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(0, /**/);
 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(0, /**/);
 889                                         }
 890                                 }
 891                         }
 892                         else if (sub.str(4) == "{") // transforms '\{' into '{'
 893                                 replace = "{";
 894                         else if (sub.str(4) == "}")
 895                                 replace = "}";
 896                         else if (sub.str(4) == "%")
 897                                 replace = "%";
 898                         else {
 899                                 AccentsIterator it_ac = accents.find(sub.str(4));
 900                                 if (it_ac == accents.end()) {
 901                                         replace = sub.str(2);
 902                                 }
 903                                 else {
 904                                         replace = it_ac->second;
 905                                 }
 906                         }
 907                 }
 908                 if (lastpos < (size_t) sub.position(2))
 909                         s += t.substr(lastpos, sub.position(2) - lastpos);
 910                 s += replace;
 911                 lastpos = sub.position(2) + sub.length(2);
 912         }
 913         if (lastpos == 0)
 914                 s = t;
 915         else if (lastpos < t.length())
 916                 s += t.substr(lastpos, t.length() - lastpos);
 917         // Handle quotes in regex
 918         // substitute all '„', '“', '»', '«' with '"'
 919         // and all '‚', '‘', '›', '‹' with "\'"
 920         static std::regex plainquotes { R"(„|“|»|«)" };
 921         static std::regex innerquotes { R"(‚|‘|›|‹)" };
 922         t = std::regex_replace(s, plainquotes, R"(")");
 923         s = std::regex_replace(t, innerquotes, R"(')");
 924         //LYXERR0("correctRegex output '" << s << "'");
 925         return s;
 926 }
 927
 928 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
 929 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
 930 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
 931 string escape_for_regex(string s, bool withformat)
 932 {
 933         size_t lastpos = 0;
 934         string result = "";
 935         while (lastpos < s.size()) {
 936                 size_t regex_pos = s.find("\\regexp{", lastpos);
 937                 if (regex_pos == string::npos) {
 938                         regex_pos = s.size();
 939                 }
 940                 if (regex_pos > lastpos) {
 941                         result += string2regex(s.substr(lastpos, regex_pos-lastpos));
 942                         lastpos = regex_pos;
 943                         if (lastpos == s.size())
 944                                 break;
 945                 }
 946                 size_t end_pos = s.find("\\endregexp", regex_pos + 8);
 947                 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
 948                 lastpos = end_pos + 13;
 949         }
 950         return result;
 951 }
 952
 953
 954 /// Wrapper for lyx::regex_replace with simpler interface
 955 bool regex_replace(string const & s, string & t, string const & searchstr,
 956                    string const & replacestr)
 957 {
 958         regex e(searchstr, regex_constants::ECMAScript);
 959         ostringstream oss;
 960         ostream_iterator<char, char> it(oss);
 961         regex_replace(it, s.begin(), s.end(), e, replacestr);
 962         // tolerate t and s be references to the same variable
 963         bool rv = (s != oss.str());
 964         t = oss.str();
 965         return rv;
 966 }
 967
 968 class MatchResult {
 969 public:
 970         enum range {
 971                 newIsTooFar,
 972                 newIsBetter,
 973                 newIsInvalid
 974         };
 975         int match_len;
 976         int match_prefix;
 977         int match2end;
 978         int pos;
 979         int leadsize;
 980         int pos_len;
 981         int searched_size;
 982         vector <string> result = vector <string>();
 983         MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {}
 984 };
 985
 986 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
 987 {
 988         if (newres.match2end < oldres.match2end)
 989                 return MatchResult::newIsTooFar;
 990         if (newres.match_len < oldres.match_len)
 991                 return MatchResult::newIsTooFar;
 992
 993         if (newres.match_len == oldres.match_len) {
 994                 if (newres.match2end == oldres.match2end)
 995                         return MatchResult::newIsBetter;
 996         }
 997         return MatchResult::newIsInvalid;
 998 }
 999
1000 /** The class performing a match between a position in the document and the FindAdvOptions.
1001  **/
1002
1003 class MatchStringAdv {
1004 public:
1005         MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
1006
1007         /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
1008          ** constructor as opt.search, under the opt.* options settings.
1009          **
1010          ** @param at_begin
1011          **     If set, then match is searched only against beginning of text starting at cur.
1012          **     If unset, then match is searched anywhere in text starting at cur.
1013          **
1014          ** @return
1015          ** The length of the matching text, or zero if no match was found.
1016          **/
1017         MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1018 #if QTSEARCH
1019         bool regexIsValid;
1020         string regexError;
1021 #endif
1022
1023 public:
1024         /// buffer
1025         lyx::Buffer * p_buf;
1026         /// first buffer on which search was started
1027         lyx::Buffer * const p_first_buf;
1028         /// options
1029         FindAndReplaceOptions const & opt;
1030
1031 private:
1032         /// Auxiliary find method (does not account for opt.matchword)
1033         MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
1034         void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
1035
1036         /** Normalize a stringified or latexified LyX paragraph.
1037          **
1038          ** Normalize means:
1039          ** <ul>
1040          **   <li>if search is not casesensitive, then lowercase the string;
1041          **   <li>remove any newline at begin or end of the string;
1042          **   <li>replace any newline in the middle of the string with a simple space;
1043          **   <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
1044          ** </ul>
1045          **
1046          ** @todo Normalization should also expand macros, if the corresponding
1047          ** search option was checked.
1048          **/
1049         string normalize(docstring const & s, bool ignore_fomat) const;
1050         // normalized string to search
1051         string par_as_string;
1052         // regular expression to use for searching
1053         // regexp2 is same as regexp, but prefixed with a ".*?"
1054 #if QTSEARCH
1055         QRegularExpression regexp;
1056         QRegularExpression regexp2;
1057 #else
1058         regex regexp;
1059         regex regexp2;
1060 #endif
1061         // leading format material as string
1062         string lead_as_string;
1063         // par_as_string after removal of lead_as_string
1064         string par_as_string_nolead;
1065         // unmatched open braces in the search string/regexp
1066         int open_braces;
1067         // number of (.*?) subexpressions added at end of search regexp for closing
1068         // environments, math mode, styles, etc...
1069         int close_wildcards;
1070 public:
1071         // Are we searching with regular expressions ?
1072         bool use_regexp = false;
1073         static int valid_matches;
1074         static vector <string> matches;
1075         void FillResults(MatchResult &found_mr);
1076 };
1077
1078 int MatchStringAdv::valid_matches = 0;
1079 vector <string> MatchStringAdv::matches = vector <string>(10);
1080
1081 void MatchStringAdv::FillResults(MatchResult &found_mr)
1082 {
1083         if (found_mr.match_len > 0) {
1084                 valid_matches = found_mr.result.size();
1085                 for (size_t i = 0; i < found_mr.result.size(); i++)
1086                         matches[i] = found_mr.result[i];
1087         } else
1088                 valid_matches = 0;
1089 }
1090
1091 static docstring buffer_to_latex(Buffer & buffer)
1092 {
1093         //OutputParams runparams(&buffer.params().encoding());
1094         OutputParams runparams(encodings.fromLyXName("utf8"));
1095         odocstringstream ods;
1096         otexstream os(ods);
1097         runparams.nice = true;
1098         runparams.flavor = Flavor::XeTeX;
1099         runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1100         // No side effect of file copying and image conversion
1101         runparams.dryrun = true;
1102         if (ignoreFormats.getDeleted())
1103                 runparams.for_search = OutputParams::SearchWithoutDeleted;
1104         else
1105                 runparams.for_search = OutputParams::SearchWithDeleted;
1106         if (ignoreFormats.getNonContent()) {
1107                 runparams.for_search |= OutputParams::SearchNonOutput;
1108         }
1109         pit_type const endpit = buffer.paragraphs().size();
1110         for (pit_type pit = 0; pit != endpit; ++pit) {
1111                 TeXOnePar(buffer, buffer.text(), pit, os, runparams, string(), -1, -1, true);
1112                 LYXERR(Debug::FINDVERBOSE, "searchString up to here: " << ods.str());
1113         }
1114         return ods.str();
1115 }
1116
1117 static string latexNamesToUtf8(docstring strIn)
1118 {
1119         string addtmp = to_utf8(strIn);
1120         static regex const rmAcc("(\\\\)*("
1121                                          "\\\\([A-Za-z]+\\{.\\})"       // e.g. "ddot{A}" == sub.str(3)
1122                                         "|\\\\([A-Za-z]+)( |\\{\\})?"   // e.g. "LyX", "LyX{}", "LyX " == sub.str(4)
1123                                         ")"
1124                                 );
1125         size_t lastpos = 0;
1126         smatch sub;
1127         string replace;
1128         string add("");
1129         if (accents.empty())
1130                 buildAccentsMap();
1131         for (sregex_iterator it_add(addtmp.begin(), addtmp.end(), rmAcc), end; it_add != end; ++it_add) {
1132                 sub = *it_add;
1133                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
1134                         continue;
1135                 }
1136                 else {
1137                         string key;
1138                         if (sub.length(3) > 0)
1139                                 key = sub.str(3);
1140                         else
1141                                 key = sub.str(4);
1142                         AccentsIterator it_ac = accents.find(key);
1143                         if (it_ac == accents.end()) {
1144                                 replace = sub.str(2);
1145                         }
1146                         else {
1147                                 replace = it_ac->second;
1148                         }
1149                 }
1150                 if (lastpos < (size_t) sub.position(2))
1151                         add += addtmp.substr(lastpos, sub.position(2) - lastpos);
1152                 add += replace;
1153                 lastpos = sub.position(2) + sub.length(2);
1154         }
1155         if (lastpos == 0)
1156                 add = addtmp;
1157         else if (addtmp.length() > lastpos)
1158                 add += addtmp.substr(lastpos, addtmp.length() - lastpos);
1159         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1160                         << add << "'");
1161         return add;
1162 }
1163
1164 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1165 {
1166         docstring str;
1167         if (!opt.ignoreformat) {
1168                 str = buffer_to_latex(buffer);
1169         } else {
1170                 // OutputParams runparams(&buffer.params().encoding());
1171                 OutputParams runparams(encodings.fromLyXName("utf8"));
1172                 runparams.nice = true;
1173                 runparams.flavor = Flavor::XeTeX;
1174                 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1175                 runparams.dryrun = true;
1176                 int option = AS_STR_INSETS |AS_STR_PLAINTEXT;
1177                 if (ignoreFormats.getDeleted()) {
1178                         option |= AS_STR_SKIPDELETE;
1179                         runparams.for_search = OutputParams::SearchWithoutDeleted;
1180                 }
1181                 else {
1182                         runparams.for_search = OutputParams::SearchWithDeleted;
1183                 }
1184                 if (ignoreFormats.getNonContent()) {
1185                         runparams.for_search |= OutputParams::SearchNonOutput;
1186                 }
1187                 string t("");
1188                 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1189                         Paragraph const & par = buffer.paragraphs().at(pit);
1190                         string add = latexNamesToUtf8(par.asString(pos_type(0), par.size(),
1191                                                                 option,
1192                                                                 &runparams));
1193                         LYXERR(Debug::FINDVERBOSE, "Adding to search string: '"
1194                                 << add << "'");
1195                         t += add;
1196                 }
1197                 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1198                 while (regex_replace(t, t, "\\\\(text|lyxmathsym|ensuremath)\\{([^\\}]*)\\}", "$2"));
1199                 str = from_utf8(t);
1200         }
1201         return str;
1202 }
1203
1204
1205 /// Return separation pos between the leading material and the rest
1206 static size_t identifyLeading(string const & s)
1207 {
1208         string t = s;
1209         // @TODO Support \item[text]
1210         // Kornel: Added textsl, textsf, textit, texttt and noun
1211         // + allow to search for colored text too
1212         while (regex_replace(t, t, "^\\\\(("
1213                              "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1214                              "lyxaddress|lyxrightaddress|"
1215                              "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1216                              "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1217                              "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1218                              "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1219                || regex_replace(t, t, "^\\$", "")
1220                || regex_replace(t, t, "^\\\\\\[", "")
1221                || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1222                || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1223                ;
1224         LYXERR(Debug::FINDVERBOSE, "  after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1225         return s.find(t);
1226 }
1227
1228 /*
1229  * Given a latexified string, retrieve some handled features
1230  * The features of the regex will later be compared with the features
1231  * of the searched text. If the regex features are not a
1232  * subset of the analized, then, in not format ignoring search
1233  * we can early stop the search in the relevant inset.
1234  */
1235 typedef map<string, bool> Features;
1236
1237 static Features identifyFeatures(string const & s)
1238 {
1239         static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1240         static regex const valid("^("
1241                 "("
1242                         "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1243                                 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1244                                 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1245                         "(u|uu)line|(s|x)out|uwave|"
1246                         "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1247                 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1248         smatch sub;
1249         bool displ = true;
1250         Features info;
1251
1252         for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1253                 sub = *it;
1254                 if (displ) {
1255                         if (sub.str(1).compare("regexp") == 0) {
1256                                 displ = false;
1257                                 continue;
1258                         }
1259                         string token = sub.str(1);
1260                         smatch sub2;
1261                         if (regex_match(token, sub2, valid)) {
1262                                 info[token] = true;
1263                         }
1264                         else {
1265                                 // ignore
1266                         }
1267                 }
1268                 else {
1269                         if (sub.str(1).compare("endregexp") == 0) {
1270                                 displ = true;
1271                                 continue;
1272                         }
1273                 }
1274         }
1275         return info;
1276 }
1277
1278 /*
1279  * defines values features of a key "\\[a-z]+{"
1280  */
1281 class KeyInfo {
1282 public:
1283         enum KeyType {
1284                 /* Char type with content discarded
1285                  * like \hspace{1cm} */
1286                 noContent,
1287                 /* Char, like \backslash */
1288                 isChar,
1289                 /* replace starting backslash with '#' */
1290                 isText,
1291                 /* \part, \section*, ... */
1292                 isSectioning,
1293                 /* title, author etc */
1294                 isTitle,
1295                 /* \foreignlanguage{ngerman}, ... */
1296                 isMain,
1297                 /* inside \code{}
1298                  * to discard language in content */
1299                 noMain,
1300                 isRegex,
1301                 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1302                 isMath,
1303                 /* fonts, colors, markups, ... */
1304                 isStandard,
1305                 /* footnotesize, ... large, ...
1306                  * Ignore all of them */
1307                 isSize,
1308                 invalid,
1309                 /* inputencoding, ...
1310                  * Discard also content, because they do not help in search */
1311                 doRemove,
1312                 /* twocolumns, ...
