3 * This file is part of LyX, the document processor.
4 * License details can be found in the file COPYING.
6 * \author Lars Gullik Bjønnes
9 * \author Alfredo Braunstein
10 * \author Tommaso Cucinotta
11 * \author Kornel Benko
13 * Full author contact details are available in file CREDITS.
21 #include "BufferList.h"
22 #include "BufferParams.h"
23 #include "BufferView.h"
26 #include "CutAndPaste.h"
27 #include "FuncRequest.h"
29 #include "output_latex.h"
30 #include "OutputParams.h"
31 #include "Paragraph.h"
35 #include "frontends/Application.h"
36 #include "frontends/alert.h"
38 #include "mathed/InsetMath.h"
39 #include "mathed/InsetMathHull.h"
40 #include "mathed/MathData.h"
41 #include "mathed/MathStream.h"
42 #include "mathed/MathSupport.h"
44 #include "support/debug.h"
45 #include "support/docstream.h"
46 #include "support/FileName.h"
47 #include "support/gettext.h"
48 #include "support/lassert.h"
49 #include "support/lstrings.h"
50 #include "support/textutils.h"
55 //#define ResultsDebug
56 #define USE_QT_FOR_SEARCH
57 #if defined(USE_QT_FOR_SEARCH)
58 #include <QtCore> // sets QT_VERSION
59 #if (QT_VERSION >= 0x050000)
60 #include <QRegularExpression>
70 using namespace lyx::support;
75 // Helper class for deciding what should be ignored
79 IgnoreFormats() = default;
81 bool getFamily() const { return ignoreFamily_; }
83 bool getSeries() const { return ignoreSeries_; }
85 bool getShape() const { return ignoreShape_; }
87 bool getUnderline() const { return ignoreUnderline_; }
89 bool getMarkUp() const { return ignoreMarkUp_; }
91 bool getStrikeOut() const { return ignoreStrikeOut_; }
93 bool getSectioning() const { return ignoreSectioning_; }
95 bool getFrontMatter() const { return ignoreFrontMatter_; }
97 bool getColor() const { return ignoreColor_; }
99 bool getLanguage() const { return ignoreLanguage_; }
101 void setIgnoreFormat(string const & type, bool value);
105 bool ignoreFamily_ = false;
107 bool ignoreSeries_ = false;
109 bool ignoreShape_ = false;
111 bool ignoreUnderline_ = false;
113 bool ignoreMarkUp_ = false;
115 bool ignoreStrikeOut_ = false;
117 bool ignoreSectioning_ = false;
119 bool ignoreFrontMatter_ = false;
121 bool ignoreColor_ = false;
123 bool ignoreLanguage_ = false;
127 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
129 if (type == "color") {
130 ignoreColor_ = value;
132 else if (type == "language") {
133 ignoreLanguage_ = value;
135 else if (type == "sectioning") {
136 ignoreSectioning_ = value;
137 ignoreFrontMatter_ = value;
139 else if (type == "font") {
140 ignoreSeries_ = value;
141 ignoreShape_ = value;
142 ignoreFamily_ = value;
144 else if (type == "series") {
145 ignoreSeries_ = value;
147 else if (type == "shape") {
148 ignoreShape_ = value;
150 else if (type == "family") {
151 ignoreFamily_ = value;
153 else if (type == "markup") {
154 ignoreMarkUp_ = value;
156 else if (type == "underline") {
157 ignoreUnderline_ = value;
159 else if (type == "strike") {
160 ignoreStrikeOut_ = value;
164 // The global variable that can be changed from outside
165 IgnoreFormats ignoreFormats;
168 void setIgnoreFormat(string const & type, bool value)
170 ignoreFormats.setIgnoreFormat(type, value);
176 bool parse_bool(docstring & howto)
181 howto = split(howto, var, ' ');
189 MatchString(docstring const & s, bool cs, bool mw)
190 : str(s), case_sens(cs), whole_words(mw)
193 // returns true if the specified string is at the specified position
194 // del specifies whether deleted strings in ct mode will be considered
195 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
197 return par.find(str, case_sens, whole_words, pos, del);
205 // match whole words only
210 int findForward(DocIterator & cur, MatchString const & match,
211 bool find_del = true)
213 for (; cur; cur.forwardChar())
214 if (cur.inTexted()) {
215 int len = match(cur.paragraph(), cur.pos(), find_del);
223 int findBackwards(DocIterator & cur, MatchString const & match,
224 bool find_del = true)
228 if (cur.inTexted()) {
229 int len = match(cur.paragraph(), cur.pos(), find_del);
238 bool searchAllowed(docstring const & str)
241 frontend::Alert::error(_("Search error"), _("Search string is empty"));
248 bool findOne(BufferView * bv, docstring const & searchstr,
249 bool case_sens, bool whole, bool forward,
