // inset when match at start of it)
4154         DocIterator old_cur(cur.buffer());
4155         MatchResult mres;
4156         static MatchResult fail = MatchResult();
4157         MatchResult max_match;
4158         // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
4159         // Happens with e.g. hyperlinks
4160         // either one sees "http://www.bla.bla" or nothing
4161         // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
4162         // and although we search for only 3 chars, we find the whole hyperlink inset
4163         bool at_begin = (expected.match_prefix == 0);
4164         if (!match.opt.forward && match.opt.ignoreformat) {
4165                 if (expected.pos > 0)
4166                         return fail;
4167         }
4168         LASSERT(at_begin, /**/);
4169         if (expected.match_len > 0 && at_begin) {
4170                 // Search for deepest match
4171                 old_cur = cur;
4172                 max_match = expected;
4173                 do {
4174                         size_t d = cur.depth();
4175                         cur.forwardPos();
4176                         if (!cur)
4177                                 break;
4178                         if (cur.depth() < d)
4179                                 break;
4180                         if (cur.depth() == d)
4181                                 break;
4182                         size_t lastd = d;
4183                         while (cur && cur.depth() > lastd) {
4184                                 lastd = cur.depth();
4185                                 mres = match(cur, -1, at_begin);
4186                                 displayMres(mres, "Checking innermost", cur);
4187                                 if (mres.match_len > 0)
4188                                         break;
4189                                 // maybe deeper?
4190                                 cur.forwardPos();
4191                         }
4192                         if (mres.match_len < expected.match_len)
4193                                 break;
4194                         max_match = mres;
4195                         old_cur = cur;;
4196                 } while(1);
4197                 cur = old_cur;
4198         }
4199         else {
4200                 // (expected.match_len <= 0)
4201                 mres = match(cur);      /* match valid only if not searching whole words */
4202                 displayMres(mres, "Start with negative match", cur);
4203                 max_match = mres;
4204         }
4205         if (max_match.match_len <= 0) return fail;
4206         LYXERR(Debug::FINDVERBOSE, "Ok");
4207
4208         // Compute the match length
4209         int len = 1;
4210         if (cur.pos() + len > cur.lastpos())
4211                 return fail;
4212
4213         LASSERT(match.use_regexp, /**/);
4214         {
4215                 int minl = 1;
4216                 int maxl = cur.lastpos() - cur.pos();
4217                 // Greedy behaviour while matching regexps
4218                 while (maxl > minl) {
4219                         MatchResult mres2;
4220                         mres2 = match(cur, len, at_begin);
4221                         displayMres(mres2, "Finalize loop", cur);
4222                         int actual_match_len = mres2.match_len;
4223                         if (actual_match_len >= max_match.match_len) {
4224                                 // actual_match_len > max_match _can_ happen,
4225                                 // if the search area splits
4226                                 // some following word so that the regex
4227                                 // (e.g. 'r.*r\b' matches 'r' from the middle of the
4228                                 // splitted word)
4229                                 // This means, the len value is too big
4230                                 actual_match_len = max_match.match_len;
4231                                 max_match = mres2;
4232                                 max_match.match_len = actual_match_len;
4233                                 maxl = len;
4234                                 if (maxl - minl < 4)
4235                                         len = (maxl + minl)/2;
4236                                 else
4237                                         len = minl + (maxl - minl + 3)/4;
4238                         }
4239                         else {
4240                                 // (actual_match_len < max_match.match_len)
4241                                 minl = len + 1;
4242                                 len = (maxl + minl)/2;
4243                         }
4244                 }
4245                 len = minl;
4246                 old_cur = cur;
4247                 // Search for real start of matched characters
4248                 while (len > 1) {
4249                         MatchResult actual_match;
4250                         do {
4251                                 cur.forwardPos();
4252                         } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
4253                         if (cur.depth() < old_cur.depth()) {
4254                                 // Outer inset?