1313                  * like remove, but also all arguments */
1314                 removeWithArg,
1315                 /* item, listitem */
1316                 isList,
1317                 /* tex, latex, ... like isChar */
1318                 isIgnored,
1319                 /* like \lettrine[lines=5]{}{} */
1320                 cleanToStart,
1321                 // like isStandard, but always remove head
1322                 headRemove,
1323                 /* End of arguments marker for lettrine,
1324                  * so that they can be ignored */
1325                 endArguments
1326         };
1327         KeyInfo() = default;
1328         KeyInfo(KeyType type, int parcount, bool disable)
1329                 : keytype(type),
1330                   parenthesiscount(parcount),
1331                   disabled(disable) {}
1332         KeyType keytype = invalid;
1333         string head;
1334         int _tokensize = -1;
1335         int _tokenstart = -1;
1336         int _dataStart = -1;
1337         int _dataEnd = -1;
1338         int parenthesiscount = 1;
1339         bool disabled = false;
1340         bool used = false; /* by pattern */
1341 };
1342
1343 class Border {
1344 public:
1345         Border(int l=0, int u=0) : low(l), upper(u) {}
1346         int low;
1347         int upper;
1348 };
1349
1350 #define MAXOPENED 30
1351 class Intervall {
1352         bool isPatternString_;
1353 public:
1354         explicit Intervall(bool isPattern, string const & p)
1355                 : isPatternString_(isPattern), par(p), ignoreidx(-1),
1356                   actualdeptindex(0), hasTitle(false), langcount(0)
1357         {
1358                 depts[0] = 0;
1359                 closes[0] = 0;
1360         }
1361
1362         string par;
1363         int ignoreidx;
1364         static vector<Border> borders;
1365         int depts[MAXOPENED];
1366         int closes[MAXOPENED];
1367         int actualdeptindex;
1368         int previousNotIgnored(int) const;
1369         int nextNotIgnored(int) const;
1370         void handleOpenP(int i);
1371         void handleCloseP(int i, bool closingAllowed);
1372         void resetOpenedP(int openPos);
1373         void addIntervall(int upper);
1374         void addIntervall(int low, int upper); /* if explicit */
1375         void removeAccents();
1376         void setForDefaultLang(KeyInfo const & defLang) const;
1377         int findclosing(int start, int end, char up, char down, int repeat);
1378         void handleParentheses(int lastpos, bool closingAllowed);
1379         bool hasTitle;
1380         // Number of disabled language specs up
1381         // to current position in actual interval
1382         int langcount;
1383         int isOpeningPar(int pos) const;
1384         string titleValue;
1385         void output(ostringstream &os, int lastpos);
1386         // string show(int lastpos);
1387 };
1388
1389 vector<Border> Intervall::borders = vector<Border>(30);
1390
1391 int Intervall::isOpeningPar(int pos) const
1392 {
1393         if ((pos < 0) || (size_t(pos) >= par.size()))
1394                 return 0;
1395         if (par[pos] != '{')
1396                 return 0;
1397         if (size_t(pos) + 2 >= par.size())
1398                 return 1;
1399         if (par[pos+2] != '}')
1400                 return 1;
1401         if (par[pos+1] == '[' || par[pos+1] == ']')
1402                 return 3;
1403         return 1;
1404 }
1405
1406 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1407 {
1408         // Enable the use of first token again
1409         if (ignoreidx >= 0) {
1410                 int value = defLang._tokenstart + defLang._tokensize;
1411                 int borderidx = 0;
1412                 if (hasTitle)
1413                         borderidx = 1;
1414                 if (value > 0) {
1415                         if (borders[borderidx].low < value)
1416                                 borders[borderidx].low = value;
1417                         if (borders[borderidx].upper < value)
1418                                 borders[borderidx].upper = value;
1419                 }
1420         }
1421 }
1422
1423 static void checkDepthIndex(int val)
1424 {
1425         static int maxdepthidx = MAXOPENED-2;
1426         static int lastmaxdepth = 0;
1427         if (val > lastmaxdepth) {
1428                 LYXERR(Debug::INFO, "Depth reached " << val);
1429                 lastmaxdepth = val;
1430         }
1431         if (val > maxdepthidx) {
1432                 maxdepthidx = val;
1433                 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1434         }
1435 }
1436
1437 #if 0
1438 // Not needed, because borders are now dynamically expanded
1439 static void checkIgnoreIdx(int val)
1440 {
1441         static int lastmaxignore = -1;
1442         if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1443                 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1444                 lastmaxignore = val;
1445         }
1446 }
1447 #endif
1448
1449 /*
1450  * Expand the region of ignored parts of the input latex string
1451  * The region is only relevant in output()
1452  */
1453 void Intervall::addIntervall(int low, int upper)
1454 {
1455         int idx;
1456         if (low == upper) return;
1457         for (idx = ignoreidx+1; idx > 0; --idx) {
1458                 if (low > borders[idx-1].upper) {
1459                         break;
1460                 }
1461         }
1462         Border br(low, upper);
1463         if (idx > ignoreidx) {
1464                 if (borders.size() <= size_t(idx)) {
1465                         borders.push_back(br);
1466                 }
1467                 else {
1468                         borders[idx] = br;
1469                 }
1470                 ignoreidx = idx;
1471                 // checkIgnoreIdx(ignoreidx);
1472                 return;
1473         }
1474         else {
1475                 // Expand only if one of the new bound is inside the interwall
1476                 // We know here that br.low > borders[idx-1].upper
1477                 if (br.upper < borders[idx].low) {
1478                         // We have to insert at this pos
1479                         if (size_t(ignoreidx+1) >= borders.size()) {
1480                                 borders.push_back(borders[ignoreidx]);
1481                         }
1482                         else {
1483                                 borders[ignoreidx+1] = borders[ignoreidx];
1484                         }
1485                         for (int i = ignoreidx; i > idx; --i) {
1486                                 borders[i] = borders[i-1];
1487                         }
1488                         borders[idx] = br;
1489                         ignoreidx += 1;
1490                         // checkIgnoreIdx(ignoreidx);
1491                         return;
1492                 }
1493                 // Here we know, that we are overlapping
1494                 if (br.low > borders[idx].low)
1495                         br.low = borders[idx].low;
1496                 // check what has to be concatenated
1497                 int count = 0;
1498                 for (int i = idx; i <= ignoreidx; i++) {
1499                         if (br.upper >= borders[i].low) {
1500                                 count++;
1501                                 if (br.upper < borders[i].upper)
1502                                         br.upper = borders[i].upper;
1503                         }
1504                         else {
1505                                 break;
1506                         }
1507                 }
1508                 // count should be >= 1 here
1509                 borders[idx] = br;
1510                 if (count > 1) {
1511                         for (int i = idx + count; i <= ignoreidx; i++) {
1512                                 borders[i-count+1] = borders[i];
1513                         }
1514                         ignoreidx -= count - 1;
1515                         return;
1516                 }
1517         }
1518 }
1519
1520 static void buildaccent(string n, string param, string values)
1521 {
1522         stringstream s(n);
1523         string name;
1524         const char delim = '|';
1525         while (getline(s, name, delim)) {
1526                 size_t start = 0;
1527                 for (char c : param) {
1528                         string key = name + "{" + c + "}";
1529                         // get the corresponding utf8-value
1530                         if ((values[start] & 0xc0) != 0xc0) {
1531                                 // should not happen, utf8 encoding starts at least with 11xxxxxx
1532                                 // but value for '\dot{i}' is 'i', which is ascii
1533                                 if ((values[start] & 0x80) == 0) {
1534                                         // is ascii
1535                                         accents[key] = values.substr(start, 1);
1536                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1537                                 }
1538                                 start++;
1539                                 continue;
1540                         }
1541                         for (int j = 1; ;j++) {
1542                                 if (start + j >= values.size()) {
1543                                         accents[key] = values.substr(start, j);
1544                                         start = values.size() - 1;
1545                                         break;
1546                                 }
1547                                 else if ((values[start+j] & 0xc0) != 0x80) {
1548                                         // This is the first byte of following utf8 char
1549                                         accents[key] = values.substr(start, j);
1550                                         start += j;
1551                                         // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1552                                         break;
1553                                 }
1554                         }
1555                 }
1556         }
1557 }
1558
1559 // Helper function
1560 static string getutf8(unsigned uchar)
1561 {
1562 #define maxc 5
1563         string ret = string();
1564         char c[maxc] = {0};
1565         if (uchar <= 0x7f) {
1566                 c[maxc-1] = uchar & 0x7f;
1567         }
1568         else {
1569                 unsigned char rest = 0x40;
1570                 unsigned char first = 0x80;
1571                 int start = maxc-1;
1572                 for (int i = start; i >=0; --i) {
1573                         if (uchar < rest) {
1574                                 c[i] = first + uchar;
1575                                 break;
1576                         }
1577                         c[i] = 0x80 | (uchar &  0x3f);
1578                         uchar >>= 6;
1579                         rest >>= 1;
1580                         first >>= 1;
1581                         first |= 0x80;
1582                 }
1583         }
1584         for (int i = 0; i < maxc; i++) {
1585                 if (c[i] == 0) continue;
1586                 ret += c[i];
1587         }
1588         return(ret);
1589 }
1590
1591 static void addAccents(string latex_in, string unicode_out)
1592 {
1593         latex_in = latex_in.substr(1);
1594         AccentsIterator it_ac = accents.find(latex_in);
1595         if (it_ac == accents.end()) {
1596                 accents[latex_in] = unicode_out;
1597         }
1598         else {
1599                 LYXERR0("Key " << latex_in  << " already set");
1600         }
1601 }
1602
1603 void static fillMissingUnicodesymbols()
1604 {
1605         addAccents("\\pounds", getutf8(0x00a3));
1606         addAccents("\\textsterling", getutf8(0x00a3));
1607         addAccents("\\textyen", getutf8(0x00a5));
1608         addAccents("\\yen", getutf8(0x00a5));
1609         addAccents("\\textsection", getutf8(0x00a7));
1610         addAccents("\\mathsection", getutf8(0x00a7));
1611         addAccents("\\textcopyright", getutf8(0x00a9));
1612         addAccents("\\copyright", getutf8(0x00a9));
1613         addAccents("\\textlnot", getutf8(0x00ac));
1614         addAccents("\\neg", getutf8(0x00ac));
1615         addAccents("\\textregistered", getutf8(0x00ae));
1616         addAccents("\\circledR", getutf8(0x00ae));
1617         addAccents("\\textpm", getutf8(0x00b1));
1618         addAccents("\\pm", getutf8(0x00b1));
1619         addAccents("\\textparagraph", getutf8(0x00b6));
1620         addAccents("\\mathparagraph", getutf8(0x00b6));
1621         addAccents("\\textperiodcentered", getutf8(0x00b7));
1622         addAccents("\\texttimes", getutf8(0x00d7));
1623         addAccents("\\times", getutf8(0x00d7));
1624         addAccents("\\O", getutf8(0x00d8));
1625         addAccents("\\dh", getutf8(0x00f0));
1626         addAccents("\\eth", getutf8(0x00f0));
1627         addAccents("\\textdiv", getutf8(0x00f7));
1628         addAccents("\\div", getutf8(0x00f7));
1629         addAccents("\\o", getutf8(0x00f8));
1630         addAccents("\\textcrlambda", getutf8(0x019b));
1631         addAccents("\\j", getutf8(0x0237));
1632         addAccents("\\textrevepsilon", getutf8(0x025c));
1633         addAccents("\\textbaru", getutf8(0x0289));
1634         addAccents("\\textquoteleft", getutf8(0x02bb));
1635         addAccents("\\textGamma", getutf8(0x0393));
1636         addAccents("\\Gamma", getutf8(0x0393));
1637         addAccents("\\textDelta", getutf8(0x0394));
1638         addAccents("\\Delta", getutf8(0x0394));
1639         addAccents("\\textTheta", getutf8(0x0398));
1640         addAccents("\\Theta", getutf8(0x0398));
1641         addAccents("\\textLambda", getutf8(0x039b));
1642         addAccents("\\Lambda", getutf8(0x039b));
1643         addAccents("\\textXi", getutf8(0x039e));
1644         addAccents("\\Xi", getutf8(0x039e));
1645         addAccents("\\textPi", getutf8(0x03a0));
1646         addAccents("\\Pi", getutf8(0x03a0));
1647         addAccents("\\textSigma", getutf8(0x03a3));
1648         addAccents("\\Sigma", getutf8(0x03a3));
1649         addAccents("\\textUpsilon", getutf8(0x03a5));
1650         addAccents("\\Upsilon", getutf8(0x03a5));
1651         addAccents("\\textPhi", getutf8(0x03a6));
1652         addAccents("\\Phi", getutf8(0x03a6));
1653         addAccents("\\textPsi", getutf8(0x03a8));
1654         addAccents("\\Psi", getutf8(0x03a8));
1655         addAccents("\\textOmega", getutf8(0x03a9));
1656         addAccents("\\Omega", getutf8(0x03a9));
1657         addAccents("\\textalpha", getutf8(0x03b1));
1658         addAccents("\\alpha", getutf8(0x03b1));
1659         addAccents("\\textbeta", getutf8(0x03b2));
1660         addAccents("\\beta", getutf8(0x03b2));
1661         addAccents("\\textgamma", getutf8(0x03b3));
1662         addAccents("\\gamma", getutf8(0x03b3));
1663         addAccents("\\textdelta", getutf8(0x03b4));
1664         addAccents("\\delta", getutf8(0x03b4));
1665         addAccents("\\textepsilon", getutf8(0x03b5));
1666         addAccents("\\varepsilon", getutf8(0x03b5));
1667         addAccents("\\textzeta", getutf8(0x03b6));
1668         addAccents("\\zeta", getutf8(0x03b6));
1669         addAccents("\\texteta", getutf8(0x03b7));
1670         addAccents("\\eta", getutf8(0x03b7));
1671         addAccents("\\texttheta", getutf8(0x03b8));
1672         addAccents("\\theta", getutf8(0x03b8));
1673         addAccents("\\textiota", getutf8(0x03b9));
1674         addAccents("\\iota", getutf8(0x03b9));
1675         addAccents("\\textkappa", getutf8(0x03ba));
1676         addAccents("\\kappa", getutf8(0x03ba));
1677         addAccents("\\textlambda", getutf8(0x03bb));
1678         addAccents("\\lambda", getutf8(0x03bb));
1679         addAccents("\\textmu", getutf8(0x03bc));
1680         addAccents("\\mu", getutf8(0x03bc));
1681         addAccents("\\textnu", getutf8(0x03bd));
1682         addAccents("\\nu", getutf8(0x03bd));
1683         addAccents("\\textxi", getutf8(0x03be));
1684         addAccents("\\xi", getutf8(0x03be));
1685         addAccents("\\textpi", getutf8(0x03c0));
1686         addAccents("\\pi", getutf8(0x03c0));
1687         addAccents("\\textrho", getutf8(0x03c1));
1688         addAccents("\\rho", getutf8(0x03c1));
1689         addAccents("\\textfinalsigma", getutf8(0x03c2));
1690         addAccents("\\varsigma", getutf8(0x03c2));
1691         addAccents("\\textsigma", getutf8(0x03c3));
1692         addAccents("\\sigma", getutf8(0x03c3));
1693         addAccents("\\texttau", getutf8(0x03c4));
1694         addAccents("\\tau", getutf8(0x03c4));
1695         addAccents("\\textupsilon", getutf8(0x03c5));
1696         addAccents("\\upsilon", getutf8(0x03c5));
1697         addAccents("\\textphi", getutf8(0x03c6));
1698         addAccents("\\varphi", getutf8(0x03c6));
1699         addAccents("\\textchi", getutf8(0x03c7));
1700         addAccents("\\chi", getutf8(0x03c7));
1701         addAccents("\\textpsi", getutf8(0x03c8));
1702         addAccents("\\psi", getutf8(0x03c8));
1703         addAccents("\\textomega", getutf8(0x03c9));
1704         addAccents("\\omega", getutf8(0x03c9));
1705         addAccents("\\textdigamma", getutf8(0x03dd));
1706         addAccents("\\digamma", getutf8(0x03dd));
1707         addAccents("\\hebalef", getutf8(0x05d0));
1708         addAccents("\\aleph", getutf8(0x05d0));
1709         addAccents("\\hebbet", getutf8(0x05d1));
1710         addAccents("\\beth", getutf8(0x05d1));
1711         addAccents("\\hebgimel", getutf8(0x05d2));
1712         addAccents("\\gimel", getutf8(0x05d2));
1713         addAccents("\\hebdalet", getutf8(0x05d3));
1714         addAccents("\\daleth", getutf8(0x05d3));
1715         addAccents("\\hebhe", getutf8(0x05d4));
1716         addAccents("\\hebvav", getutf8(0x05d5));
1717         addAccents("\\hebzayin", getutf8(0x05d6));
1718         addAccents("\\hebhet", getutf8(0x05d7));
1719         addAccents("\\hebtet", getutf8(0x05d8));
1720         addAccents("\\hebyod", getutf8(0x05d9));
1721         addAccents("\\hebfinalkaf", getutf8(0x05da));
1722         addAccents("\\hebkaf", getutf8(0x05db));
1723         addAccents("\\heblamed", getutf8(0x05dc));
1724         addAccents("\\hebfinalmem", getutf8(0x05dd));
1725         addAccents("\\hebmem", getutf8(0x05de));
1726         addAccents("\\hebfinalnun", getutf8(0x05df));
1727         addAccents("\\hebnun", getutf8(0x05e0));
1728         addAccents("\\hebsamekh", getutf8(0x05e1));
1729         addAccents("\\hebayin", getutf8(0x05e2));
1730         addAccents("\\hebfinalpe", getutf8(0x05e3));
1731         addAccents("\\hebpe", getutf8(0x05e4));
1732         addAccents("\\hebfinaltsadi", getutf8(0x05e5));
1733         addAccents("\\hebtsadi", getutf8(0x05e6));
1734         addAccents("\\hebqof", getutf8(0x05e7));
1735         addAccents("\\hebresh", getutf8(0x05e8));
1736         addAccents("\\hebshin", getutf8(0x05e9));
1737         addAccents("\\hebtav", getutf8(0x05ea));
1738
1739         // Thai characters
1740         addAccents("\\thaiKoKai", getutf8(0x0e01));
1741         addAccents("\\thaiKhoKhai", getutf8(0x0e02));
1742         addAccents("\\thaiKhoKhuat", getutf8(0x0e03));
1743         addAccents("\\thaiKhoKhwai", getutf8(0x0e04));
1744         addAccents("\\thaiKhoKhon", getutf8(0x0e05));
1745         addAccents("\\thaiKhoRakhang", getutf8(0x0e06));
1746         addAccents("\\thaiNgoNgu", getutf8(0x0e07));
1747         addAccents("\\thaiChoChan", getutf8(0x0e08));
1748         addAccents("\\thaiChoChing", getutf8(0x0e09));
1749         addAccents("\\thaiChoChang", getutf8(0x0e0a));
1750         addAccents("\\thaiSoSo", getutf8(0x0e0b));
1751         addAccents("\\thaiChoChoe", getutf8(0x0e0c));
1752         addAccents("\\thaiYoYing", getutf8(0x0e0d));
1753         addAccents("\\thaiDoChada", getutf8(0x0e0e));
1754         addAccents("\\thaiToPatak", getutf8(0x0e0f));
1755         addAccents("\\thaiThoThan", getutf8(0x0e10));
1756         addAccents("\\thaiThoNangmontho", getutf8(0x0e11));
1757         addAccents("\\thaiThoPhuthao", getutf8(0x0e12));
1758         addAccents("\\thaiNoNen", getutf8(0x0e13));
1759         addAccents("\\thaiDoDek", getutf8(0x0e14));
1760         addAccents("\\thaiToTao", getutf8(0x0e15));
1761         addAccents("\\thaiThoThung", getutf8(0x0e16));
1762         addAccents("\\thaiThoThahan", getutf8(0x0e17));
1763         addAccents("\\thaiThoThong", getutf8(0x0e18));
1764         addAccents("\\thaiNoNu", getutf8(0x0e19));
1765         addAccents("\\thaiBoBaimai", getutf8(0x0e1a));
1766         addAccents("\\thaiPoPla", getutf8(0x0e1b));
1767         addAccents("\\thaiPhoPhung", getutf8(0x0e1c));
1768         addAccents("\\thaiFoFa", getutf8(0x0e1d));
1769         addAccents("\\thaiPhoPhan", getutf8(0x0e1e));
1770         addAccents("\\thaiFoFan", getutf8(0x0e1f));
1771         addAccents("\\thaiPhoSamphao", getutf8(0x0e20));
1772         addAccents("\\thaiMoMa", getutf8(0x0e21));
1773         addAccents("\\thaiYoYak", getutf8(0x0e22));
1774         addAccents("\\thaiRoRua", getutf8(0x0e23));
1775         addAccents("\\thaiRu", getutf8(0x0e24));
1776         addAccents("\\thaiLoLing", getutf8(0x0e25));
1777         addAccents("\\thaiLu", getutf8(0x0e26));
1778         addAccents("\\thaiWoWaen", getutf8(0x0e27));
1779         addAccents("\\thaiSoSala", getutf8(0x0e28));
1780         addAccents("\\thaiSoRusi", getutf8(0x0e29));
1781         addAccents("\\thaiSoSua", getutf8(0x0e2a));
1782         addAccents("\\thaiHoHip", getutf8(0x0e2b));
1783         addAccents("\\thaiLoChula", getutf8(0x0e2c));
1784         addAccents("\\thaiOAng", getutf8(0x0e2d));
1785         addAccents("\\thaiHoNokhuk", getutf8(0x0e2e));
1786         addAccents("\\thaiPaiyannoi", getutf8(0x0e2f));
1787         addAccents("\\thaiSaraA", getutf8(0x0e30));
1788         addAccents("\\thaiMaiHanakat", getutf8(0x0e31));
1789         addAccents("\\thaiSaraAa", getutf8(0x0e32));
1790         addAccents("\\thaiSaraAm", getutf8(0x0e33));
1791         addAccents("\\thaiSaraI", getutf8(0x0e34));
1792         addAccents("\\thaiSaraIi", getutf8(0x0e35));
1793         addAccents("\\thaiSaraUe", getutf8(0x0e36));
1794         addAccents("\\thaiSaraUee", getutf8(0x0e37));
1795         addAccents("\\thaiSaraU", getutf8(0x0e38));
1796         addAccents("\\thaiSaraUu", getutf8(0x0e39));
1797         addAccents("\\thaiPhinthu", getutf8(0x0e3a));
1798         addAccents("\\thaiSaraE", getutf8(0x0e40));
1799         addAccents("\\thaiSaraAe", getutf8(0x0e41));
1800         addAccents("\\thaiSaraO", getutf8(0x0e42));
1801         addAccents("\\thaiSaraAiMaimuan", getutf8(0x0e43));
1802         addAccents("\\thaiSaraAiMaimalai", getutf8(0x0e44));
1803         addAccents("\\thaiLakkhangyao", getutf8(0x0e45));
1804         addAccents("\\thaiMaiyamok", getutf8(0x0e46));