250 bool find_del = true, bool check_wrap = false)
252 if (!searchAllowed(searchstr))
255 DocIterator cur = forward
256 ? bv->cursor().selectionEnd()
257 : bv->cursor().selectionBegin();
259 MatchString const match(searchstr, case_sens, whole);
261 int match_len = forward
262 ? findForward(cur, match, find_del)
263 : findBackwards(cur, match, find_del);
266 bv->putSelectionAt(cur, match_len, !forward);
267 else if (check_wrap) {
268 DocIterator cur_orig(bv->cursor());
271 q = _("End of file reached while searching forward.\n"
272 "Continue searching from the beginning?");
274 q = _("Beginning of file reached while searching backward.\n"
275 "Continue searching from the end?");
276 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
277 q, 0, 1, _("&Yes"), _("&No"));
278 if (wrap_answer == 0) {
280 bv->cursor().clear();
281 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
283 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
284 bv->cursor().backwardPos();
286 bv->clearSelection();
287 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
290 bv->cursor().setCursor(cur_orig);
294 return match_len > 0;
298 int replaceAll(BufferView * bv,
299 docstring const & searchstr, docstring const & replacestr,
300 bool case_sens, bool whole)
302 Buffer & buf = bv->buffer();
304 if (!searchAllowed(searchstr) || buf.isReadonly())
307 DocIterator cur_orig(bv->cursor());
309 MatchString const match(searchstr, case_sens, whole);
312 int const rsize = replacestr.size();
313 int const ssize = searchstr.size();
316 cur.setCursor(doc_iterator_begin(&buf));
317 int match_len = findForward(cur, match, false);
318 while (match_len > 0) {
319 // Backup current cursor position and font.
320 pos_type const pos = cur.pos();
321 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
323 int striked = ssize -
324 cur.paragraph().eraseChars(pos, pos + match_len,
325 buf.params().track_changes);
326 cur.paragraph().insert(pos, replacestr, font,
327 Change(buf.params().track_changes
329 : Change::UNCHANGED));
330 for (int i = 0; i < rsize + striked; ++i)
333 match_len = findForward(cur, match, false);
336 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
338 cur_orig.fixIfBroken();
339 bv->setCursor(cur_orig);
345 // the idea here is that we are going to replace the string that
346 // is selected IF it is the search string.
347 // if there is a selection, but it is not the search string, then
348 // we basically ignore it. (FIXME We ought to replace only within
350 // if there is no selection, then:
351 // (i) if some search string has been provided, then we find it.
352 // (think of how the dialog works when you hit "replace" the
354 // (ii) if no search string has been provided, then we treat the
355 // word the cursor is in as the search string. (why? i have no
356 // idea.) but this only works in text?
358 // returns the number of replacements made (one, if any) and
359 // whether anything at all was done.
360 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
361 docstring const & replacestr, bool case_sens,
362 bool whole, bool forward, bool findnext)
364 Cursor & cur = bv->cursor();
365 if (!cur.selection()) {
366 // no selection, non-empty search string: find it
367 if (!searchstr.empty()) {
368 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
369 return make_pair(found, 0);
371 // empty search string
374 return make_pair(false, 0);
375 // select current word and treat it as the search string.
376 // This causes a minor bug as undo will restore this selection,
377 // which the user did not create (#8986).
378 cur.innerText()->selectWord(cur, WHOLE_WORD);
379 searchstr = cur.selectionAsString(false);
382 // if we still don't have a search string, report the error
384 if (!searchAllowed(searchstr))
385 return make_pair(false, 0);
387 bool have_selection = cur.selection();
388 docstring const selected = cur.selectionAsString(false);
391 ? searchstr == selected
392 : compare_no_case(searchstr, selected) == 0;
394 // no selection or current selection is not search word:
395 // just find the search word
396 if (!have_selection || !match) {
397 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
398 return make_pair(found, 0);
401 // we're now actually ready to replace. if the buffer is
402 // read-only, we can't, though.