4255                                 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
4256                                 break;
4257                         }
4258                         if (cur.pos() != old_cur.pos()) {
4259                                 // OK, forwarded 1 pos in actual inset
4260                                 actual_match = match(cur, len-1, at_begin);
4261                                 if (actual_match.match_len == max_match.match_len) {
4262                                         // Ha, got it! The shorter selection has the same match length
4263                                         len--;
4264                                         old_cur = cur;
4265                                         max_match = actual_match;
4266                                 }
4267                                 else {
4268                                         // OK, the shorter selection matches less chars, revert to previous value
4269                                         cur = old_cur;
4270                                         break;
4271                                 }
4272                         }
4273                         else {
4274                                 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
4275                                 actual_match = match(cur, len, at_begin);
4276                                 if (actual_match.match_len == max_match.match_len) {
4277                                         old_cur = cur;
4278                                         max_match = actual_match;
4279                                 }
4280                         }
4281                 }
4282                 if (len == 0)
4283                         return fail;
4284                 else {
4285                         max_match.pos_len = len;
4286                         displayMres(max_match, "SEARCH RESULT", cur)
4287                                         return max_match;
4288                 }
4289         }
4290 }
4291
4292 /// Finds forward
4293 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
4294 {
4295         if (!cur)
4296                 return 0;
4297         bool repeat = false;
4298         DocIterator orig_cur;   // to be used if repeat not successful
4299         MatchResult orig_mres;
4300         while (!theApp()->longOperationCancelled() && cur) {
4301                 //(void) findAdvForwardInnermost(cur);
4302                 LYXERR(Debug::FINDVERBOSE, "findForwardAdv() cur: " << cur);
4303                 MatchResult mres = match(cur, -1, false);
4304                 string msg = "Starting";
4305                 if (repeat)
4306                         msg = "Repeated";
4307                 displayMres(mres, msg + " findForwardAdv", cur)
4308                                 int match_len = mres.match_len;
4309                 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
4310                         LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
4311                         match_len = 0;
4312                 }
4313                 if (match_len <= 0) {
4314                         // This should exit nested insets, if any, or otherwise undefine the currsor.
4315                         cur.pos() = cur.lastpos();
4316                         LYXERR(Debug::FINDVERBOSE, "Advancing pos: cur=" << cur);
4317                         cur.forwardPos();
4318                 }
4319                 else {  // match_len > 0
4320                         // Try to find the begin of searched string
4321                         int increment;
4322                         int firstInvalid = cur.lastpos() - cur.pos();
4323                         {
4324                                 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
4325                                 int incrcur = (firstInvalid + 1 )*3/4;
4326                                 if (incrcur < incrmatch)
4327                                         increment = incrcur;
4328                                 else
4329                                         increment = incrmatch;
4330                                 if (increment < 1)
4331                                         increment = 1;
4332                         }
4333                         LYXERR(Debug::FINDVERBOSE, "Set increment to " << increment);
4334                         while (increment > 0) {
4335                                 DocIterator old_cur = cur;
4336                                 if (cur.pos() + increment >= cur.lastpos()) {
4337                                         increment /= 2;
4338                                         continue;
4339                                 }
4340                                 cur.pos() = cur.pos() + increment;
4341                                 MatchResult mres2 = match(cur, -1, false);
4342                                 displayMres(mres2, "findForwardAdv loop", cur)
4343                                                 switch (interpretMatch(mres, mres2)) {
4344                                         case MatchResult::newIsTooFar:
4345                                                 // behind the expected match
4346                                                 firstInvalid = increment;
4347                                                 cur = old_cur;
4348                                                 increment /= 2;
4349                                                 break;
4350                                         case MatchResult::newIsBetter:
4351                                                 // not reached yet, but cur.pos()+increment is bettert
4352                                                 mres = mres2;
4353                                                 firstInvalid -= increment;
4354                                                 if (increment > firstInvalid*3/4)
4355                                                         increment = firstInvalid*3/4;
4356                                                 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
4357                                                         if (increment >= mres2.match_prefix)
4358                                                                 increment = (mres2.match_prefix+1)*3/4;
4359                                                 }
4360                                                 break;
4361                                         default:
4362                                                 // Todo@
4363                                                 // Handle not like MatchResult::newIsTooFar
4364                                                 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
4365                                                 firstInvalid--;
4366                                                 increment = increment*3/4;
4367                                                 cur = old_cur;
4368                                                 break;
4369                                 }
4370                         }
4371                         if (mres.match_len > 0) {
4372                                 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
4373                                         // The match seems to indicate some deeper level
4374                                         repeat = true;
4375                                         orig_cur = cur;
4376                                         orig_mres = mres;
4377                                         cur.forwardPos();
4378                                         continue;
4379                                 }
4380                         }
4381                         else if (repeat) {
4382                                 // should never be reached.