1805         addAccents("\\thaiMaitaikhu", getutf8(0x0e47));
1806         addAccents("\\thaiMaiEk", getutf8(0x0e48));
1807         addAccents("\\thaiMaiTho", getutf8(0x0e49));
1808         addAccents("\\thaiMaiTri", getutf8(0x0e4a));
1809         addAccents("\\thaiMaiChattawa", getutf8(0x0e4b));
1810         addAccents("\\thaiThanthakhat", getutf8(0x0e4c));
1811         addAccents("\\thaiNikhahit", getutf8(0x0e4d));
1812         addAccents("\\thaiYamakkan", getutf8(0x0e4e));
1813         addAccents("\\thaiFongman", getutf8(0x0e4f));
1814         addAccents("\\thaizero", getutf8(0x0e50));
1815         addAccents("\\thaione", getutf8(0x0e51));
1816         addAccents("\\thaitwo", getutf8(0x0e52));
1817         addAccents("\\thaithree", getutf8(0x0e53));
1818         addAccents("\\thaifour", getutf8(0x0e54));
1819         addAccents("\\thaifive", getutf8(0x0e55));
1820         addAccents("\\thaisix", getutf8(0x0e56));
1821         addAccents("\\thaiseven", getutf8(0x0e57));
1822         addAccents("\\thaieight", getutf8(0x0e58));
1823         addAccents("\\thainine", getutf8(0x0e59));
1824         addAccents("\\thaiAngkhankhu", getutf8(0x0e5a));
1825         addAccents("\\thaiKhomut", getutf8(0x0e5b));
1826         addAccents("\\dag", getutf8(0x2020));
1827         addAccents("\\dagger", getutf8(0x2020));
1828         addAccents("\\textdagger", getutf8(0x2020));
1829         addAccents("\\ddag", getutf8(0x2021));
1830         addAccents("\\ddagger", getutf8(0x2021));
1831         addAccents("\\textdaggerdbl", getutf8(0x2021));
1832         addAccents("\\textbullet", getutf8(0x2022));
1833         addAccents("\\bullet", getutf8(0x2022));
1834         addAccents("\\dots", getutf8(0x2026));
1835         addAccents("\\ldots", getutf8(0x2026));
1836         addAccents("\\textellipsis", getutf8(0x2026));
1837         addAccents("\\textasciiacute", getutf8(0x2032));
1838         addAccents("\\prime", getutf8(0x2032));
1839         addAccents("\\textacutedbl", getutf8(0x2033));
1840         addAccents("\\dprime", getutf8(0x2033));
1841         addAccents("\\textasciigrave", getutf8(0x2035));
1842         addAccents("\\backprime", getutf8(0x2035));
1843         addAccents("\\textsubcircum{ }", getutf8(0x2038));
1844         addAccents("\\caretinsert", getutf8(0x2038));
1845         addAccents("\\textasteriskcentered", getutf8(0x204e));
1846         addAccents("\\ast", getutf8(0x204e));
1847         addAccents("\\textmho", getutf8(0x2127));
1848         addAccents("\\mho", getutf8(0x2127));
1849         addAccents("\\textleftarrow", getutf8(0x2190));
1850         addAccents("\\leftarrow", getutf8(0x2190));
1851         addAccents("\\textuparrow", getutf8(0x2191));
1852         addAccents("\\uparrow", getutf8(0x2191));
1853         addAccents("\\textrightarrow", getutf8(0x2192));
1854         addAccents("\\rightarrow", getutf8(0x2192));
1855         addAccents("\\textdownarrow", getutf8(0x2193));
1856         addAccents("\\downarrow", getutf8(0x2193));
1857         addAccents("\\textglobrise", getutf8(0x2197));
1858         addAccents("\\nearrow", getutf8(0x2197));
1859         addAccents("\\textglobfall", getutf8(0x2198));
1860         addAccents("\\searrow", getutf8(0x2198));
1861         addAccents("\\textsurd", getutf8(0x221a));
1862         addAccents("\\surd", getutf8(0x221a));
1863         addAccents("\\textbigcircle", getutf8(0x25ef));
1864         addAccents("\\bigcirc", getutf8(0x25ef));
1865         addAccents("\\FiveStar", getutf8(0x2605));
1866         addAccents("\\bigstar", getutf8(0x2605));
1867         addAccents("\\FiveStarOpen", getutf8(0x2606));
1868         addAccents("\\bigwhitestar", getutf8(0x2606));
1869         addAccents("\\Checkmark", getutf8(0x2713));
1870         addAccents("\\checkmark", getutf8(0x2713));
1871         addAccents("\\CrossMaltese", getutf8(0x2720));
1872         addAccents("\\maltese", getutf8(0x2720));
1873         addAccents("\\textlangle", getutf8(0x27e8));
1874         addAccents("\\langle", getutf8(0x27e8));
1875         addAccents("\\textrangle", getutf8(0x27e9));
1876         addAccents("\\rangle", getutf8(0x27e9));
1877 }
1878
1879 static void buildAccentsMap()
1880 {
1881         accents["imath"] = "ı";
1882         accents["i"] = "ı";
1883         accents["jmath"] = "ȷ";
1884         accents["cdot"] = "·";
1885         accents["textasciicircum"] = "^";
1886         accents["mathcircumflex"] = "^";
1887         accents["sim"] = "~";
1888         accents["guillemotright"] = "»";
1889         accents["guillemotleft"] = "«";
1890         accents["hairspace"]     = getutf8(0xf0000);    // select from free unicode plane 15
1891         accents["thinspace"]     = getutf8(0xf0002);    // and used _only_ by findadv
1892         accents["negthinspace"]  = getutf8(0xf0003);    // to omit backslashed latex macros
1893         accents["medspace"]      = getutf8(0xf0004);    // See https://en.wikipedia.org/wiki/Private_Use_Areas
1894         accents["negmedspace"]   = getutf8(0xf0005);
1895         accents["thickspace"]    = getutf8(0xf0006);
1896         accents["negthickspace"] = getutf8(0xf0007);
1897         accents["lyx"]           = getutf8(0xf0010);    // Used logos
1898         accents["LyX"]           = getutf8(0xf0010);
1899         accents["tex"]           = getutf8(0xf0011);
1900         accents["TeX"]           = getutf8(0xf0011);
1901         accents["latex"]         = getutf8(0xf0012);
1902         accents["LaTeX"]         = getutf8(0xf0012);
1903         accents["latexe"]        = getutf8(0xf0013);
1904         accents["LaTeXe"]        = getutf8(0xf0013);
1905         accents["lyxarrow"]      = getutf8(0xf0020);
1906         accents["braceleft"]     = getutf8(0xf0030);
1907         accents["braceright"]    = getutf8(0xf0031);
1908         accents["backslash lyx"]           = getutf8(0xf0010);  // Used logos inserted with starting \backslash
1909         accents["backslash LyX"]           = getutf8(0xf0010);
1910         accents["backslash tex"]           = getutf8(0xf0011);
1911         accents["backslash TeX"]           = getutf8(0xf0011);
1912         accents["backslash latex"]         = getutf8(0xf0012);
1913         accents["backslash LaTeX"]         = getutf8(0xf0012);
1914         accents["backslash latexe"]        = getutf8(0xf0013);
1915         accents["backslash LaTeXe"]        = getutf8(0xf0013);
1916         accents["backslash lyxarrow"]      = getutf8(0xf0020);
1917         accents["ddot{\\imath}"] = "ï";
1918         buildaccent("ddot", "aAeEhHiIoOtuUwWxXyY",
1919                     "äÄëËḧḦïÏöÖẗüÜẅẄẍẌÿŸ");   // umlaut
1920         buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1921                     "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ");      // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1922         accents["acute{\\imath}"] = "í";
1923         buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1924                     "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1925         buildaccent("dacute|H|h", "oOuU", "őŐűŰ");  // double acute
1926         buildaccent("mathring|r", "aAuUwy",
1927                     "åÅůŮẘẙ");  // ring
1928         accents["check{\\imath}"] = "ǐ";
1929         accents["check{\\jmath}"] = "ǰ";
1930         buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1931                     "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ");        // caron
1932         accents["hat{\\imath}"] = "î";
1933         accents["hat{\\jmath}"] = "ĵ";
1934         buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1935                     "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ");  // circ
1936         accents["bar{\\imath}"] = "ī";
1937         buildaccent("bar|=", "aAeEiIoOuUyY",
1938                     "āĀēĒīĪōŌūŪȳȲ");        // macron
1939         accents["tilde{\\imath}"] = "ĩ";
1940         buildaccent("tilde", "aAeEiInNoOuUvVyY",
1941                     "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ");  // tilde
1942         accents["breve{\\imath}"] = "ĭ";
1943         buildaccent("breve|u", "aAeEgGiIoOuU",
1944                     "ăĂĕĔğĞĭĬŏŎŭŬ");        // breve
1945         accents["grave{\\imath}"] = "ì";
1946         buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1947                     "àÀèÈìÌòÒùÙǹǸẁẀỳỲ");    // grave
1948         buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1949                     "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ");      // dot below
1950         buildaccent("ogonek|k", "AaEeIiUuOo",
1951                     "ĄąĘęĮįŲųǪǫ");    // ogonek
1952         buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1953                     "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ");        // cedilla
1954         buildaccent("subring|textsubring", "Aa",
1955                     "Ḁḁ");  // subring
1956         buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1957                     "ḒḓḘḙḼḽṊṋṰṱṶṷ");    // subcircum
1958         buildaccent("subtilde|textsubtilde", "EeIiUu",
1959                     "ḚḛḬḭṴṵ");      // subtilde
1960         accents["dgrave{\\imath}"] = "ȉ";
1961         accents["textdoublegrave{\\i}"] = "ȉ";
1962         buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1963                     "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1964         accents["rcap{\\imath}"] = "ȋ";
1965         accents["textroundcap{\\i}"] = "ȋ";
1966         buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1967                     "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1968         buildaccent("slashed", "oO",
1969                     "øØ"); // slashed
1970         fillMissingUnicodesymbols(); // Add some still not handled entries contained in 'unicodesynbols'
1971         // LYXERR0("Number of accents " << accents.size());
1972 }
1973
1974 /*
1975  * Created accents in math or regexp environment
1976  * are macros, but we need the utf8 equivalent
1977  */
1978 void Intervall::removeAccents()
1979 {
1980         if (accents.empty())
1981                 buildAccentsMap();
1982         static regex const accre("\\\\("
1983                                  "([\\S]|[A-Za-z]+)\\{[^\\{\\}]+\\}"
1984                                  "|("
1985                                  "(backslash ([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))"
1986                                  "|[A-Za-z]+"
1987                                  ")"
1988                                  "(?![a-zA-Z]))");
1989         smatch sub;
1990         for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1991                 sub = *itacc;
1992                 string key = sub.str(1);
1993                 AccentsIterator it_ac = accents.find(key);
1994                 if (it_ac != accents.end()) {
1995                         string val = it_ac->second;
1996                         size_t pos = sub.position(size_t(0));
1997                         for (size_t i = 0; i < val.size(); i++) {
1998                                 par[pos+i] = val[i];
1999                         }
2000                         // Remove possibly following space too
2001                         if (par[pos+sub.str(0).size()] == ' ')
2002                                 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
2003                         else
2004                                 addIntervall(pos+val.size(), pos + sub.str(0).size());
2005                         for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
2006                                 // remove traces of any remaining chars
2007                                 par[i] = ' ';
2008                         }
2009                 }
2010                 else {
2011                         LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
2012                 }
2013         }
2014 }
2015
2016 void Intervall::handleOpenP(int i)
2017 {
2018         actualdeptindex++;
2019         depts[actualdeptindex] = i+1;
2020         closes[actualdeptindex] = -1;
2021         checkDepthIndex(actualdeptindex);
2022 }
2023
2024 void Intervall::handleCloseP(int i, bool closingAllowed)
2025 {
2026         if (actualdeptindex <= 0) {
2027                 if (! closingAllowed)
2028                         LYXERR(Debug::FINDVERBOSE, "Bad closing parenthesis in latex");  /* should not happen, but the latex input may be wrong */
2029                 // if we are at the very end
2030                 addIntervall(i, i+1);
2031         }
2032         else {
2033                 closes[actualdeptindex] = i+1;
2034                 actualdeptindex--;
2035         }
2036 }
2037
2038 void Intervall::resetOpenedP(int openPos)
2039 {
2040         // Used as initializer for foreignlanguage entry
2041         actualdeptindex = 1;
2042         depts[1] = openPos+1;
2043         closes[1] = -1;
2044 }
2045
2046 int Intervall::previousNotIgnored(int start) const
2047 {
2048         int idx = 0;                          /* int intervalls */
2049         for (idx = ignoreidx; idx >= 0; --idx) {
2050                 if (start > borders[idx].upper)
2051                         return start;
2052                 if (start >= borders[idx].low)
2053                         start = borders[idx].low-1;
2054         }
2055         return start;
2056 }
2057
2058 int Intervall::nextNotIgnored(int start) const
2059 {
2060         int idx = 0;                          /* int intervalls */
2061         for (idx = 0; idx <= ignoreidx; idx++) {
2062                 if (start < borders[idx].low)
2063                         return start;
2064                 if (start < borders[idx].upper)
2065                         start = borders[idx].upper;
2066         }
2067         return start;
2068 }
2069
2070 typedef unordered_map<string, KeyInfo> KeysMap;
2071 typedef unordered_map<string, KeyInfo>::const_iterator KeysIterator;
2072 typedef vector< KeyInfo> Entries;
2073 static KeysMap keys = unordered_map<string, KeyInfo>();
2074
2075 class LatexInfo {
2076 private:
2077         int entidx_;
2078         Entries entries_;
2079         Intervall interval_;
2080         void buildKeys(bool);
2081         void buildEntries(bool);
2082         void makeKey(const string &, KeyInfo, bool isPatternString);
2083         void processRegion(int start, int region_end); /*  remove {} parts */
2084         void removeHead(KeyInfo const &, int count=0);
2085
2086 public:
2087         LatexInfo(string const & par, bool isPatternString)
2088                 : entidx_(-1), interval_(isPatternString, par)
2089         {
2090                 buildKeys(isPatternString);
2091                 entries_ = vector<KeyInfo>();
2092                 buildEntries(isPatternString);
2093         }
2094         int getFirstKey() {
2095                 entidx_ = 0;
2096                 if (entries_.empty()) {
2097                         return -1;
2098                 }
2099                 if (entries_[0].keytype == KeyInfo::isTitle) {
2100                         interval_.hasTitle = true;
2101                         if (! entries_[0].disabled) {
2102                                 interval_.titleValue = entries_[0].head;
2103                         }
2104                         else {
2105                                 interval_.titleValue = "";
2106                         }
2107                         removeHead(entries_[0]);
2108                         if (entries_.size() > 1)
2109                                 return 1;
2110                         else
2111                                 return -1;
2112                 }
2113                 return 0;
2114         }
2115         int getNextKey() {
2116                 entidx_++;
2117                 if (int(entries_.size()) > entidx_) {
2118                         return entidx_;
2119                 }
2120                 else {
2121                         return -1;
2122                 }
2123         }
2124         bool setNextKey(int idx) {
2125                 if ((idx == entidx_) && (entidx_ >= 0)) {
2126                         entidx_--;
2127                         return true;
2128                 }
2129                 else
2130                         return false;
2131         }
2132         int find(int start, KeyInfo::KeyType keytype) const {
2133                 if (start < 0)
2134                         return -1;
2135                 int tmpIdx = start;
2136                 while (tmpIdx < int(entries_.size())) {
2137                         if (entries_[tmpIdx].keytype == keytype)
2138                                 return tmpIdx;
2139                         tmpIdx++;
2140                 }
2141                 return -1;
2142         }
2143         int process(ostringstream & os, KeyInfo const & actual);
2144         int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
2145         // string show(int lastpos) { return interval.show(lastpos);}
2146         int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);}
2147         KeyInfo &getKeyInfo(int keyinfo) {
2148                 static KeyInfo invalidInfo = KeyInfo();
2149                 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
2150                         return invalidInfo;
2151                 else
2152                         return entries_[keyinfo];
2153         }
2154         void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);}
2155         void addIntervall(int low, int up) { interval_.addIntervall(low, up); }
2156 };
2157
2158
2159 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
2160 {
2161         int skip = 0;
2162         int depth = 0;
2163         for (int i = start; i < end; i += 1 + skip) {
2164                 char c;
2165                 c = par[i];
2166                 skip = 0;
2167                 if (c == '\\') skip = 1;
2168                 else if (c == up) {
2169                         depth++;
2170                 }
2171                 else if (c == down) {
2172                         if (depth == 0) {
2173                                 repeat--;
2174                                 if ((repeat <= 0) || (par[i+1] != up))
2175                                         return i;
2176                         }
2177                         --depth;
2178                 }
2179         }
2180         return end;
2181 }
2182
2183 class MathInfo {
2184         class MathEntry {
2185         public:
2186                 string wait;
2187                 size_t mathEnd;
2188                 size_t mathpostfixsize;
2189                 size_t mathStart;
2190                 size_t mathprefixsize;
2191                 size_t mathSize;
2192         };
2193         size_t actualIdx_;
2194         vector<MathEntry> entries_;
2195 public:
2196         MathInfo() {
2197                 actualIdx_ = 0;
2198         }
2199         void insert(string const & wait, size_t start, size_t prefixsize, size_t end, size_t postfixsize) {
2200                 MathEntry m = MathEntry();
2201                 m.wait = wait;
2202                 m.mathStart = start;
2203                 m.mathprefixsize = prefixsize;
2204                 m.mathEnd = end + postfixsize;
2205                 m.mathpostfixsize = postfixsize;
2206                 m.mathSize = m.mathEnd - m.mathStart;
2207                 entries_.push_back(m);
2208         }
2209         bool empty() const { return entries_.empty(); }
2210         size_t getEndPos() const {
2211                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2212                         return 0;
2213                 }
2214                 return entries_[actualIdx_].mathEnd;
2215         }
2216         size_t getStartPos() const {
2217                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2218                         return 100000;                    /*  definitely enough? */
2219                 }
2220                 return entries_[actualIdx_].mathStart;
2221         }
2222         size_t getPrefixSize() const {
2223                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2224                         return 0;
2225                 }
2226                 return entries_[actualIdx_].mathprefixsize;
2227         }
2228         size_t getPostfixSize() const {
2229                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2230                         return 0;
2231                 }
2232                 return entries_[actualIdx_].mathpostfixsize;
2233         }
2234         size_t getFirstPos() {
2235                 actualIdx_ = 0;
2236                 return getStartPos();
2237         }
2238         size_t getSize() const {
2239                 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
2240                         return size_t(0);
2241                 }
2242                 return entries_[actualIdx_].mathSize;
2243         }
2244         void incrEntry() { actualIdx_++; }
2245 };
2246
2247 void LatexInfo::buildEntries(bool isPatternString)
2248 {
2249         static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|x?x?alignat)\\*?\\})(\\{[0-9]+\\})?)");
2250         static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
2251         static bool disableLanguageOverride = false;
2252         smatch sub, submath;
2253         bool evaluatingRegexp = false;
2254         MathInfo mi;
2255         bool evaluatingMath = false;
2256         bool evaluatingCode = false;
2257         size_t codeEnd = 0;
2258         bool evaluatingOptional = false;
2259         size_t optionalEnd = 0;
2260         int codeStart = -1;
2261         KeyInfo found;
2262         bool math_end_waiting = false;
2263         size_t math_pos = 10000;
2264         size_t math_prefix_size = 1;
2265         string math_end;
2266         static vector<string> usedText = vector<string>();
2267         static bool removeMathHull = false;
2268
2269         interval_.removeAccents();
2270
2271         for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
2272                 submath = *itmath;
2273                 if ((submath.position(2) - submath.position(0)) %2 == 1) {
2274                         // prefixed by odd count of '\\'
2275                         continue;
2276                 }
2277                 if (math_end_waiting) {
2278                         size_t pos = submath.position(size_t(2));
2279                         if ((math_end == "$") &&
2280                                         (submath.str(2) == "$")) {
2281                                 mi.insert("$", math_pos, 1, pos, 1);
2282                                 math_end_waiting = false;
2283                         }
2284                         else if ((math_end == "\\]") &&
2285                                  (submath.str(2) == "\\]")) {
2286                                 mi.insert("\\]", math_pos, 2, pos, 2);