403 if (bv->buffer().isReadonly())
404 return make_pair(false, 0);
406 cap::replaceSelectionWithString(cur, replacestr);
408 cur.pos() += replacestr.length();
409 LASSERT(cur.pos() <= cur.lastpos(),
410 cur.pos() = cur.lastpos());
413 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
415 return make_pair(true, 1);
421 docstring const find2string(docstring const & search,
422 bool casesensitive, bool matchword, bool forward)
426 << int(casesensitive) << ' '
427 << int(matchword) << ' '
433 docstring const replace2string(docstring const & replace,
434 docstring const & search,
435 bool casesensitive, bool matchword,
436 bool all, bool forward, bool findnext)
439 ss << replace << '\n'
441 << int(casesensitive) << ' '
442 << int(matchword) << ' '
444 << int(forward) << ' '
450 bool lyxfind(BufferView * bv, FuncRequest const & ev)
452 if (!bv || ev.action() != LFUN_WORD_FIND)
455 //lyxerr << "find called, cmd: " << ev << endl;
457 // data is of the form
459 // <casesensitive> <matchword> <forward>"
461 docstring howto = split(ev.argument(), search, '\n');
463 bool casesensitive = parse_bool(howto);
464 bool matchword = parse_bool(howto);
465 bool forward = parse_bool(howto);
467 return findOne(bv, search, casesensitive, matchword, forward, true, true);
471 bool lyxreplace(BufferView * bv,
472 FuncRequest const & ev, bool has_deleted)
474 if (!bv || ev.action() != LFUN_WORD_REPLACE)
477 // data is of the form
480 // <casesensitive> <matchword> <all> <forward> <findnext>"
483 docstring howto = split(ev.argument(), rplc, '\n');
484 howto = split(howto, search, '\n');
486 bool casesensitive = parse_bool(howto);
487 bool matchword = parse_bool(howto);
488 bool all = parse_bool(howto);
489 bool forward = parse_bool(howto);
490 bool findnext = howto.empty() ? true : parse_bool(howto);
495 int replace_count = 0;
497 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
498 update = replace_count > 0;
501 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
503 replace_count = rv.second;
506 Buffer const & buf = bv->buffer();
508 // emit message signal.
509 buf.message(_("String not found."));
511 if (replace_count == 0) {
512 buf.message(_("String found."));
513 } else if (replace_count == 1) {
514 buf.message(_("String has been replaced."));
516 docstring const str =
517 bformat(_("%1$d strings have been replaced."), replace_count);
521 } else if (findnext) {
522 // if we have deleted characters, we do not replace at all, but
523 // rather search for the next occurence
524 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
527 bv->message(_("String not found."));
533 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
535 for (; cur; cur.forwardPos())
536 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
540 DocIterator cur_orig(bv->cursor());
541 docstring q = _("End of file reached while searching forward.\n"
542 "Continue searching from the beginning?");
543 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
544 q, 0, 1, _("&Yes"), _("&No"));
545 if (wrap_answer == 0) {
546 bv->cursor().clear();
547 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
548 bv->clearSelection();
549 cur.setCursor(bv->cursor().selectionBegin());
550 if (findNextChange(bv, cur, false))
553 bv->cursor().setCursor(cur_orig);
560 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
562 for (cur.backwardPos(); cur; cur.backwardPos()) {
563 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
568 DocIterator cur_orig(bv->cursor());
569 docstring q = _("Beginning of file reached while searching backward.\n"
570 "Continue searching from the end?");
571 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
572 q, 0, 1, _("&Yes"), _("&No"));
573 if (wrap_answer == 0) {
574 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
575 bv->cursor().backwardPos();
576 bv->clearSelection();
577 cur.setCursor(bv->cursor().selectionBegin());
578 if (findPreviousChange(bv, cur, false))
581 bv->cursor().setCursor(cur_orig);
588 bool selectChange(Cursor & cur, bool forward)
590 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
592 Change ch = cur.paragraph().lookupChange(cur.pos());
594 CursorSlice tip1 = cur.top();
595 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
596 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
597 if (!ch2.isSimilarTo(ch))
600 CursorSlice tip2 = cur.top();
601 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
603 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
604 if (!ch2.isSimilarTo(ch)) {
605 // take a step forward to correctly set the selection
613 cur.bv().mouseSetCursor(cur, false);
615 cur.bv().mouseSetCursor(cur, true);
623 bool findChange(BufferView * bv, bool forward)
626 cur.setCursor(forward ? bv->cursor().selectionEnd()
627 : bv->cursor().selectionBegin());
628 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
629 return selectChange(cur, forward);
634 bool findNextChange(BufferView * bv)
636 return findChange(bv, true);
640 bool findPreviousChange(BufferView * bv)
642 return findChange(bv, false);
649 typedef vector<pair<string, string> > Escapes;
651 /// A map of symbols and their escaped equivalent needed within a regex.