4383                                 cur = orig_cur;
4384                                 mres = orig_mres;
4385                         }
4386                         // LYXERR0("Leaving first loop");
4387                         LYXERR(Debug::FINDVERBOSE, "Finalizing 1");
4388                         MatchResult found_match = findAdvFinalize(cur, match, mres);
4389                         if (found_match.match_len > 0) {
4390                                 match.FillResults(found_match);
4391                                 return found_match.pos_len;
4392                         }
4393                         else {
4394                                 // try next possible match
4395                                 cur.forwardPos();
4396                                 repeat = false;
4397                                 continue;
4398                         }
4399                 }
4400         }
4401         return 0;
4402 }
4403
4404
4405 /// Find the most backward consecutive match within same paragraph while searching backwards.
4406 MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, MatchResult &expected)
4407 {
4408         DocIterator cur_begin = cur;
4409         cur_begin.pos() = 0;
4410         DocIterator tmp_cur = cur;
4411         MatchResult mr = findAdvFinalize(tmp_cur, match, expected);
4412         Inset & inset = cur.inset();
4413         for (; cur != cur_begin; cur.backwardPos()) {
4414                 LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): cur=" << cur);
4415                 DocIterator new_cur = cur;
4416                 new_cur.backwardPos();
4417                 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
4418                         break;
4419                 MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
4420                 if (new_mr.match_len == mr.match_len)
4421                         break;
4422                 mr = new_mr;
4423         }
4424         LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): exiting with cur=" << cur);
4425         return mr;
4426 }
4427
4428
4429 /// Finds backwards
4430 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
4431 {
4432         if (! cur)
4433                 return 0;
4434         // Backup of original position
4435         DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4436         if (cur == cur_begin)
4437                 return 0;
4438         cur.backwardPos();
4439         DocIterator cur_orig(cur);
4440         bool pit_changed = false;
4441         do {
4442                 cur.pos() = 0;
4443                 MatchResult found_match = match(cur, -1, false);
4444
4445                 if (found_match.match_len > 0) {
4446                         if (pit_changed)
4447                                 cur.pos() = cur.lastpos();
4448                         else
4449                                 cur.pos() = cur_orig.pos();
4450                         LYXERR(Debug::FINDVERBOSE, "findBackAdv2: cur: " << cur);
4451                         DocIterator cur_prev_iter;
4452                         do {
4453                                 found_match = match(cur);
4454                                 LYXERR(Debug::FINDVERBOSE, "findBackAdv3: found_match="
4455                                        << (found_match.match_len > 0) << ", cur: " << cur);
4456                                 if (found_match.match_len > 0) {
4457                                         MatchResult found_mr = findMostBackwards(cur, match, found_match);
4458                                         if (found_mr.pos_len > 0) {
4459                                                 match.FillResults(found_mr);
4460                                                 return found_mr.pos_len;
4461                                         }
4462                                 }
4463
4464                                 // Stop if begin of document reached
4465                                 if (cur == cur_begin)
4466                                         break;
4467                                 cur_prev_iter = cur;
4468                                 cur.backwardPos();
4469                         } while (true);
4470                 }
4471                 if (cur == cur_begin)
4472                         break;
4473                 if (cur.pit() > 0)
4474                         --cur.pit();
4475                 else
4476                         cur.backwardPos();
4477                 pit_changed = true;
4478         } while (!theApp()->longOperationCancelled());
4479         return 0;
4480 }
4481
4482
4483 } // namespace
4484
4485