2287                                 math_end_waiting = false;
2288                         }
2289                         else if ((submath.str(3).compare("end") == 0) &&
2290                                  (submath.str(5).compare(math_end) == 0)) {
2291                                 mi.insert(math_end, math_pos, math_prefix_size, pos, submath.str(2).length());
2292                                 math_end_waiting = false;
2293                         }
2294                         else
2295                                 continue;
2296                 }
2297                 else {
2298                         if (submath.str(3).compare("begin") == 0) {
2299                                 math_end_waiting = true;
2300                                 math_end = submath.str(5);
2301                                 math_pos = submath.position(size_t(2));
2302                                 math_prefix_size = submath.str(2).length();
2303                         }
2304                         else if (submath.str(2).compare("\\[") == 0) {
2305                                 math_end_waiting = true;
2306                                 math_end = "\\]";
2307                                 math_pos = submath.position(size_t(2));
2308                         }
2309                         else if (submath.str(2) == "$") {
2310                                 size_t pos = submath.position(size_t(2));
2311                                 math_end_waiting = true;
2312                                 math_end = "$";
2313                                 math_pos = pos;
2314                         }
2315                 }
2316         }
2317         // Ignore language if there is math somewhere in pattern-string
2318         if (isPatternString) {
2319                 for (auto s: usedText) {
2320                         // Remove entries created in previous search runs
2321                         keys.erase(s);
2322                 }
2323                 usedText = vector<string>();
2324                 if (! mi.empty()) {
2325                         // Disable language
2326                         keys["foreignlanguage"].disabled = true;
2327                         disableLanguageOverride = true;
2328                         removeMathHull = false;
2329                 }
2330                 else {
2331                         removeMathHull = true;  // used later if not isPatternString
2332                         disableLanguageOverride = false;
2333                 }
2334         }
2335         else {
2336                 if (disableLanguageOverride) {
2337                         keys["foreignlanguage"].disabled = true;
2338                 }
2339         }
2340         math_pos = mi.getFirstPos();
2341         for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
2342                 sub = *it;
2343                 if ((sub.position(2) - sub.position(0)) %2 == 1) {
2344                         // prefixed by odd count of '\\'
2345                         continue;
2346                 }
2347                 string key = sub.str(5);
2348                 if (key == "") {
2349                         if (sub.str(2)[0] == '\\')
2350                                 key = sub.str(2)[1];
2351                         else {
2352                                 key = sub.str(2);
2353                         }
2354                 }
2355                 KeysIterator it_key = keys.find(key);
2356                 if (it_key != keys.end()) {
2357                         if (it_key->second.keytype == KeyInfo::headRemove) {
2358                                 KeyInfo found1 = it_key->second;
2359                                 found1.disabled = true;
2360                                 found1.head = "\\" + key + "{";
2361                                 found1._tokenstart = sub.position(size_t(2));
2362                                 found1._tokensize = found1.head.length();
2363                                 found1._dataStart = found1._tokenstart + found1.head.length();
2364                                 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
2365                                 found1._dataEnd = endpos;
2366                                 removeHead(found1);
2367                                 continue;
2368                         }
2369                 }
2370                 if (evaluatingRegexp) {
2371                         if (sub.str(3).compare("endregexp") == 0) {
2372                                 evaluatingRegexp = false;
2373                                 // found._tokenstart already set
2374                                 found._dataEnd = sub.position(size_t(2)) + 13;
2375                                 found._dataStart = found._dataEnd;
2376                                 found._tokensize = found._dataEnd - found._tokenstart;
2377                                 found.parenthesiscount = 0;
2378                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2379                         }
2380                         else {
2381                                 continue;
2382                         }
2383                 }
2384                 else {
2385                         if (evaluatingMath) {
2386                                 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
2387                                         continue;
2388                                 evaluatingMath = false;
2389                                 mi.incrEntry();
2390                                 math_pos = mi.getStartPos();
2391                         }
2392                         if (it_key == keys.end()) {
2393                                 found = KeyInfo(KeyInfo::isStandard, 0, true);
2394                                 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
2395                                 found = KeyInfo(KeyInfo::isText, 0, false);
2396                                 if (isPatternString) {
2397                                         found.keytype = KeyInfo::isChar;
2398                                         found.disabled = false;
2399                                         found.used = true;
2400                                 }
2401                                 keys[key] = found;
2402                                 usedText.push_back(key);
2403                         }
2404                         else
2405                                 found = keys[key];
2406                         if (key.compare("regexp") == 0) {
2407                                 evaluatingRegexp = true;
2408                                 found._tokenstart = sub.position(size_t(2));
2409                                 found._tokensize = 0;
2410                                 continue;
2411                         }
2412                 }
2413                 // Handle the other params of key
2414                 if (found.keytype == KeyInfo::isIgnored)
2415                         continue;
2416                 else if (found.keytype == KeyInfo::isMath) {
2417                         if (size_t(sub.position(size_t(2))) == math_pos) {
2418                                 found = keys[key];
2419                                 found._tokenstart = sub.position(size_t(2));
2420                                 found._tokensize = mi.getSize();
2421                                 found._dataEnd = found._tokenstart + found._tokensize;
2422                                 found._dataStart = found._dataEnd;
2423                                 found.parenthesiscount = 0;
2424                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2425                                 if (removeMathHull) {
2426                                         interval_.addIntervall(found._tokenstart, found._tokenstart + mi.getPrefixSize());
2427                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize(), found._dataEnd);
2428                                 }
2429                                 else {
2430                                         // Treate all math constructs as simple math
2431                                         interval_.par[found._tokenstart] = '$';
2432                                         interval_.par[found._dataEnd - mi.getPostfixSize()] = '$';
2433                                         interval_.addIntervall(found._tokenstart + 1, found._tokenstart + mi.getPrefixSize());
2434                                         interval_.addIntervall(found._dataEnd - mi.getPostfixSize() + 1, found._dataEnd);
2435                                 }
2436                                 evaluatingMath = true;
2437                         }
2438                         else {
2439                                 // begin|end of unknown env, discard
2440                                 // First handle tables
2441                                 // longtable|tabular
2442                                 bool discardComment;
2443                                 found = keys[key];
2444                                 found.keytype = KeyInfo::doRemove;
2445                                 if ((sub.str(7).compare("longtable") == 0) ||
2446                                                 (sub.str(7).compare("tabular") == 0)) {
2447                                         discardComment = true;        /* '%' */
2448                                 }
2449                                 else {
2450                                         discardComment = false;
2451                                         static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
2452                                         smatch sub2;
2453                                         string token = sub.str(7);
2454                                         if (regex_match(token, sub2, removeArgs)) {
2455                                                 found.keytype = KeyInfo::removeWithArg;
2456                                         }
2457                                 }
2458                                 // discard spaces before pos(2)
2459                                 int pos = sub.position(size_t(2));
2460                                 int count;
2461                                 for (count = 0; pos - count > 0; count++) {
2462                                         char c = interval_.par[pos-count-1];
2463                                         if (discardComment) {
2464                                                 if ((c != ' ') && (c != '%'))
2465                                                         break;
2466                                         }
2467                                         else if (c != ' ')
2468                                                 break;
2469                                 }
2470                                 found._tokenstart = pos - count;
2471                                 if (sub.str(3).compare(0, 5, "begin") == 0) {
2472                                         size_t pos1 = pos + sub.str(2).length();
2473                                         if (sub.str(7).compare("cjk") == 0) {
2474                                                 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2475                                                 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2476                                                         pos1 += 2;
2477                                                 found.keytype = KeyInfo::isMain;
2478                                                 found._dataStart = pos1;
2479                                                 found._dataEnd = interval_.par.length();
2480                                                 found.disabled = keys["foreignlanguage"].disabled;
2481                                                 found.used = keys["foreignlanguage"].used;
2482                                                 found._tokensize = pos1 - found._tokenstart;
2483                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2484                                         }
2485                                         else {
2486                                                 // Swallow possible optional params
2487                                                 while (interval_.par[pos1] == '[') {
2488                                                         pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2489                                                 }
2490                                                 // Swallow also the eventual parameter
2491                                                 if (interval_.par[pos1] == '{') {
2492                                                         found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2493                                                 }
2494                                                 else {
2495                                                         found._dataEnd = pos1;
2496                                                 }
2497                                                 found._dataStart = found._dataEnd;
2498                                                 found._tokensize = count + found._dataEnd - pos;
2499                                                 found.parenthesiscount = 0;
2500                                                 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2501                                                 found.disabled = true;
2502                                         }
2503                                 }
2504                                 else {
2505                                         // Handle "\end{...}"
2506                                         found._dataStart = pos + sub.str(2).length();
2507                                         found._dataEnd = found._dataStart;
2508                                         found._tokensize = count + found._dataEnd - pos;
2509                                         found.parenthesiscount = 0;
2510                                         found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2511                                         found.disabled = true;
2512                                 }
2513                         }
2514                 }
2515                 else if (found.keytype != KeyInfo::isRegex) {
2516                         found._tokenstart = sub.position(size_t(2));
2517                         if (found.parenthesiscount == 0) {
2518                                 // Probably to be discarded
2519                                 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2520                                 char following = interval_.par[following_pos];
2521                                 if (following == ' ')
2522                                         found.head = "\\" + sub.str(5) + " ";
2523                                 else if (following == '=') {
2524                                         // like \uldepth=1000pt
2525                                         found.head = sub.str(2);
2526                                 }
2527                                 else
2528                                         found.head = "\\" + key;
2529                                 found._tokensize = found.head.length();
2530                                 found._dataEnd = found._tokenstart + found._tokensize;
2531                                 found._dataStart = found._dataEnd;
2532                         }
2533                         else {
2534                                 int params = found._tokenstart + key.length() + 1;
2535                                 if (evaluatingOptional) {
2536                                         if (size_t(found._tokenstart) > optionalEnd) {
2537                                                 evaluatingOptional = false;
2538                                         }
2539                                         else {
2540                                                 found.disabled = true;
2541                                         }
2542                                 }
2543                                 int optend = params;
2544                                 while (interval_.par[optend] == '[') {
2545                                         // discard optional parameters
2546                                         optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2547                                 }
2548                                 if (optend > params) {
2549                                         key += interval_.par.substr(params, optend-params);
2550                                         evaluatingOptional = true;
2551                                         optionalEnd = optend;
2552                                         if (found.keytype == KeyInfo::isSectioning) {
2553                                                 // Remove optional values (but still keep in header)
2554                                                 interval_.addIntervall(params, optend);
2555                                         }
2556                                 }
2557                                 string token = sub.str(7);
2558                                 int closings;
2559                                 if (interval_.par[optend] != '{') {
2560                                         closings = 0;
2561                                         found.parenthesiscount = 0;
2562                                         found.head = "\\" + key;
2563                                 }
2564                                 else
2565                                         closings = found.parenthesiscount;
2566                                 if (found.parenthesiscount == 1) {
2567                                         found.head = "\\" + key + "{";
2568                                 }
2569                                 else if (found.parenthesiscount > 1) {
2570                                         if (token != "") {
2571                                                 found.head = sub.str(2) + "{";
2572                                                 closings = found.parenthesiscount - 1;
2573                                         }
2574                                         else {
2575                                                 found.head = "\\" + key + "{";
2576                                         }
2577                                 }
2578                                 found._tokensize = found.head.length();
2579                                 found._dataStart = found._tokenstart + found.head.length();
2580                                 if (found.keytype == KeyInfo::doRemove) {
2581                                         if (closings > 0) {
2582                                                 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2583                                                 if (endpar >= interval_.par.length())
2584                                                         found._dataStart = interval_.par.length();
2585                                                 else
2586                                                         found._dataStart = endpar;
2587                                                 found._tokensize = found._dataStart - found._tokenstart;
2588                                         }
2589                                         else {
2590                                                 found._dataStart = found._tokenstart + found._tokensize;
2591                                         }
2592                                         closings = 0;
2593                                 }
2594                                 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2595                                         found._dataStart += 15;
2596                                 }
2597                                 size_t endpos;
2598                                 if (closings < 1)
2599                                         endpos = found._dataStart - 1;
2600                                 else
2601                                         endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2602                                 if (found.keytype == KeyInfo::isList) {
2603                                         // Check if it really is list env
2604                                         static regex const listre("^([a-z]+)$");
2605                                         smatch sub2;
2606                                         if (!regex_match(token, sub2, listre)) {
2607                                                 // Change the key of this entry. It is not in a list/item environment
2608                                                 found.keytype = KeyInfo::endArguments;
2609                                         }
2610                                 }
2611                                 if (found.keytype == KeyInfo::noMain) {
2612                                         evaluatingCode = true;
2613                                         codeEnd = endpos;
2614                                         codeStart = found._dataStart;
2615                                 }
2616                                 else if (evaluatingCode) {
2617                                         if (size_t(found._dataStart) > codeEnd)
2618                                                 evaluatingCode = false;
2619                                         else if (found.keytype == KeyInfo::isMain) {
2620                                                 // Disable this key, treate it as standard
2621                                                 found.keytype = KeyInfo::isStandard;
2622                                                 found.disabled = true;
2623                                                 if ((codeEnd +1 >= interval_.par.length()) &&
2624                                                                 (found._tokenstart == codeStart)) {
2625                                                         // trickery, because the code inset starts
2626                                                         // with \selectlanguage ...
2627                                                         codeEnd = endpos;
2628                                                         if (entries_.size() > 1) {
2629                                                                 entries_[entries_.size()-1]._dataEnd = codeEnd;
2630                                                         }
2631                                                 }
2632                                         }
2633                                 }
2634                                 if ((endpos == interval_.par.length()) &&
2635                                                 (found.keytype == KeyInfo::doRemove)) {
2636                                         // Missing closing => error in latex-input?
