src/lyxfind.cpp

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