4486 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4487                                  DocIterator const & cur, int len)
4488 {
4489         if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4490                 return docstring();
4491         if (!opt.ignoreformat)
4492                 return latexifyFromCursor(cur, len);
4493         else
4494                 return stringifyFromCursor(cur, len);
4495 }
4496
4497
4498 FindAndReplaceOptions::FindAndReplaceOptions(
4499                 docstring const & _find_buf_name, bool _casesensitive,
4500                 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4501                 docstring const & _repl_buf_name, bool _keep_case,
4502                 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4503         : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4504           forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4505           repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4506 {
4507 }
4508
4509
4510 namespace {
4511
4512
4513 /** Check if 'len' letters following cursor are all non-lowercase */
4514 static bool allNonLowercase(Cursor const & cur, int len)
4515 {
4516         pos_type beg_pos = cur.selectionBegin().pos();
4517         pos_type end_pos = cur.selectionBegin().pos() + len;
4518         if (len > cur.lastpos() + 1 - beg_pos) {
4519                 LYXERR(Debug::FINDVERBOSE, "This should not happen, more debug needed");
4520                 len = cur.lastpos() + 1 - beg_pos;
4521                 end_pos = beg_pos + len;
4522         }
4523         for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4524                 if (isLowerCase(cur.paragraph().getChar(pos)))
4525                         return false;
4526         return true;
4527 }
4528
4529
4530 /** Check if first letter is upper case and second one is lower case */
4531 static bool firstUppercase(Cursor const & cur)
4532 {
4533         char_type ch1, ch2;
4534         pos_type pos = cur.selectionBegin().pos();
4535         if (pos >= cur.lastpos() - 1) {
4536                 LYXERR(Debug::FINDVERBOSE, "No upper-case at cur: " << cur);
4537                 return false;
4538         }
4539         ch1 = cur.paragraph().getChar(pos);
4540         ch2 = cur.paragraph().getChar(pos + 1);
4541         bool result = isUpperCase(ch1) && isLowerCase(ch2);
4542         LYXERR(Debug::FINDVERBOSE, "firstUppercase(): "
4543                << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4544                << ch2 << "(" << char(ch2) << ")"
4545                << ", result=" << result << ", cur=" << cur);
4546         return result;
4547 }
4548
4549
4550 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4551  **
4552  ** \fixme What to do with possible further paragraphs in replace buffer ?
4553  **/
4554 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4555 {
4556         ParagraphList::iterator pit = buffer.paragraphs().begin();
4557         LASSERT(!pit->empty(), /**/);
4558         pos_type right = pos_type(1);
4559         pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4560         right = pit->size();
4561         pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4562 }
4563 } // namespace
4564
4565 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4566 {
4567         // Should replace the string "$" + std::to_string(matchnum) with replacement
4568         // if the char '$' is not prefixed with odd number of char '\\'
4569         static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4570         string s;
4571         size_t lastpos = 0;
4572         smatch sub;
4573         for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4574                 sub = *it;
4575                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4576                         continue;
4577                 int num = stoi(sub.str(3), nullptr, 10);
4578                 if (num >= maxmatchnum)
4579                         continue;
4580                 if (lastpos < (size_t) sub.position(2))
4581                         s += t.substr(lastpos, sub.position(2) - lastpos);
4582                 s += replacements[num];
4583                 lastpos = sub.position(2) + sub.length(2);
4584         }
4585         if (lastpos == 0)
4586                 return false;
4587         else if (lastpos < t.length())
4588                 s += t.substr(lastpos, t.length() - lastpos);
4589         t = s;
4590         return true;
4591 }
4592
4593 ///
4594 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4595 {
4596         Cursor & cur = bv->cursor();
4597         if (opt.repl_buf_name.empty()