2637                                         // therefore do not delete remaining data
2638                                         found._dataStart -= 1;
2639                                         found._dataEnd = found._dataStart;
2640                                 }
2641                                 else
2642                                         found._dataEnd = endpos;
2643                         }
2644                         if (isPatternString) {
2645                                 keys[key].used = true;
2646                         }
2647                 }
2648                 entries_.push_back(found);
2649         }
2650 }
2651
2652 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2653 {
2654         stringstream s(keysstring);
2655         string key;
2656         const char delim = '|';
2657         while (getline(s, key, delim)) {
2658                 KeyInfo keyII(keyI);
2659                 if (isPatternString) {
2660                         keyII.used = false;
2661                 }
2662                 else if ( !keys[key].used)
2663                         keyII.disabled = true;
2664                 keys[key] = keyII;
2665         }
2666 }
2667
2668 void LatexInfo::buildKeys(bool isPatternString)
2669 {
2670
2671         static bool keysBuilt = false;
2672         if (keysBuilt && !isPatternString) return;
2673
2674         // Keys to ignore in any case
2675         makeKey("text|textcyrillic|lyxmathsym|ensuremath", KeyInfo(KeyInfo::headRemove, 1, true), true);
2676         // Known standard keys with 1 parameter.
2677         // Split is done, if not at start of region
2678         makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2679         makeKey("textbf",               KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2680         makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2681         makeKey("uuline|uline|uwave",   KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2682         makeKey("emph|noun",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2683         makeKey("sout|xout",            KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2684
2685         makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2686                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2687         makeKey("section*|subsection*|subsubsection*|paragraph*",
2688                 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2689         makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2690         makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2691         // Regex
2692         makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2693
2694         // Split is done, if not at start of region
2695         makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2696         makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2697
2698         // Split is done always.
2699         makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2700
2701         // Known charaters
2702         // No split
2703         makeKey("backslash|textbackslash|slash",  KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2704         makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2705         makeKey("textasciiacute|texemdash",       KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2706         makeKey("dots|ldots",                     KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2707         // Spaces
2708         makeKey("quad|qquad|hfill|dotfill",               KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2709         makeKey("textvisiblespace|nobreakspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2710         makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2711         makeKey("thickspace|medspace|thinspace",          KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2712         // Skip
2713         // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2714         // Custom space/skip, remove the content (== length value)
2715         makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2716         // Found in fr/UserGuide.lyx
2717         makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2718         // quotes
2719         makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2720         makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2721         // Known macros to remove (including their parameter)
2722         // No split
2723         makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2724         makeKey("addtocounter|setlength",                 KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2725         // handle like standard keys with 1 parameter.
2726         makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2727
2728         // Ignore deleted text
2729         makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2730         // but preserve added text
2731         makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2732
2733         // Macros to remove, but let the parameter survive
2734         // No split
2735         makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2736
2737         // Remove language spec from content of these insets
2738         makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2739
2740         // Same effect as previous, parameter will survive (because there is no one anyway)
2741         // No split
2742         makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2743         // Remove table decorations
2744         makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2745         // Discard shape-header.
2746         // For footnote or shortcut too, because of lang settings
2747         // and wrong handling if used 'KeyInfo::noMain'
2748         makeKey("circlepar|diamondpar|heartpar|nutpar",  KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2749         makeKey("trianglerightpar|hexagonpar|starpar",   KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2750         makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2751         makeKey("triangleleftpar|shapepar|dropuppar",    KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2752         makeKey("hphantom|vphantom|note|footnote|shortcut|include|includegraphics",     KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2753         makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2754         // like ('tiny{}' or '\tiny ' ... )
2755         makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2756
2757         // Survives, like known character
2758         // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2759         makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2760         makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2761
2762         makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2763         makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2764         makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2765
2766         makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2767         // Remove RTL/LTR marker
2768         makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2769         makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2770         makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2771         makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2772         makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2773         makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2774         makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2775         if (isPatternString) {
2776                 // Allow the first searched string to rebuild the keys too
2777                 keysBuilt = false;
2778         }
2779         else {
2780                 // no need to rebuild again
2781                 keysBuilt = true;
2782         }
2783 }
2784
2785 /*
2786  * Keep the list of actual opened parentheses actual
2787  * (e.g. depth == 4 means there are 4 '{' not processed yet)
2788  */
2789 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2790 {
2791         int skip = 0;
2792         for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2793                 char c;
2794                 c = par[i];
2795                 skip = 0;
2796                 if (c == '\\') skip = 1;
2797                 else if (c == '{') {
2798                         handleOpenP(i);
2799                 }
2800                 else if (c == '}') {
2801                         handleCloseP(i, closingAllowed);
2802                 }
2803         }
2804 }
2805
2806 #if (0)
2807 string Intervall::show(int lastpos)
2808 {
2809         int idx = 0;                          /* int intervalls */
2810         string s;
2811         int i = 0;
2812         for (idx = 0; idx <= ignoreidx; idx++) {
2813                 while (i < lastpos) {
2814                         int printsize;
2815                         if (i <= borders[idx].low) {
2816                                 if (borders[idx].low > lastpos)
2817                                         printsize = lastpos - i;
2818                                 else
2819                                         printsize = borders[idx].low - i;
2820                                 s += par.substr(i, printsize);
2821                                 i += printsize;
2822                                 if (i >= borders[idx].low)
2823                                         i = borders[idx].upper;
2824                         }
2825                         else {
2826                                 i = borders[idx].upper;
2827                                 break;
2828                         }
2829                 }
2830         }
2831         if (lastpos > i) {
2832                 s += par.substr(i, lastpos-i);
2833         }
2834         return s;
2835 }
2836 #endif
2837
2838 void Intervall::output(ostringstream &os, int lastpos)
2839 {
2840         // get number of chars to output
2841         int idx = 0;                          /* int intervalls */
2842         int i = 0;
2843         int printed = 0;
2844         string startTitle = titleValue;
2845         for (idx = 0; idx <= ignoreidx; idx++) {
2846                 if (i < lastpos) {
2847                         if (i <= borders[idx].low) {
2848                                 int printsize;
2849                                 if (borders[idx].low > lastpos)
2850                                         printsize = lastpos - i;
2851                                 else
2852                                         printsize = borders[idx].low - i;
2853                                 if (printsize > 0) {
2854                                         os << startTitle << par.substr(i, printsize);
2855                                         i += printsize;
2856                                         printed += printsize;
2857                                         startTitle = "";
2858                                 }
2859                                 handleParentheses(i, false);
2860                                 if (i >= borders[idx].low)
2861                                         i = borders[idx].upper;
2862                         }
2863                         else {
2864                                 i = borders[idx].upper;
2865                         }
2866                 }
2867                 else
2868                         break;
2869         }
2870         if (lastpos > i) {
2871                 os << startTitle << par.substr(i, lastpos-i);
2872                 printed += lastpos-i;
2873         }
2874         handleParentheses(lastpos, false);
2875         int startindex;
2876         if (keys["foreignlanguage"].disabled)
2877                 startindex = actualdeptindex-langcount;
2878         else
2879                 startindex = actualdeptindex;
2880         for (int i = startindex; i > 0; --i) {
2881                 os << "}";
2882         }
2883         if (hasTitle && (printed > 0))
2884                 os << "}";
2885         if (! isPatternString_)
2886                 os << "\n";
2887         handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2888 }
2889
2890 void LatexInfo::processRegion(int start, int region_end)
2891 {
2892         while (start < region_end) {          /* Let {[} and {]} survive */
2893                 int cnt = interval_.isOpeningPar(start);
2894                 if (cnt == 1) {
2895                         // Closing is allowed past the region
2896                         int closing = interval_.findclosing(start+1, interval_.par.length());
2897                         interval_.addIntervall(start, start+1);
2898                         interval_.addIntervall(closing, closing+1);
2899                 }
2900                 else if (cnt == 3)
2901                         start += 2;
2902                 start = interval_.nextNotIgnored(start+1);
2903         }
2904 }
2905
2906 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2907 {
2908         if (actual.parenthesiscount == 0) {
2909                 // "{\tiny{} ...}" ==> "{{} ...}"
2910                 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2911         }
2912         else {
2913                 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2914                 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2915                 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2916         }
2917 }
2918
2919 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2920 {
2921         int nextKeyIdx = 0;
2922         switch (actual.keytype)
2923         {
2924         case KeyInfo::isTitle: {
2925                 removeHead(actual);
2926                 nextKeyIdx = getNextKey();
2927                 break;
2928         }
2929         case KeyInfo::cleanToStart: {
2930                 actual._dataEnd = actual._dataStart;
2931                 nextKeyIdx = getNextKey();
2932                 // Search for end of arguments
2933                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2934                 if (tmpIdx > 0) {
2935                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2936                                 entries_[i].disabled = true;
2937                         }
2938                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
2939                 }
2940                 while (interval_.par[actual._dataEnd] == ' ')
2941                         actual._dataEnd++;
2942                 interval_.addIntervall(0, actual._dataEnd+1);
2943                 interval_.actualdeptindex = 0;
2944                 interval_.depts[0] = actual._dataEnd+1;
2945                 interval_.closes[0] = -1;
2946                 break;
2947         }
2948         case KeyInfo::isText:
2949                 interval_.par[actual._tokenstart] = '#';
2950                 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2951                 nextKeyIdx = getNextKey();
2952                 break;
2953         case KeyInfo::noContent: {          /* char like "\hspace{2cm}" */
2954                 if (actual.disabled)
2955                         interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2956                 else
2957                         interval_.addIntervall(actual._dataStart, actual._dataEnd);
2958         }
2959                 // fall through
2960         case KeyInfo::isChar: {
2961                 nextKeyIdx = getNextKey();
2962                 break;
2963         }
2964         case KeyInfo::isSize: {
2965                 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2966                         if (actual.parenthesiscount == 0)
2967                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2968                         else {
2969                                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2970                         }
2971                         nextKeyIdx = getNextKey();
2972                 } else {
2973                         // Here _dataStart points to '{', so correct it
2974                         actual._dataStart += 1;
2975                         actual._tokensize += 1;
2976                         actual.parenthesiscount = 1;
2977                         if (interval_.par[actual._dataStart] == '}') {
2978                                 // Determine the end if used like '{\tiny{}...}'
2979                                 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2980                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2981                         }
2982                         else {
2983                                 // Determine the end if used like '\tiny{...}'
2984                                 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2985                         }
2986                         // Split on this key if not at start
2987                         int start = interval_.nextNotIgnored(previousStart);
2988                         if (start < actual._tokenstart) {
2989                                 interval_.output(os, actual._tokenstart);
2990                                 interval_.addIntervall(start, actual._tokenstart);
2991                         }
2992                         // discard entry if at end of actual
2993                         nextKeyIdx = process(os, actual);
2994                 }
2995                 break;
2996         }
2997         case KeyInfo::endArguments: {
2998                 // Remove trailing '{}' too
2999                 actual._dataStart += 1;
3000                 actual._dataEnd += 1;
3001                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3002                 nextKeyIdx = getNextKey();
3003                 break;
3004         }
3005         case KeyInfo::noMain:
3006                 // fall through
3007         case KeyInfo::isStandard: {
3008                 if (actual.disabled) {
3009                         removeHead(actual);
3010                         processRegion(actual._dataStart, actual._dataStart+1);
3011                         nextKeyIdx = getNextKey();
3012                 } else {
3013                         // Split on this key if not at datastart of calling entry
3014                         int start = interval_.nextNotIgnored(previousStart);
3015                         if (start < actual._tokenstart) {
3016                                 interval_.output(os, actual._tokenstart);
3017                                 interval_.addIntervall(start, actual._tokenstart);
3018                         }
3019                         // discard entry if at end of actual
3020                         nextKeyIdx = process(os, actual);
3021                 }
3022                 break;
3023         }
3024         case KeyInfo::removeWithArg: {
3025                 nextKeyIdx = getNextKey();
3026                 // Search for end of arguments
3027                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3028                 if (tmpIdx > 0) {
3029                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3030                                 entries_[i].disabled = true;
3031                         }
3032                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3033                 }
3034                 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
3035                 break;
3036         }
3037         case KeyInfo::doRemove: {
3038                 // Remove the key with all parameters and following spaces
3039                 size_t pos;
3040                 size_t start;
3041                 if (interval_.par[actual._dataEnd-1] == ' ' || interval_.par[actual._dataEnd-1] == '}')
3042                         start = actual._dataEnd;
3043                 else
3044                         start = actual._dataEnd+1;
3045                 for (pos = start; pos < interval_.par.length(); pos++) {
3046                         if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
3047                                 break;
3048                 }
3049                 // Remove also enclosing parentheses [] and {}
3050                 int numpars = 0;
3051                 int spaces = 0;
3052                 while (actual._tokenstart > numpars) {
3053                         if (pos+numpars >= interval_.par.size())
3054                                 break;
3055                         else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
3056                                 numpars++;
3057                         else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
3058                                 numpars++;
3059                         else
3060                                 break;
3061                 }
3062                 if (numpars > 0) {
3063                         if (interval_.par[pos+numpars] == ' ')
3064                                 spaces++;
3065                 }
3066
3067                 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
3068                 nextKeyIdx = getNextKey();
3069                 break;
3070         }
3071         case KeyInfo::isList: {
3072                 // Discard space before _tokenstart
3073                 int count;
3074                 for (count = 0; count < actual._tokenstart; count++) {
3075                         if (interval_.par[actual._tokenstart-count-1] != ' ')
3076                                 break;
3077                 }
3078                 nextKeyIdx = getNextKey();
3079                 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
3080                 if (tmpIdx > 0) {
3081                         // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
3082                         // with arguments
3083                         // How else can we catch this one?
3084                         for (int i = nextKeyIdx; i <= tmpIdx; i++) {
3085                                 entries_[i].disabled = true;
3086                         }
3087                         actual._dataEnd = entries_[tmpIdx]._dataEnd;
3088                 }
3089                 else if (nextKeyIdx > 0) {
3090                         // Ignore any lang entries inside data region
3091                         for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
3092                                 if (entries_[i].keytype == KeyInfo::isMain)
3093                                         entries_[i].disabled = true;
3094                         }
3095                 }
3096                 if (actual.disabled) {
3097                         interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
3098                 }
3099                 else {
3100                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3101                 }
3102                 if (interval_.par[actual._dataEnd+1] == '[') {
3103                         int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
3104                         if ((interval_.par[actual._dataEnd+2] == '{') &&
3105                                         (interval_.par[posdown-1] == '}')) {
3106                                 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
3107                                 interval_.addIntervall(posdown-1, posdown+1);
3108                         }
3109                         else {
3110                                 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
3111                                 interval_.addIntervall(posdown, posdown+1);
3112                         }
3113                         int blk = interval_.nextNotIgnored(actual._dataEnd+1);
3114                         if (blk > posdown) {
3115                                 // Discard at most 1 space after empty item
3116                                 int count;
3117                                 for (count = 0; count < 1; count++) {
3118                                         if (interval_.par[blk+count] != ' ')
3119                                                 break;
3120                                 }
3121                                 if (count > 0)
3122                                         interval_.addIntervall(blk, blk+count);
3123                         }
3124                 }
3125                 break;
3126         }
3127         case KeyInfo::isSectioning: {
3128                 // Discard spaces before _tokenstart
3129                 int count;
3130                 int val = actual._tokenstart;
3131                 for (count = 0; count < actual._tokenstart;) {
3132                         val = interval_.previousNotIgnored(val-1);
3133                         if (val < 0 || interval_.par[val] != ' ')
3134                                 break;
3135                         else {
3136                                 count = actual._tokenstart - val;
3137                         }
3138                 }
3139                 if (actual.disabled) {
3140                         removeHead(actual, count);
3141                         nextKeyIdx = getNextKey();
3142                 } else {
3143                         interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
3144                         nextKeyIdx = process(os, actual);
3145                 }
3146                 break;
3147         }
3148         case KeyInfo::isMath: {
3149                 // Same as regex, use the content unchanged
3150                 nextKeyIdx = getNextKey();
3151                 break;
3152         }
3153         case KeyInfo::isRegex: {
3154                 // DO NOT SPLIT ON REGEX
3155                 // Do not disable
3156                 nextKeyIdx = getNextKey();
3157                 break;
3158         }
3159         case KeyInfo::isIgnored: {
3160                 // Treat like a character for now
3161                 nextKeyIdx = getNextKey();
3162                 break;
3163         }
3164         case KeyInfo::isMain: {
3165                 if (interval_.par.substr(actual._dataStart, 2) == "% ")
3166                         interval_.addIntervall(actual._dataStart, actual._dataStart+2);
3167                 if (actual._tokenstart > 0) {
3168                         int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
3169                         if ((prev >= 0) && interval_.par[prev] == '%')
3170                                 interval_.addIntervall(prev, prev+1);
3171                 }
3172                 if (actual.disabled) {
3173                         removeHead(actual);
3174                         interval_.langcount++;
3175                         if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
3176                                         (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
3177                                 // Discard also the space before math-equation
3178                                 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
3179                         }
3180                         nextKeyIdx = getNextKey();
3181                         // interval.resetOpenedP(actual._dataStart-1);
3182                 }
3183                 else {
3184                         if (actual._tokenstart < 26) {
3185                                 // for the first (and maybe dummy) language
3186                                 interval_.setForDefaultLang(actual);
3187                         }
3188                         interval_.resetOpenedP(actual._dataStart-1);
3189                 }
3190                 break;
3191         }
3192         case KeyInfo::invalid:
3193         case KeyInfo::headRemove:
3194                 // These two cases cannot happen, already handled
3195                 // fall through
3196         default: {
3197                 // LYXERR(Debug::INFO, "Unhandled keytype");
3198                 nextKeyIdx = getNextKey();
3199                 break;
3200         }
3201         }
3202         return nextKeyIdx;
3203 }
3204
3205 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
3206 {
3207         int end = interval_.nextNotIgnored(actual._dataEnd);
3208         int oldStart = actual._dataStart;
3209         int nextKeyIdx = getNextKey();
3210         while (true) {
3211                 if ((nextKeyIdx < 0) ||
3212                                 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
3213                                 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
3214                         if (oldStart <= end) {
3215                                 processRegion(oldStart, end);
3216                                 oldStart = end+1;
3217                         }
3218                         break;
3219                 }
3220                 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
3221
3222                 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
3223                         (void) dispatch(os, actual._dataStart, nextKey);
3224                         end = nextKey._tokenstart;
3225                         break;
3226                 }
3227                 processRegion(oldStart, nextKey._tokenstart);
3228                 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
3229
3230                 oldStart = nextKey._dataEnd+1;
3231         }
3232         // now nextKey is either invalid or is outside of actual._dataEnd
3233         // output the remaining and discard myself
3234         if (oldStart <= end) {
3235                 processRegion(oldStart, end);
3236         }
3237         if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
3238                 end += 1;
3239                 // This is the normal case.
