src/lyxfind.cpp

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