4598                         || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4599                         || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4600                 return 0;
4601
4602         DocIterator sel_beg = cur.selectionBegin();
4603         DocIterator sel_end = cur.selectionEnd();
4604         if (&sel_beg.inset() != &sel_end.inset()
4605                         || sel_beg.pit() != sel_end.pit()
4606                         || sel_beg.idx() != sel_end.idx())
4607                 return 0;
4608         int sel_len = sel_end.pos() - sel_beg.pos();
4609         LYXERR(Debug::FINDVERBOSE, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4610                << ", sel_len: " << sel_len << endl);
4611         if (sel_len == 0)
4612                 return 0;
4613         LASSERT(sel_len > 0, return 0);
4614
4615         if (!matchAdv(sel_beg, sel_len).match_len)
4616                 return 0;
4617
4618         // Build a copy of the replace buffer, adapted to the KeepCase option
4619         Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4620         ostringstream oss;
4621         repl_buffer_orig.write(oss);
4622         string lyx = oss.str();
4623         if (matchAdv.valid_matches > 0)
4624                 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4625         Buffer repl_buffer(string(), false);
4626         repl_buffer.setInternal(true);
4627         repl_buffer.setUnnamed(true);
4628         LASSERT(repl_buffer.readString(lyx), return 0);
4629         if (opt.keep_case && sel_len >= 2) {
4630                 LYXERR(Debug::FINDVERBOSE, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4631                 if (cur.inTexted()) {
4632                         if (firstUppercase(cur))
4633                                 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4634                         else if (allNonLowercase(cur, sel_len))
4635                                 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4636                 }
4637         }
4638         cap::cutSelection(cur, false);
4639         if (cur.inTexted()) {
4640                 repl_buffer.changeLanguage(
4641                                         repl_buffer.language(),
4642                                         cur.getFont().language());
4643                 LYXERR(Debug::FINDVERBOSE, "Replacing by pasteParagraphList()ing repl_buffer");
4644                 LYXERR(Debug::FINDVERBOSE, "Before pasteParagraphList() cur=" << cur << endl);
4645                 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4646                                         repl_buffer.params().documentClassPtr(),
4647                                         repl_buffer.params().authors(),
4648                                         bv->buffer().errorList("Paste"));
4649                 LYXERR(Debug::FINDVERBOSE, "After pasteParagraphList() cur=" << cur << endl);
4650                 sel_len = repl_buffer.paragraphs().begin()->size();
4651         } else if (cur.inMathed()) {
4652                 odocstringstream ods;
4653                 otexstream os(ods);
4654                 // OutputParams runparams(&repl_buffer.params().encoding());
4655                 OutputParams runparams(encodings.fromLyXName("utf8"));
4656                 runparams.nice = false;
4657                 setFindParams(runparams);
4658                 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams, string(), -1, -1, true);
4659                 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4660                 docstring repl_latex = ods.str();
4661                 LYXERR(Debug::FINDVERBOSE, "Latexified replace_buffer: '" << repl_latex << "'");
4662                 string s;
4663                 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4664                 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4665                 repl_latex = from_utf8(s);
4666                 LYXERR(Debug::FINDVERBOSE, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4667                 MathData ar(cur.buffer());
4668                 asArray(repl_latex, ar, Parse::NORMAL);
4669                 cur.insert(ar);
4670                 sel_len = ar.size();
4671                 LYXERR(Debug::FINDVERBOSE, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4672         }
4673         if (cur.pos() >= sel_len)
4674                 cur.pos() -= sel_len;
4675         else
4676                 cur.pos() = 0;
4677         LYXERR(Debug::FINDVERBOSE, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4678         bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4679         bv->processUpdateFlags(Update::Force);
4680         return 1;
4681 }
4682
4683
4684 /// Perform a FindAdv operation.