652 /// @note Beware of order
653 Escapes const & get_regexp_escapes()
655 typedef std::pair<std::string, std::string> P;
657 static Escapes escape_map;
658 if (escape_map.empty()) {
659 escape_map.push_back(P("$", "_x_$"));
660 escape_map.push_back(P("{", "_x_{"));
661 escape_map.push_back(P("}", "_x_}"));
662 escape_map.push_back(P("[", "_x_["));
663 escape_map.push_back(P("]", "_x_]"));
664 escape_map.push_back(P("(", "_x_("));
665 escape_map.push_back(P(")", "_x_)"));
666 escape_map.push_back(P("+", "_x_+"));
667 escape_map.push_back(P("*", "_x_*"));
668 escape_map.push_back(P(".", "_x_."));
669 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
670 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
671 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
672 escape_map.push_back(P("_x_", "\\"));
677 /// A map of lyx escaped strings and their unescaped equivalent.
678 Escapes const & get_lyx_unescapes()
680 typedef std::pair<std::string, std::string> P;
682 static Escapes escape_map;
683 if (escape_map.empty()) {
684 escape_map.push_back(P("\\%", "%"));
685 escape_map.push_back(P("\\{", "{"));
686 escape_map.push_back(P("\\}", "}"));
687 escape_map.push_back(P("\\mathcircumflex ", "^"));
688 escape_map.push_back(P("\\mathcircumflex", "^"));
689 escape_map.push_back(P("\\backslash ", "\\"));
690 escape_map.push_back(P("\\backslash", "\\"));
691 escape_map.push_back(P("\\sim ", "~"));
692 escape_map.push_back(P("\\sim", "~"));
697 /// A map of escapes turning a regexp matching text to one matching latex.
698 Escapes const & get_regexp_latex_escapes()
700 typedef std::pair<std::string, std::string> P;
702 static Escapes escape_map;
703 if (escape_map.empty()) {
704 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
705 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
706 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
707 escape_map.push_back(P("\\[", "\\{\\[\\}"));
708 escape_map.push_back(P("\\]", "\\{\\]\\}"));
709 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
710 escape_map.push_back(P("%", "\\\\\\%"));
711 escape_map.push_back(P("#", "\\\\#"));
716 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
717 ** the found occurrence were escaped.
719 string apply_escapes(string s, Escapes const & escape_map)
721 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
722 Escapes::const_iterator it;
723 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
724 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
725 unsigned int pos = 0;
726 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
727 s.replace(pos, it->first.length(), it->second);
728 LYXERR(Debug::FIND, "After escape: " << s);
729 pos += it->second.length();
730 // LYXERR(Debug::FIND, "pos: " << pos);
733 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
738 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
739 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
740 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
741 string escape_for_regex(string s, bool match_latex)
744 while (pos < s.size()) {
745 size_t new_pos = s.find("\\regexp{", pos);
746 if (new_pos == string::npos)
751 LYXERR(Debug::FIND, "new_pos: " << new_pos);
752 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
753 LYXERR(Debug::FIND, "t [lyx]: " << t);
754 t = apply_escapes(t, get_regexp_escapes());
755 LYXERR(Debug::FIND, "t [rxp]: " << t);
756 s.replace(pos, new_pos - pos, t);
757 new_pos = pos + t.size();
758 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
759 LYXERR(Debug::FIND, "new_pos: " << new_pos);
760 if (new_pos == s.size())
763 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
764 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
765 LYXERR(Debug::FIND, "end_pos: " << end_pos);
766 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
767 LYXERR(Debug::FIND, "t in regexp : " << t);
768 t = apply_escapes(t, get_lyx_unescapes());
769 LYXERR(Debug::FIND, "t in regexp after unescapes [lyx]: " << t);
771 t = apply_escapes(t, get_regexp_latex_escapes());
772 LYXERR(Debug::FIND, "t in regexp after latex_escapes [ltx]: " << t);
774 if (end_pos == s.size()) {
775 s.replace(new_pos, end_pos - new_pos, t);
776 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
779 s.replace(new_pos, end_pos + 13 - new_pos, t);
780 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
781 pos = new_pos + t.size();
782 LYXERR(Debug::FIND, "pos: " << pos);
788 /// Wrapper for lyx::regex_replace with simpler interface
789 bool regex_replace(string const & s, string & t, string const & searchstr,
790 string const & replacestr)
792 regex e(searchstr, regex_constants::ECMAScript);
794 ostream_iterator<char, char> it(oss);
795 regex_replace(it, s.begin(), s.end(), e, replacestr);
796 // tolerate t and s be references to the same variable
797 bool rv = (s != oss.str());
803 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
805 ** Verify that closed braces exactly match open braces. This avoids that, for example,
806 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
809 ** Number of open braces that must remain open at the end for the verification to succeed.