3240                 // But if using the firstlanguage, the closing may be missing
3241         }
3242         // get minimum of 'end' and  'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
3243         int output_end;
3244         if (actual._dataEnd < end)
3245                 output_end = interval_.nextNotIgnored(actual._dataEnd);
3246         else if (interval_.par.size() > (size_t) end)
3247                 output_end = interval_.nextNotIgnored(end);
3248         else
3249                 output_end = interval_.par.size();
3250         if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
3251                 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
3252         }
3253         // Remove possible empty data
3254         int dstart = interval_.nextNotIgnored(actual._dataStart);
3255         while (interval_.isOpeningPar(dstart) == 1) {
3256                 interval_.addIntervall(dstart, dstart+1);
3257                 int dend = interval_.findclosing(dstart+1, output_end);
3258                 interval_.addIntervall(dend, dend+1);
3259                 dstart = interval_.nextNotIgnored(dstart+1);
3260         }
3261         if (dstart < output_end)
3262                 interval_.output(os, output_end);
3263         if (nextKeyIdx < 0)
3264                 interval_.addIntervall(0, end);
3265         else
3266                 interval_.addIntervall(actual._tokenstart, end);
3267         return nextKeyIdx;
3268 }
3269
3270 string splitOnKnownMacros(string par, bool isPatternString)
3271 {
3272         ostringstream os;
3273         LatexInfo li(par, isPatternString);
3274         // LYXERR(Debug::INFO, "Berfore split: " << par);
3275         KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
3276         DummyKey.head = "";
3277         DummyKey._tokensize = 0;
3278         DummyKey._dataStart = 0;
3279         DummyKey._dataEnd = par.length();
3280         DummyKey.disabled = true;
3281         int firstkeyIdx = li.getFirstKey();
3282         string s;
3283         if (firstkeyIdx >= 0) {
3284                 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
3285                 DummyKey._tokenstart = firstKey._tokenstart;
3286                 int nextkeyIdx;
3287                 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
3288                         // Use dummy firstKey
3289                         firstKey = DummyKey;
3290                         (void) li.setNextKey(firstkeyIdx);
3291                 }
3292                 else {
3293                         if (par.substr(firstKey._dataStart, 2) == "% ")
3294                                 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
3295                 }
3296                 nextkeyIdx = li.process(os, firstKey);
3297                 while (nextkeyIdx >= 0) {
3298                         // Check for a possible gap between the last
3299                         // entry and this one
3300                         int datastart = li.nextNotIgnored(firstKey._dataStart);
3301                         KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
3302                         if ((nextKey._tokenstart > datastart)) {
3303                                 // Handle the gap
3304                                 firstKey._dataStart = datastart;
3305                                 firstKey._dataEnd = par.length();
3306                                 (void) li.setNextKey(nextkeyIdx);
3307                                 // Fake the last opened parenthesis
3308                                 li.setForDefaultLang(firstKey);
3309                                 nextkeyIdx = li.process(os, firstKey);
3310                         }
3311                         else {
3312                                 if (nextKey.keytype != KeyInfo::isMain) {
3313                                         firstKey._dataStart = datastart;
3314                                         firstKey._dataEnd = nextKey._dataEnd+1;
3315                                         (void) li.setNextKey(nextkeyIdx);
3316                                         li.setForDefaultLang(firstKey);
3317                                         nextkeyIdx = li.process(os, firstKey);
3318                                 }
3319                                 else {
3320                                         nextkeyIdx = li.process(os, nextKey);
3321                                 }
3322                         }
3323                 }
3324                 // Handle the remaining
3325                 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
3326                 firstKey._dataEnd = par.length();
3327                 // Check if ! empty
3328                 if ((firstKey._dataStart < firstKey._dataEnd) &&
3329                                 (par[firstKey._dataStart] != '}')) {
3330                         li.setForDefaultLang(firstKey);
3331                         (void) li.process(os, firstKey);
3332                 }
3333                 s = os.str();
3334                 // return string definitelly impossible to match, but should be known
3335         }
3336         else
3337                 s = par;                            /* no known macros found */
3338         // LYXERR(Debug::INFO, "After split: " << s);
3339         return s;
3340 }
3341
3342 /*
3343  * Try to unify the language specs in the latexified text.
3344  * Resulting modified string is set to "", if
3345  * the searched tex does not contain all the features in the search pattern
3346  */
3347 static string correctlanguagesetting(string par, bool isPatternString, bool withformat, lyx::Buffer *pbuf = nullptr)
3348 {
3349         static Features regex_f;
3350         static int missed = 0;
3351         static bool regex_with_format = false;
3352
3353         int parlen = par.length();
3354
3355         while ((parlen > 0) && (par[parlen-1] == '\n')) {
3356                 parlen--;
3357         }
3358         if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
3359                 // Happens to be there in case of description or labeling environment
3360                 parlen--;
3361         }
3362         string result;
3363         if (withformat) {
3364                 // Split the latex input into pieces which
3365                 // can be digested by our search engine
3366                 LYXERR(Debug::FINDVERBOSE, "input: \"" << par << "\"");
3367                 if (isPatternString && (pbuf != nullptr)) { // Check if we should disable/enable test for language
3368                         // We check for polyglossia, because in runparams.flavor we use Flavor::XeTeX
3369                         string doclang = pbuf->params().language->polyglossia();
3370                         static regex langre("\\\\(foreignlanguage)\\{([^\\}]+)\\}");
3371                         smatch sub;
3372                         bool toIgnoreLang = true;
3373                         for (sregex_iterator it(par.begin(), par.end(), langre), end; it != end; ++it) {
3374                                 sub = *it;
3375                                 if (sub.str(2) != doclang) {
3376                                         toIgnoreLang = false;
3377                                         break;
3378                                 }
3379                         }
3380                         setIgnoreFormat("language", toIgnoreLang, false);
3381
3382                 }
3383                 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
3384                 LYXERR(Debug::FINDVERBOSE, "After splitOnKnownMacros:\n\"" << result << "\"");
3385         }
3386         else
3387                 result = par.substr(0, parlen);
3388         if (isPatternString) {
3389                 missed = 0;
3390                 if (withformat) {
3391                         regex_f = identifyFeatures(result);
3392                         string features = "";
3393                         for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3394                                 string a = it->first;
3395                                 regex_with_format = true;
3396                                 features += " " + a;
3397                                 // LYXERR(Debug::INFO, "Identified regex format:" << a);
3398                         }
3399                         LYXERR(Debug::FINDVERBOSE, "Identified Features" << features);
3400
3401                 }
3402         } else if (regex_with_format) {
3403                 Features info = identifyFeatures(result);
3404                 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3405                         string a = it->first;
3406                         bool b = it->second;
3407                         if (b && ! info[a]) {
3408                                 missed++;
3409                                 LYXERR(Debug::FINDVERBOSE, "Missed(" << missed << " " << a <<", srclen = " << parlen );
3410                                 return "";
3411                         }
3412                 }
3413
3414         }
3415         else {
3416                 // LYXERR(Debug::INFO, "No regex formats");
3417         }
3418         return result;
3419 }
3420
3421
3422 // Remove trailing closure of math, macros and environments, so to catch parts of them.
3423 static void identifyClosing(string & t, bool ignoreformat)
3424 {
3425         do {
3426                 LYXERR(Debug::FINDVERBOSE, "identifyClosing(): t now is '" << t << "'");
3427                 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
3428                         continue;
3429                 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
3430                         continue;
3431                 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]+\\*?\\}$", "$1"))
3432                         continue;
3433                 if (! ignoreformat) {
3434                         if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1"))
3435                                 continue;
3436                 }
3437                 break;
3438         } while (true);
3439         return;
3440 }
3441
3442 static int num_replaced = 0;
3443 static bool previous_single_replace = true;
3444
3445 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
3446 {
3447 #if QTSEARCH
3448         if (regexp_str.empty() || regexp2_str.empty()) {
3449                 regexIsValid = false;
3450                 regexError = "Invalid empty regex";
3451                 return;
3452         }
3453         // Handle \w properly
3454         QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3455         if (! opt.casesensitive) {
3456                 popts |= QRegularExpression::CaseInsensitiveOption;
3457         }
3458         regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3459         regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3460         regexError = "";
3461         if (regexp.isValid() && regexp2.isValid()) {
3462                 regexIsValid = true;
3463                 // Check '{', '}' pairs inside the regex
3464                 int balanced = 0;
3465                 int skip = 1;
3466                 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3467                         char c = par_as_string[i];
3468                         if (c == '\\') {
3469                                 skip = 2;
3470                                 continue;
3471                         }
3472                         if (c == '{')
3473                                 balanced++;
3474                         else if (c == '}') {
3475                                 balanced--;
3476                                 if (balanced < 0)
3477                                         break;
3478                         }
3479                         skip = 1;
3480                 }
3481                 if (balanced != 0) {
3482                         regexIsValid = false;
3483                         regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3484                 }
3485         }
3486         else {
3487                 regexIsValid = false;
3488                 if (!regexp.isValid())
3489                         regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3490                 else
3491                         regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3492         }
3493 #else
3494         (void)par_as_string;
3495         if (opt.casesensitive) {
3496                 regexp = regex(regexp_str);
3497                 regexp2 = regex(regexp2_str);
3498         }
3499         else {
3500                 regexp = regex(regexp_str, std::regex_constants::icase);
3501                 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3502         }
3503 #endif
3504 }
3505
3506 static void modifyRegexForMatchWord(string &t)
3507 {
3508         string s("");
3509         regex wordre("(\\\\)*((\\.|\\\\b))");
3510         size_t lastpos = 0;
3511         smatch sub;
3512         for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3513                 sub = *it;
3514                 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3515                         continue;
3516                 }
3517                 else if (sub.str(2) == "\\\\b")
3518                         return;
3519                 if (lastpos < (size_t) sub.position(2))
3520                         s += t.substr(lastpos, sub.position(2) - lastpos);
3521                 s += "\\S";
3522                 lastpos = sub.position(2) + sub.length(2);
3523         }
3524         if (lastpos == 0) {
3525                 s = "\\b" + t + "\\b";
3526                 t = s;
3527                 return;
3528         }
3529         else if (lastpos < t.length())
3530                 s += t.substr(lastpos, t.length() - lastpos);
3531         t = "\\b" + s + "\\b";
3532 }
3533
3534 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3535         : p_buf(&buf), p_first_buf(&buf), opt(opt)
3536 {
3537         Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3538         docstring const & ds = stringifySearchBuffer(find_buf, opt);
3539         if (ds.empty() ) {
3540                 CreateRegexp(opt, "", "", "");
3541                 return;
3542         }
3543         use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3544         if (opt.replace_all && previous_single_replace) {
3545                 previous_single_replace = false;
3546                 num_replaced = 0;
3547         }
3548         else if (!opt.replace_all) {
3549                 num_replaced = 0;       // count number of replaced strings
3550                 previous_single_replace = true;
3551         }
3552         // When using regexp, braces are hacked already by escape_for_regex()
3553         par_as_string = normalize(ds, opt.ignoreformat);
3554         open_braces = 0;
3555         close_wildcards = 0;
3556
3557         size_t lead_size = 0;
3558         // correct the language settings
3559         par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat, &buf);
3560         if (par_as_string.empty()) {
3561                 CreateRegexp(opt, "", "", "");
3562                 return;
3563         }
3564         opt.matchAtStart = false;
3565         if (!use_regexp) {
3566                 identifyClosing(par_as_string, opt.ignoreformat); // Removes math closings ($, ], ...) at end of string
3567                 if (opt.ignoreformat) {
3568                         lead_size = 0;
3569                 }
3570                 else {
3571                         lead_size = identifyLeading(par_as_string);
3572                 }
3573                 lead_as_string = par_as_string.substr(0, lead_size);
3574                 string lead_as_regex_string = string2regex(lead_as_string);
3575                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3576                 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3577                 /* Handle whole words too in this case
3578                 */
3579                 if (opt.matchword) {
3580                         par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3581                         opt.matchword = false;
3582                 }
3583                 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3584                 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3585                 CreateRegexp(opt, regexp_str, regexp2_str);
3586                 use_regexp = true;
3587                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3588                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3589                 return;
3590         }
3591
3592         if (!opt.ignoreformat) {
3593                 lead_size = identifyLeading(par_as_string);
3594                 LYXERR(Debug::FINDVERBOSE, "Lead_size: " << lead_size);
3595                 lead_as_string = par_as_string.substr(0, lead_size);
3596                 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3597         }
3598
3599         // Here we are using regexp
3600         LASSERT(use_regexp, /**/);
3601         {
3602                 string lead_as_regexp;
3603                 if (lead_size > 0) {
3604                         lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3605                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "\\$$", "");
3606                         (void)regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3607                         par_as_string = par_as_string_nolead;
3608                         LYXERR(Debug::FINDVERBOSE, "lead_as_regexp is '" << lead_as_regexp << "'");
3609                         LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3610                 }
3611                 // LYXERR(Debug::FINDVERBOSE, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3612                 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3613                 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3614                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3615                 ++close_wildcards;
3616                 size_t lng = par_as_string.size();
3617                 if (!opt.ignoreformat) {
3618                         // Remove extra '\}' at end if not part of \{\.\}
3619                         while(lng > 2) {
3620                                 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3621                                         if (lng >= 6) {
3622                                                 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3623                                                         break;
3624                                         }
3625                                         lng -= 2;
3626                                         open_braces++;
3627                                 }
3628                                 else
3629                                         break;
3630                         }
3631                         if (lng < par_as_string.size())
3632                                 par_as_string = par_as_string.substr(0,lng);
3633                 }
3634                 LYXERR(Debug::FINDVERBOSE, "par_as_string after correctRegex is '" << par_as_string << "'");
3635                 if ((lng > 0) && (par_as_string[0] == '^')) {
3636                         par_as_string = par_as_string.substr(1);
3637                         --lng;
3638                         opt.matchAtStart = true;
3639                 }
3640                 // LYXERR(Debug::FINDVERBOSE, "par_as_string now is '" << par_as_string << "'");
3641                 // LYXERR(Debug::FINDVERBOSE, "Open braces: " << open_braces);
3642                 // LYXERR(Debug::FINDVERBOSE, "Replaced text (to be used as regex): " << par_as_string);
3643
3644                 // If entered regexp must match at begin of searched string buffer
3645                 // Kornel: Added parentheses to use $1 for size of the leading string
3646                 string regexp_str;
3647                 string regexp2_str;
3648                 {
3649                         // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3650                         // Unfortunately is '\1', '\2', etc not working for strings with extra format
3651                         // so the convert has no effect in that case
3652                         for (int i = 7; i > 0; --i) {
3653                                 string orig = "\\\\" + std::to_string(i);
3654                                 string dest = "\\" + std::to_string(i+2);
3655                                 while (regex_replace(par_as_string, par_as_string, orig, dest));
3656                         }
3657                         if (opt.matchword) {
3658                                 modifyRegexForMatchWord(par_as_string);
3659                                 opt.matchword = false;
3660                         }
3661                         regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3662                         regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3663                 }
3664                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3665                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3666                 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3667         }
3668 }
3669
3670 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3671 {
3672         MatchResult mres;
3673
3674         mres.searched_size = len;
3675         if (at_begin &&
3676                         (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3677                 return mres;
3678
3679         docstring docstr = stringifyFromForSearch(opt, cur, len);
3680         string str;
3681         str = normalize(docstr, opt.ignoreformat);
3682         if (!opt.ignoreformat) {
3683                 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3684                 // remove closing '}' and '\n' to allow for use of '$' in regex
3685                 size_t lng = str.size();
3686                 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3687                         lng--;
3688                 if (lng != str.size()) {
3689                         str = str.substr(0, lng);
3690                 }
3691         }
3692         if (str.empty()) {
3693                 mres.match_len = -1;
3694                 return mres;
3695         }
3696         LYXERR(Debug::FINDVERBOSE, "After normalization: Matching against:\n'" << str << "'");
3697
3698         LASSERT(use_regexp, /**/);
3699         {
3700                 // use_regexp always true
3701                 LYXERR(Debug::FINDVERBOSE, "Searching in regexp mode: at_begin=" << at_begin);
3702 #if QTSEARCH
3703                 QString qstr = QString::fromStdString(str);
3704                 QRegularExpression const *p_regexp;
3705                 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3706                 if (at_begin) {
3707                         p_regexp = &regexp;
3708                 } else {
3709                         p_regexp = &regexp2;
3710                 }
3711                 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3712                 if (!match.hasMatch())
3713                         return mres;
3714 #else
3715                 regex const *p_regexp;
3716                 regex_constants::match_flag_type flags;
3717                 if (at_begin) {
3718                         flags = regex_constants::match_continuous;
3719                         p_regexp = &regexp;
3720                 } else {
3721                         flags = regex_constants::match_default;
3722                         p_regexp = &regexp2;
3723                 }
3724                 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3725                 if (re_it == sregex_iterator())
3726                         return mres;
3727                 match_results<string::const_iterator> const & m = *re_it;
3728 #endif
3729                 // Whole found string, including the leading
3730                 // std: m[0].second - m[0].first
3731                 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3732                 //
3733                 // Size of the leading string
3734                 // std: m[1].second - m[1].first
3735                 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3736                 int leadingsize = 0;
3737 #if QTSEARCH
3738                 if (match.lastCapturedIndex() > 0) {
3739                         leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3740                 }
3741
3742 #else
3743                 if (m.size() > 2) {
3744                         leadingsize = m[1].second - m[1].first;
3745                 }
3746 #endif
3747 #if QTSEARCH
3748                 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3749                 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3750                 // because of different number of closing at end of string
3751                 // we have to 'unify' the length of the post-match.