4685 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4686 {
4687         DocIterator cur;
4688         int pos_len = 0;
4689
4690         // e.g., when invoking word-findadv from mini-buffer wither with
4691         //       wrong options syntax or before ever opening advanced F&R pane
4692         if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4693                 return false;
4694
4695         try {
4696                 MatchStringAdv matchAdv(bv->buffer(), opt);
4697 #if QTSEARCH
4698                 if (!matchAdv.regexIsValid) {
4699                         bv->message(lyx::from_utf8(matchAdv.regexError));
4700                         return(false);
4701                 }
4702 #endif
4703                 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4704                 if (length > 0)
4705                         bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4706                 num_replaced += findAdvReplace(bv, opt, matchAdv);
4707                 cur = bv->cursor();
4708                 if (opt.forward)
4709                         pos_len = findForwardAdv(cur, matchAdv);
4710                 else
4711                         pos_len = findBackwardsAdv(cur, matchAdv);
4712         } catch (exception & ex) {
4713                 bv->message(from_utf8(ex.what()));
4714                 return false;
4715         }
4716
4717         if (pos_len == 0) {
4718                 if (num_replaced > 0) {
4719                         switch (num_replaced)
4720                         {
4721                         case 1:
4722                                 bv->message(_("One match has been replaced."));
4723                                 break;
4724                         case 2:
4725                                 bv->message(_("Two matches have been replaced."));
4726                                 break;
4727                         default:
4728                                 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4729                                 break;
4730                         }
4731                         num_replaced = 0;
4732                 }
4733                 else {
4734                         bv->message(_("Match not found."));
4735                 }
4736                 return false;
4737         }
4738
4739         if (num_replaced > 0)
4740                 bv->message(_("Match has been replaced."));
4741         else
4742                 bv->message(_("Match found."));
4743
4744         if (cur.pos() + pos_len > cur.lastpos()) {
4745                 // Prevent crash in bv->putSelectionAt()
4746                 // Should never happen, maybe LASSERT() here?
4747                 pos_len = cur.lastpos() - cur.pos();
4748         }
4749         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4750         bv->putSelectionAt(cur, pos_len, !opt.forward);
4751
4752         return true;
4753 }
4754
4755
4756 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4757 {
4758         os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4759            << opt.casesensitive << ' '
4760            << opt.matchword << ' '
4761            << opt.forward << ' '
4762            << opt.expandmacros << ' '
4763            << opt.ignoreformat << ' '
4764            << opt.replace_all << ' '
4765            << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4766            << opt.keep_case << ' '
4767            << int(opt.scope) << ' '
4768            << int(opt.restr);
4769
4770         LYXERR(Debug::FINDVERBOSE, "built: " << os.str());
4771
4772         return os;
4773 }
4774
4775
4776 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4777 {
4778         // LYXERR(Debug::FINDVERBOSE, "parsing");
4779         string s;
4780         string line;
4781         getline(is, line);
4782         while (line != "EOSS") {
4783                 if (! s.empty())
4784                         s = s + "\n";
4785                 s = s + line;
4786                 if (is.eof())   // Tolerate malformed request
4787                         break;
4788                 getline(is, line);
4789         }
4790         // LYXERR(Debug::FINDVERBOSE, "file_buf_name: '" << s << "'");
4791         opt.find_buf_name = from_utf8(s);
4792         is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4793         is.get();       // Waste space before replace string
4794         s = "";
4795         getline(is, line);
4796         while (line != "EOSS") {
4797                 if (! s.empty())
4798                         s = s + "\n";
4799                 s = s + line;
4800                 if (is.eof())   // Tolerate malformed request
4801                         break;
4802                 getline(is, line);
4803         }
4804         // LYXERR(Debug::FINDVERBOSE, "repl_buf_name: '" << s << "'");
4805         opt.repl_buf_name = from_utf8(s);
4806         is >> opt.keep_case;
4807         int i;
4808         is >> i;
4809         opt.scope = FindAndReplaceOptions::SearchScope(i);
4810         is >> i;
4811         opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4812
4813         /*
4814         LYXERR(Debug::FINDVERBOSE, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4815                << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4816                << opt.scope << ' ' << opt.restr);
4817         */
4818         return is;
4819 }
4820
4821 } // namespace lyx