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