3752                 // Done by ignoring closing parenthesis and linefeeds at string end
3753                 int matchend = match.capturedEnd(0);
3754                 size_t strsize = qstr.size();
3755                 if (!opt.ignoreformat) {
3756                         while (mres.match_len > 0) {
3757                                 QChar c = qstr.at(matchend - 1);
3758                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3759                                         mres.match_len--;
3760                                         matchend--;
3761                                 }
3762                                 else
3763                                         break;
3764                         }
3765                         while (strsize > (size_t) match.capturedEnd(0)) {
3766                                 QChar c = qstr.at(strsize-1);
3767                                 if ((c == '\n') || (c == '}')) {
3768                                         --strsize;
3769                                 }
3770                                 else
3771                                         break;
3772                         }
3773                 }
3774                 // LYXERR0(qstr.toStdString());
3775                 mres.match2end = strsize - matchend;
3776                 mres.pos = match.capturedStart(2);
3777 #else
3778                 mres.match_prefix = m[2].second - m[2].first;
3779                 mres.match_len = m[0].second - m[2].second;
3780                 // ignore closing parenthesis and linefeeds at string end
3781                 size_t strend = m[0].second - m[0].first;
3782                 int matchend = strend;
3783                 size_t strsize = str.size();
3784                 if (!opt.ignoreformat) {
3785                         while (mres.match_len > 0) {
3786                                 char c = str.at(matchend - 1);
3787                                 if ((c == '\n') || (c == '}') || (c == '{')) {
3788                                         mres.match_len--;
3789                                         matchend--;
3790                                 }
3791                                 else
3792                                         break;
3793                         }
3794                         while (strsize > strend) {
3795                                 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3796                                         --strsize;
3797                                 }
3798                                 else
3799                                         break;
3800                         }
3801                 }
3802                 // LYXERR0(str);
3803                 mres.match2end = strsize - matchend;
3804                 mres.pos = m[2].first - m[0].first;;
3805 #endif
3806                 if (mres.match2end < 0)
3807                         mres.match_len = 0;
3808                 mres.leadsize = leadingsize;
3809 #if QTSEARCH
3810                 if (mres.match_len > 0) {
3811                         string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3812                         mres.result.push_back(a0);
3813                         for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3814                                 mres.result.push_back(match.captured(i).toStdString());
3815                         }
3816                 }
3817 #else
3818                 if (mres.match_len > 0) {
3819                         string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3820                         mres.result.push_back(a0);
3821                         for (size_t i = 3; i < m.size(); i++) {
3822                                 mres.result.push_back(m[i]);
3823                         }
3824                 }
3825 #endif
3826                 return mres;
3827         }
3828 }
3829
3830
3831 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3832 {
3833         MatchResult mres = findAux(cur, len, at_begin);
3834         int res = mres.match_len;
3835         LYXERR(Debug::FINDVERBOSE,
3836                "res=" << res << ", at_begin=" << at_begin
3837                << ", matchAtStart=" << opt.matchAtStart
3838                << ", inTexted=" << cur.inTexted());
3839         if (opt.matchAtStart) {
3840                 if (cur.pos() != 0)
3841                         mres.match_len = 0;
3842                 else if (mres.match_prefix > 0)
3843                         mres.match_len = 0;
3844                 return mres;
3845         }
3846         else
3847                 return mres;
3848 }
3849
3850 #if 0
3851 static bool simple_replace(string &t, string from, string to)
3852 {
3853         regex repl("(\\\\)*(" + from + ")");
3854         string s("");
3855         size_t lastpos = 0;
3856         smatch sub;
3857         for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3858                 sub = *it;
3859                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3860                         continue;
3861                 if (lastpos < (size_t) sub.position(2))
3862                         s += t.substr(lastpos, sub.position(2) - lastpos);
3863                 s += to;
3864                 lastpos = sub.position(2) + sub.length(2);
3865         }
3866         if (lastpos == 0)
3867                 return false;
3868         else if (lastpos < t.length())
3869                 s += t.substr(lastpos, t.length() - lastpos);
3870         t = s;
3871         return true;
3872 }
3873 #endif
3874
3875 string MatchStringAdv::normalize(docstring const & s, bool ignore_format) const
3876 {
3877         string t;
3878         t = lyx::to_utf8(s);
3879         // Remove \n at begin
3880         while (!t.empty() && t[0] == '\n')
3881                 t = t.substr(1);
3882         // Remove \n* at end
3883         while (!t.empty() && t[t.size() - 1] == '\n') {
3884                 t = t.substr(0, t.size() - 1);
3885         }
3886         size_t pos;
3887         // Handle all other '\n'
3888         while ((pos = t.find("\n")) != string::npos) {
3889                 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3890                         // Handle '\\\n'
3891                         if (isAlnumASCII(t[pos+1])) {
3892                                 t.replace(pos-2, 3, " ");
3893                         }
3894                         else {
3895                                 t.replace(pos-2, 3, "");
3896                         }
3897                 }
3898                 else {
3899                         if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3900                                 // '\n' adjacent to non-alpha-numerics, discard
3901                                 t.replace(pos, 1, "");
3902                         }
3903                         else {
3904                                 // Replace all other \n with spaces
3905                                 t.replace(pos, 1, " ");
3906                         }
3907                         if (!ignore_format) {
3908                                 size_t count = 0;
3909                                 while ((pos > count + 1) && (t[pos - 1 -count] == '%')) {
3910                                         count++;
3911                                 }
3912                                 if (count > 0) {
3913                                         t.replace(pos - count, count, "");
3914                                 }
3915                         }
3916                 }
3917         }
3918         // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3919         // Kornel: Added textsl, textsf, textit, texttt and noun
3920         // + allow to seach for colored text too
3921         LYXERR(Debug::FINDVERBOSE, "Removing stale empty macros from: " << t);
3922         while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3923                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3924         while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3925                 LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3926         while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3927
3928         return t;
3929 }
3930
3931
3932 docstring stringifyFromCursor(DocIterator const & cur, int len)
3933 {
3934         LYXERR(Debug::FINDVERBOSE, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3935         if (cur.inTexted()) {
3936                 Paragraph const & par = cur.paragraph();
3937                 // TODO what about searching beyond/across paragraph breaks ?
3938                 // TODO Try adding a AS_STR_INSERTS as last arg
3939                 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3940                                         int(par.size()) : cur.pos() + len;
3941                 // OutputParams runparams(&cur.buffer()->params().encoding());
3942                 OutputParams runparams(encodings.fromLyXName("utf8"));
3943                 runparams.nice = true;
3944                 runparams.flavor = Flavor::XeTeX;
3945                 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3946                 // No side effect of file copying and image conversion
3947                 runparams.dryrun = true;
3948                 int option = AS_STR_INSETS | AS_STR_PLAINTEXT;
3949                 if (ignoreFormats.getDeleted()) {
3950                         option |= AS_STR_SKIPDELETE;
3951                         runparams.for_search = OutputParams::SearchWithoutDeleted;
3952                 }
3953                 else {
3954                         runparams.for_search = OutputParams::SearchWithDeleted;
3955                 }
3956                 if (ignoreFormats.getNonContent()) {
3957                         runparams.for_search |= OutputParams::SearchNonOutput;
3958                 }
3959                 LYXERR(Debug::FINDVERBOSE, "Stringifying with cur: "
3960                        << cur << ", from pos: " << cur.pos() << ", end: " << end);
3961                 docstring res = from_utf8(latexNamesToUtf8(par.asString(cur.pos(), end,
3962                                                                         option,
3963                                                                         &runparams)));
3964                 LYXERR(Debug::FIND, "Stringified text from pos(" << cur.pos() << ") len(" << len << "): " << res);
3965                 return res;
3966         } else if (cur.inMathed()) {
3967                 CursorSlice cs = cur.top();
3968                 MathData md = cs.cell();
3969                 MathData::const_iterator it_end =
3970                                 (( len == -1 || cs.pos() + len > int(md.size()))
3971                                  ? md.end()
3972                                  : md.begin() + cs.pos() + len );
3973                 MathData md2;
3974                 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
3975                         md2.push_back(*it);
3976                 docstring res = from_utf8(latexNamesToUtf8(asString(md2)));
3977                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Stringified math from pos(" << cur.pos() << ") len(" << len << "): " << res);
3978                 return res;
3979         }
3980         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
3981         return docstring();
3982 }
3983
3984 /** Computes the LaTeX export of buf starting from cur and ending len positions
3985  * after cur, if len is positive, or at the paragraph or innermost inset end
3986  * if len is -1.
3987  */
3988 docstring latexifyFromCursor(DocIterator const & cur, int len)
3989 {
3990         /*
3991         LYXERR(Debug::FINDVERBOSE, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3992         LYXERR(Debug::FINDVERBOSE, "  with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3993                << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3994         */
3995         Buffer const & buf = *cur.buffer();
3996
3997         odocstringstream ods;
3998         otexstream os(ods);
3999         //OutputParams runparams(&buf.params().encoding());
4000         OutputParams runparams(encodings.fromLyXName("utf8"));
4001         runparams.nice = false;
4002         runparams.flavor = Flavor::XeTeX;
4003         runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4004         // No side effect of file copying and image conversion
4005         runparams.dryrun = true;
4006         if (ignoreFormats.getDeleted()) {
4007                 runparams.for_search = OutputParams::SearchWithoutDeleted;
4008         }
4009         else {
4010                 runparams.for_search = OutputParams::SearchWithDeleted;
4011         }
4012         if (ignoreFormats.getNonContent()) {
4013                 runparams.for_search |= OutputParams::SearchNonOutput;
4014         }
4015
4016         if (cur.inTexted()) {
4017                 // @TODO what about searching beyond/across paragraph breaks ?
4018                 pos_type endpos = cur.paragraph().size();
4019                 if (len != -1 && endpos > cur.pos() + len)
4020                         endpos = cur.pos() + len;
4021                 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
4022                           string(), cur.pos(), endpos, true);
4023                 LYXERR(Debug::FIND, "Latexified text from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4024                 return(ods.str());
4025         } else if (cur.inMathed()) {
4026                 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
4027                 for (int s = cur.depth() - 1; s >= 0; --s) {
4028                         CursorSlice const & cs = cur[s];
4029                         if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
4030                                 TeXMathStream ws(os);
4031                                 cs.asInsetMath()->asHullInset()->header_write(ws);
4032                                 break;
4033                         }
4034                 }
4035
4036                 CursorSlice const & cs = cur.top();
4037                 MathData md = cs.cell();
4038                 MathData::const_iterator it_end =
4039                                 ((len == -1 || cs.pos() + len > int(md.size()))
4040                                  ? md.end()
4041                                  : md.begin() + cs.pos() + len);
4042                 MathData md2;
4043                 for (MathData::const_iterator it = md.begin() + cs.pos();
4044                      it != it_end; ++it)
4045                         md2.push_back(*it);
4046
4047                 ods << asString(md2);
4048                 // Retrieve the math environment type, and add '$' or '$]'
4049                 // or others (\end{equation}) accordingly
4050                 for (int s = cur.depth() - 1; s >= 0; --s) {
4051                         CursorSlice const & cs2 = cur[s];
4052                         InsetMath * inset = cs2.asInsetMath();
4053                         if (inset && inset->asHullInset()) {
4054                                 TeXMathStream ws(os);
4055                                 inset->asHullInset()->footer_write(ws);
4056                                 break;
4057                         }
4058                 }
4059                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Latexified math from pos(" << cur.pos() << ") len(" << len << "): " << ods.str());
4060         } else {
4061                 LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Don't know how to stringify from here: " << cur);
4062         }
4063         return ods.str();
4064 }
4065
4066 #if defined(ResultsDebug)
4067 // Debugging output
4068 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
4069 {
4070         LYXERR0( "from:\t\t\t" << from);
4071         string status;
4072         if (mres.pos_len > 0) {
4073                 // Set in finalize
4074                 status = "FINALSEARCH";
4075         }
4076         else {
4077                 if (mres.match_len > 0) {
4078                         if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
4079                                 status = "Good Match";
4080                         else
4081                                 status = "Matched in";
4082                 }
4083                 else
4084                         status = "MissedSearch";
4085         }
4086
4087         LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
4088         if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
4089                 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
4090         if ((mres.pos > 0) || (mres.match_prefix > 0))
4091                 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
4092         for (size_t i = 0; i < mres.result.size(); i++)
4093                 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
4094 }
4095 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
4096 #else
4097 #define displayMres(s, txt, cur)
4098 #endif
4099
4100 /** Finalize an advanced find operation, advancing the cursor to the innermost
4101  ** position that matches, plus computing the length of the matching text to
4102  ** be selected
4103  ** Return the cur.pos() difference between start and end of found match
4104  **/
4105 MatchResult findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
4106 {
4107         // Search the foremost position that matches (avoids find of entire math
4108         // inset when match at start of it)
4109         DocIterator old_cur(cur.buffer());
4110         MatchResult mres;
4111         static MatchResult fail = MatchResult();
4112         MatchResult max_match;
4113         // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
4114         // Happens with e.g. hyperlinks
4115         // either one sees "http://www.bla.bla" or nothing
4116         // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
4117         // and although we search for only 3 chars, we find the whole hyperlink inset
4118         bool at_begin = (expected.match_prefix == 0);
4119         if (!match.opt.forward && match.opt.ignoreformat) {
4120                 if (expected.pos > 0)
4121                         return fail;
4122         }
4123         LASSERT(at_begin, /**/);
4124         if (expected.match_len > 0 && at_begin) {
4125                 // Search for deepest match
4126                 old_cur = cur;
4127                 max_match = expected;
4128                 do {
4129                         size_t d = cur.depth();
4130                         cur.forwardPos();
4131                         if (!cur)
4132                                 break;
4133                         if (cur.depth() < d)
4134                                 break;
4135                         if (cur.depth() == d)
4136                                 break;
4137                         size_t lastd = d;
4138                         while (cur && cur.depth() > lastd) {
4139                                 lastd = cur.depth();
4140                                 mres = match(cur, -1, at_begin);
4141                                 displayMres(mres, "Checking innermost", cur);
4142                                 if (mres.match_len > 0)
4143                                         break;
4144                                 // maybe deeper?
4145                                 cur.forwardPos();
4146                         }
4147                         if (mres.match_len < expected.match_len)
4148                                 break;
4149                         max_match = mres;
4150                         old_cur = cur;;
4151                 } while(1);
4152                 cur = old_cur;
4153         }
4154         else {
4155                 // (expected.match_len <= 0)
4156                 mres = match(cur);      /* match valid only if not searching whole words */
4157                 displayMres(mres, "Start with negative match", cur);
4158                 max_match = mres;
4159         }
4160         if (max_match.match_len <= 0) return fail;
4161         LYXERR(Debug::FINDVERBOSE, "Ok");
4162
4163         // Compute the match length
4164         int len = 1;
4165         if (cur.pos() + len > cur.lastpos())
4166                 return fail;
4167
4168         LASSERT(match.use_regexp, /**/);
4169         {
4170                 int minl = 1;
4171                 int maxl = cur.lastpos() - cur.pos();
4172                 // Greedy behaviour while matching regexps
4173                 while (maxl > minl) {
4174                         MatchResult mres2;
4175                         mres2 = match(cur, len, at_begin);
4176                         displayMres(mres2, "Finalize loop", cur);
4177                         int actual_match_len = mres2.match_len;
4178                         if (actual_match_len >= max_match.match_len) {
4179                                 // actual_match_len > max_match _can_ happen,
4180                                 // if the search area splits
4181                                 // some following word so that the regex
4182                                 // (e.g. 'r.*r\b' matches 'r' from the middle of the
4183                                 // splitted word)
4184                                 // This means, the len value is too big
4185                                 actual_match_len = max_match.match_len;
4186                                 max_match = mres2;
4187                                 max_match.match_len = actual_match_len;
4188                                 maxl = len;
4189                                 if (maxl - minl < 4)
4190                                         len = (maxl + minl)/2;
4191                                 else
4192                                         len = minl + (maxl - minl + 3)/4;
4193                         }
4194                         else {
4195                                 // (actual_match_len < max_match.match_len)
4196                                 minl = len + 1;
4197                                 len = (maxl + minl)/2;
4198                         }
4199                 }
4200                 len = minl;
4201                 old_cur = cur;
4202                 // Search for real start of matched characters
4203                 while (len > 1) {
4204                         MatchResult actual_match;
4205                         do {
4206                                 cur.forwardPos();
4207                         } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
4208                         if (cur.depth() < old_cur.depth()) {
4209                                 // Outer inset?
4210                                 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
4211                                 break;
4212                         }
4213                         if (cur.pos() != old_cur.pos()) {
4214                                 // OK, forwarded 1 pos in actual inset
4215                                 actual_match = match(cur, len-1, at_begin);
4216                                 if (actual_match.match_len == max_match.match_len) {
4217                                         // Ha, got it! The shorter selection has the same match length
4218                                         len--;
4219                                         old_cur = cur;
4220                                         max_match = actual_match;
4221                                 }
4222                                 else {
4223                                         // OK, the shorter selection matches less chars, revert to previous value
4224                                         cur = old_cur;
4225                                         break;
4226                                 }
4227                         }
4228                         else {
4229                                 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
4230                                 actual_match = match(cur, len, at_begin);
4231                                 if (actual_match.match_len == max_match.match_len) {
4232                                         old_cur = cur;
4233                                         max_match = actual_match;
4234                                 }
4235                         }
4236                 }
4237                 if (len == 0)
4238                         return fail;
4239                 else {
4240                         max_match.pos_len = len;
4241                         displayMres(max_match, "SEARCH RESULT", cur)
4242                                         return max_match;
4243                 }
4244         }
4245 }
4246
4247 /// Finds forward
4248 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
4249 {
4250         if (!cur)
4251                 return 0;
4252         bool repeat = false;
4253         DocIterator orig_cur;   // to be used if repeat not successful
4254         MatchResult orig_mres;
4255         while (!theApp()->longOperationCancelled() && cur) {
4256                 //(void) findAdvForwardInnermost(cur);
4257                 LYXERR(Debug::FINDVERBOSE, "findForwardAdv() cur: " << cur);
4258                 MatchResult mres = match(cur, -1, false);
4259                 string msg = "Starting";
4260                 if (repeat)
4261                         msg = "Repeated";
4262                 displayMres(mres, msg + " findForwardAdv", cur)
4263                                 int match_len = mres.match_len;
4264                 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
4265                         LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
4266                         match_len = 0;
4267                 }
4268                 if (match_len <= 0) {
4269                         // This should exit nested insets, if any, or otherwise undefine the currsor.
4270                         cur.pos() = cur.lastpos();
4271                         LYXERR(Debug::FINDVERBOSE, "Advancing pos: cur=" << cur);
4272                         cur.forwardPos();
4273                 }
4274                 else {  // match_len > 0
4275                         // Try to find the begin of searched string
4276                         int increment;
4277                         int firstInvalid = cur.lastpos() - cur.pos();
4278                         {
4279                                 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
4280                                 int incrcur = (firstInvalid + 1 )*3/4;
4281                                 if (incrcur < incrmatch)
4282                                         increment = incrcur;
4283                                 else
4284                                         increment = incrmatch;
4285                                 if (increment < 1)
4286                                         increment = 1;
4287                         }
4288                         LYXERR(Debug::FINDVERBOSE, "Set increment to " << increment);
4289                         while (increment > 0) {
4290                                 DocIterator old_cur = cur;
4291                                 if (cur.pos() + increment >= cur.lastpos()) {
4292                                         increment /= 2;
4293                                         continue;
4294                                 }
4295                                 cur.pos() = cur.pos() + increment;
4296                                 MatchResult mres2 = match(cur, -1, false);
4297                                 displayMres(mres2, "findForwardAdv loop", cur)
4298                                                 switch (interpretMatch(mres, mres2)) {
4299                                         case MatchResult::newIsTooFar:
4300                                                 // behind the expected match
4301                                                 firstInvalid = increment;
4302                                                 cur = old_cur;
4303                                                 increment /= 2;
4304                                                 break;
4305                                         case MatchResult::newIsBetter:
4306                                                 // not reached yet, but cur.pos()+increment is bettert
4307                                                 mres = mres2;
4308                                                 firstInvalid -= increment;
4309                                                 if (increment > firstInvalid*3/4)
4310                                                         increment = firstInvalid*3/4;
4311                                                 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
4312                                                         if (increment >= mres2.match_prefix)
4313                                                                 increment = (mres2.match_prefix+1)*3/4;
4314                                                 }
4315                                                 break;
4316                                         default:
4317                                                 // Todo@
4318                                                 // Handle not like MatchResult::newIsTooFar
4319                                                 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
4320                                                 firstInvalid--;
4321                                                 increment = increment*3/4;
4322                                                 cur = old_cur;
4323                                                 break;
4324                                 }
4325                         }
4326                         if (mres.match_len > 0) {
4327                                 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
4328                                         // The match seems to indicate some deeper level
4329                                         repeat = true;
4330                                         orig_cur = cur;
4331                                         orig_mres = mres;
4332                                         cur.forwardPos();
4333                                         continue;
4334                                 }
4335                         }
4336                         else if (repeat) {
4337                                 // should never be reached.
4338                                 cur = orig_cur;
4339                                 mres = orig_mres;
4340                         }
4341                         // LYXERR0("Leaving first loop");
4342                         LYXERR(Debug::FINDVERBOSE, "Finalizing 1");
4343                         MatchResult found_match = findAdvFinalize(cur, match, mres);
4344                         if (found_match.match_len > 0) {
4345                                 match.FillResults(found_match);
4346                                 return found_match.pos_len;
4347                         }
4348                         else {
4349                                 // try next possible match
4350                                 cur.forwardPos();
4351                                 repeat = false;
4352                                 continue;
4353                         }
4354                 }
4355         }
4356         return 0;
4357 }
4358
4359
4360 /// Find the most backward consecutive match within same paragraph while searching backwards.
4361 MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, MatchResult &expected)
4362 {
4363         DocIterator cur_begin = cur;
4364         cur_begin.pos() = 0;
4365         DocIterator tmp_cur = cur;
4366         MatchResult mr = findAdvFinalize(tmp_cur, match, expected);
4367         Inset & inset = cur.inset();
4368         for (; cur != cur_begin; cur.backwardPos()) {
4369                 LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): cur=" << cur);
4370                 DocIterator new_cur = cur;
4371                 new_cur.backwardPos();
4372                 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
4373                         break;
4374                 MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
4375                 if (new_mr.match_len == mr.match_len)
4376                         break;
4377                 mr = new_mr;
4378         }
4379         LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): exiting with cur=" << cur);
4380         return mr;
4381 }
4382
4383
4384 /// Finds backwards
4385 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
4386 {
4387         if (! cur)
4388                 return 0;
4389         // Backup of original position
4390         DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4391         if (cur == cur_begin)
4392                 return 0;
4393         cur.backwardPos();
4394         DocIterator cur_orig(cur);
4395         bool pit_changed = false;
4396         do {
4397                 cur.pos() = 0;
4398                 MatchResult found_match = match(cur, -1, false);
4399
4400                 if (found_match.match_len > 0) {
4401                         if (pit_changed)
4402                                 cur.pos() = cur.lastpos();
4403                         else
4404                                 cur.pos() = cur_orig.pos();
4405                         LYXERR(Debug::FINDVERBOSE, "findBackAdv2: cur: " << cur);
4406                         DocIterator cur_prev_iter;
4407                         do {
4408                                 found_match = match(cur);
4409                                 LYXERR(Debug::FINDVERBOSE, "findBackAdv3: found_match="
4410                                        << (found_match.match_len > 0) << ", cur: " << cur);
4411                                 if (found_match.match_len > 0) {
4412                                         MatchResult found_mr = findMostBackwards(cur, match, found_match);
4413                                         if (found_mr.pos_len > 0) {
4414                                                 match.FillResults(found_mr);
4415                                                 return found_mr.pos_len;
4416                                         }
4417                                 }
4418
4419                                 // Stop if begin of document reached
4420                                 if (cur == cur_begin)
4421                                         break;
4422                                 cur_prev_iter = cur;
4423                                 cur.backwardPos();
4424                         } while (true);
4425                 }
4426                 if (cur == cur_begin)
4427                         break;
4428                 if (cur.pit() > 0)
4429                         --cur.pit();
4430                 else
4431                         cur.backwardPos();
4432                 pit_changed = true;
4433         } while (!theApp()->longOperationCancelled());
4434         return 0;
4435 }
4436
4437
4438 } // namespace
4439
4440
4441 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4442                                  DocIterator const & cur, int len)
4443 {
4444         if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4445                 return docstring();
4446         if (!opt.ignoreformat)
4447                 return latexifyFromCursor(cur, len);
4448         else
4449                 return stringifyFromCursor(cur, len);
4450 }
4451
4452
4453 FindAndReplaceOptions::FindAndReplaceOptions(
4454                 docstring const & _find_buf_name, bool _casesensitive,
4455                 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4456                 docstring const & _repl_buf_name, bool _keep_case,
4457                 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4458         : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4459           forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4460           repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4461 {
4462 }
4463
4464
4465 namespace {
4466
4467
4468 /** Check if 'len' letters following cursor are all non-lowercase */
4469 static bool allNonLowercase(Cursor const & cur, int len)
4470 {
4471         pos_type beg_pos = cur.selectionBegin().pos();
4472         pos_type end_pos = cur.selectionBegin().pos() + len;
4473         if (len > cur.lastpos() + 1 - beg_pos) {
4474                 LYXERR(Debug::FINDVERBOSE, "This should not happen, more debug needed");
4475                 len = cur.lastpos() + 1 - beg_pos;
4476                 end_pos = beg_pos + len;
4477         }
4478         for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4479                 if (isLowerCase(cur.paragraph().getChar(pos)))
4480                         return false;
4481         return true;
4482 }
4483
4484
4485 /** Check if first letter is upper case and second one is lower case */
4486 static bool firstUppercase(Cursor const & cur)
4487 {
4488         char_type ch1, ch2;
4489         pos_type pos = cur.selectionBegin().pos();
4490         if (pos >= cur.lastpos() - 1) {
4491                 LYXERR(Debug::FINDVERBOSE, "No upper-case at cur: " << cur);
4492                 return false;
4493         }
4494         ch1 = cur.paragraph().getChar(pos);
4495         ch2 = cur.paragraph().getChar(pos + 1);
4496         bool result = isUpperCase(ch1) && isLowerCase(ch2);
4497         LYXERR(Debug::FINDVERBOSE, "firstUppercase(): "
4498                << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4499                << ch2 << "(" << char(ch2) << ")"
4500                << ", result=" << result << ", cur=" << cur);
4501         return result;
4502 }
4503
4504
4505 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4506  **
4507  ** \fixme What to do with possible further paragraphs in replace buffer ?
4508  **/
4509 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4510 {
4511         ParagraphList::iterator pit = buffer.paragraphs().begin();
4512         LASSERT(!pit->empty(), /**/);
4513         pos_type right = pos_type(1);
4514         pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4515         right = pit->size();
4516         pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4517 }
4518 } // namespace
4519
4520 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4521 {
4522         // Should replace the string "$" + std::to_string(matchnum) with replacement
4523         // if the char '$' is not prefixed with odd number of char '\\'
4524         static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4525         string s;
4526         size_t lastpos = 0;
4527         smatch sub;
4528         for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4529                 sub = *it;
4530                 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4531                         continue;
4532                 int num = stoi(sub.str(3), nullptr, 10);
4533                 if (num >= maxmatchnum)
4534                         continue;
4535                 if (lastpos < (size_t) sub.position(2))
4536                         s += t.substr(lastpos, sub.position(2) - lastpos);
4537                 s += replacements[num];
4538                 lastpos = sub.position(2) + sub.length(2);
4539         }
4540         if (lastpos == 0)
4541                 return false;
4542         else if (lastpos < t.length())
4543                 s += t.substr(lastpos, t.length() - lastpos);
4544         t = s;
4545         return true;
4546 }
4547
4548 ///
4549 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4550 {
4551         Cursor & cur = bv->cursor();
4552         if (opt.repl_buf_name.empty()
4553                         || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4554                         || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4555                 return 0;
4556
4557         DocIterator sel_beg = cur.selectionBegin();
4558         DocIterator sel_end = cur.selectionEnd();
4559         if (&sel_beg.inset() != &sel_end.inset()
4560                         || sel_beg.pit() != sel_end.pit()
4561                         || sel_beg.idx() != sel_end.idx())
4562                 return 0;
4563         int sel_len = sel_end.pos() - sel_beg.pos();
4564         LYXERR(Debug::FINDVERBOSE, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4565                << ", sel_len: " << sel_len << endl);
4566         if (sel_len == 0)
4567                 return 0;
4568         LASSERT(sel_len > 0, return 0);
4569
4570         if (!matchAdv(sel_beg, sel_len).match_len)
4571                 return 0;
4572
4573         // Build a copy of the replace buffer, adapted to the KeepCase option
4574         Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4575         ostringstream oss;
4576         repl_buffer_orig.write(oss);
4577         string lyx = oss.str();
4578         if (matchAdv.valid_matches > 0)
4579                 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4580         Buffer repl_buffer(string(), false);
4581         repl_buffer.setInternal(true);
4582         repl_buffer.setUnnamed(true);
4583         LASSERT(repl_buffer.readString(lyx), return 0);
4584         if (opt.keep_case && sel_len >= 2) {
4585                 LYXERR(Debug::FINDVERBOSE, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4586                 if (cur.inTexted()) {
4587                         if (firstUppercase(cur))
4588                                 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4589                         else if (allNonLowercase(cur, sel_len))
4590                                 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4591                 }
4592         }
4593         cap::cutSelection(cur, false);
4594         if (cur.inTexted()) {
4595                 repl_buffer.changeLanguage(
4596                                         repl_buffer.language(),
4597                                         cur.getFont().language());
4598                 LYXERR(Debug::FINDVERBOSE, "Replacing by pasteParagraphList()ing repl_buffer");
4599                 LYXERR(Debug::FINDVERBOSE, "Before pasteParagraphList() cur=" << cur << endl);
4600                 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4601                                         repl_buffer.params().documentClassPtr(),
4602                                         repl_buffer.params().authors(),
4603                                         bv->buffer().errorList("Paste"));
4604                 LYXERR(Debug::FINDVERBOSE, "After pasteParagraphList() cur=" << cur << endl);
4605                 sel_len = repl_buffer.paragraphs().begin()->size();
4606         } else if (cur.inMathed()) {
4607                 odocstringstream ods;
4608                 otexstream os(ods);
4609                 // OutputParams runparams(&repl_buffer.params().encoding());
4610                 OutputParams runparams(encodings.fromLyXName("utf8"));
4611                 runparams.nice = false;
4612                 runparams.flavor = Flavor::XeTeX;
4613                 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4614                 runparams.dryrun = true;
4615                 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams, string(), -1, -1, true);
4616                 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4617                 docstring repl_latex = ods.str();
4618                 LYXERR(Debug::FINDVERBOSE, "Latexified replace_buffer: '" << repl_latex << "'");
4619                 string s;
4620                 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4621                 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4622                 repl_latex = from_utf8(s);
4623                 LYXERR(Debug::FINDVERBOSE, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4624                 MathData ar(cur.buffer());
4625                 asArray(repl_latex, ar, Parse::NORMAL);
4626                 cur.insert(ar);
4627                 sel_len = ar.size();
4628                 LYXERR(Debug::FINDVERBOSE, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4629         }
4630         if (cur.pos() >= sel_len)
4631                 cur.pos() -= sel_len;
4632         else
4633                 cur.pos() = 0;
4634         LYXERR(Debug::FINDVERBOSE, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4635         bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4636         bv->processUpdateFlags(Update::Force);
4637         return 1;
4638 }
4639
4640
4641 /// Perform a FindAdv operation.
4642 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4643 {
4644         DocIterator cur;
4645         int pos_len = 0;
4646
4647         // e.g., when invoking word-findadv from mini-buffer wither with
4648         //       wrong options syntax or before ever opening advanced F&R pane
4649         if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4650                 return false;
4651
4652         try {
4653                 MatchStringAdv matchAdv(bv->buffer(), opt);
4654 #if QTSEARCH
4655                 if (!matchAdv.regexIsValid) {
4656                         bv->message(lyx::from_utf8(matchAdv.regexError));
4657                         return(false);
4658                 }
4659 #endif
4660                 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4661                 if (length > 0)
4662                         bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4663                 num_replaced += findAdvReplace(bv, opt, matchAdv);
4664                 cur = bv->cursor();
4665                 if (opt.forward)
4666                         pos_len = findForwardAdv(cur, matchAdv);
4667                 else
4668                         pos_len = findBackwardsAdv(cur, matchAdv);
4669         } catch (exception & ex) {
4670                 bv->message(from_utf8(ex.what()));
4671                 return false;
4672         }
4673
4674         if (pos_len == 0) {
4675                 if (num_replaced > 0) {
4676                         switch (num_replaced)
4677                         {
4678                         case 1:
4679                                 bv->message(_("One match has been replaced."));
4680                                 break;
4681                         case 2:
4682                                 bv->message(_("Two matches have been replaced."));
4683                                 break;
4684                         default:
4685                                 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4686                                 break;
4687                         }
4688                         num_replaced = 0;
4689                 }
4690                 else {
4691                         bv->message(_("Match not found."));
4692                 }
4693                 return false;
4694         }
4695
4696         if (num_replaced > 0)
4697                 bv->message(_("Match has been replaced."));
4698         else
4699                 bv->message(_("Match found."));
4700
4701         if (cur.pos() + pos_len > cur.lastpos()) {
4702                 // Prevent crash in bv->putSelectionAt()
4703                 // Should never happen, maybe LASSERT() here?
4704                 pos_len = cur.lastpos() - cur.pos();
4705         }
4706         LYXERR(Debug::FINDVERBOSE|Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4707         bv->putSelectionAt(cur, pos_len, !opt.forward);
4708
4709         return true;
4710 }
4711
4712
4713 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4714 {
4715         os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4716            << opt.casesensitive << ' '
4717            << opt.matchword << ' '
4718            << opt.forward << ' '
4719            << opt.expandmacros << ' '
4720            << opt.ignoreformat << ' '
4721            << opt.replace_all << ' '
4722            << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4723            << opt.keep_case << ' '
4724            << int(opt.scope) << ' '
4725            << int(opt.restr);
4726
4727         LYXERR(Debug::FINDVERBOSE, "built: " << os.str());
4728
4729         return os;
4730 }
4731
4732
4733 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4734 {
4735         // LYXERR(Debug::FINDVERBOSE, "parsing");
4736         string s;
4737         string line;
4738         getline(is, line);
4739         while (line != "EOSS") {
4740                 if (! s.empty())
4741                         s = s + "\n";
4742                 s = s + line;
4743                 if (is.eof())   // Tolerate malformed request
4744                         break;
4745                 getline(is, line);
4746         }
4747         // LYXERR(Debug::FINDVERBOSE, "file_buf_name: '" << s << "'");
4748         opt.find_buf_name = from_utf8(s);
4749         is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4750         is.get();       // Waste space before replace string
4751         s = "";
4752         getline(is, line);
4753         while (line != "EOSS") {
4754                 if (! s.empty())
4755                         s = s + "\n";
4756                 s = s + line;
4757                 if (is.eof())   // Tolerate malformed request
4758                         break;
4759                 getline(is, line);
4760         }
4761         // LYXERR(Debug::FINDVERBOSE, "repl_buf_name: '" << s << "'");
4762         opt.repl_buf_name = from_utf8(s);
4763         is >> opt.keep_case;
4764         int i;
4765         is >> i;
4766         opt.scope = FindAndReplaceOptions::SearchScope(i);
4767         is >> i;
4768         opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4769
4770         /*
4771         LYXERR(Debug::FINDVERBOSE, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4772                << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4773                << opt.scope << ' ' << opt.restr);
4774         */
4775         return is;
4776 }
4777
4778 } // namespace lyx