812 bool braces_match(QString const & beg,
815 bool braces_match(string const & beg,
821 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg.toStdString() << "'");
823 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg << "'");
825 int lastidx = beg.size();
826 for (int i=0; i < lastidx; ++i) {
827 // Skip escaped braces in the count
837 } else if (c == '{') {
839 } else if (c == '}') {
840 if (open_pars == 0) {
841 LYXERR(Debug::FIND, "Found unmatched closed brace");
847 if (open_pars != unmatched) {
848 LYXERR(Debug::FIND, "Found " << open_pars
849 << " instead of " << unmatched
850 << " unmatched open braces at the end of count");
853 LYXERR(Debug::FIND, "Braces match as expected");
871 vector <string> result = vector <string>();
872 MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1) {};
875 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
877 int range = oldres.match_len;
878 if (range < 2) range = 2;
879 if (newres.match2end < oldres.match2end - oldres.match_len)
880 return MatchResult::newIsTooFar;
881 if (newres.match_len < oldres.match_len)
882 return MatchResult::newIsTooFar;
883 if ((newres.match_len == oldres.match_len) &&
884 (newres.match2end < oldres.match2end + range) &&
885 (newres.match2end > oldres.match2end - range)) {
886 return MatchResult::newIsBetter;
888 return MatchResult::newIsInvalid;
891 /** The class performing a match between a position in the document and the FindAdvOptions.
894 class MatchStringAdv {
896 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
898 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
899 ** constructor as opt.search, under the opt.* options settings.
902 ** If set, then match is searched only against beginning of text starting at cur.
903 ** If unset, then match is searched anywhere in text starting at cur.
906 ** The length of the matching text, or zero if no match was found.
908 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
917 /// first buffer on which search was started
918 lyx::Buffer * const p_first_buf;
920 FindAndReplaceOptions const & opt;
923 /// Auxiliary find method (does not account for opt.matchword)
924 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
925 void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
927 /** Normalize a stringified or latexified LyX paragraph.
931 ** <li>if search is not casesensitive, then lowercase the string;
932 ** <li>remove any newline at begin or end of the string;
933 ** <li>replace any newline in the middle of the string with a simple space;
934 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
937 ** @todo Normalization should also expand macros, if the corresponding
938 ** search option was checked.
940 string normalize(docstring const & s) const;
941 // normalized string to search
942 string par_as_string;
943 // regular expression to use for searching
944 // regexp2 is same as regexp, but prefixed with a ".*?"
946 QRegularExpression regexp;
947 QRegularExpression regexp2;
952 // leading format material as string
953 string lead_as_string;
954 // par_as_string after removal of lead_as_string
955 string par_as_string_nolead;
956 // unmatched open braces in the search string/regexp
958 // number of (.*?) subexpressions added at end of search regexp for closing
959 // environments, math mode, styles, etc...
962 // Are we searching with regular expressions ?
964 static int valid_matches;
965 static vector <string> matches;
966 void FillResults(MatchResult &found_mr);
969 int MatchStringAdv::valid_matches = 0;
970 vector <string> MatchStringAdv::matches = vector <string>(10);
972 void MatchStringAdv::FillResults(MatchResult &found_mr)
974 if (found_mr.match_len > 0) {
975 valid_matches = found_mr.result.size();
976 for (size_t i = 0; i < found_mr.result.size(); i++)
977 matches[i] = found_mr.result[i];
983 static docstring buffer_to_latex(Buffer & buffer)
985 //OutputParams runparams(&buffer.params().encoding());
986 OutputParams runparams(encodings.fromLyXName("utf8"));
987 odocstringstream ods;
989 runparams.nice = true;
990 runparams.flavor = Flavor::XeTeX;
991 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
992 // No side effect of file copying and image conversion
993 runparams.dryrun = true;
994 runparams.for_search = true;
995 pit_type const endpit = buffer.paragraphs().size();
996 for (pit_type pit = 0; pit != endpit; ++pit) {
997 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
998 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
1004 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1007 if (!opt.ignoreformat) {
1008 str = buffer_to_latex(buffer);
1010 // OutputParams runparams(&buffer.params().encoding());
1011 OutputParams runparams(encodings.fromLyXName("utf8"));
1012 runparams.nice = true;
1013 runparams.flavor = Flavor::XeTeX;
1014 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1015 runparams.dryrun = true;
1016 runparams.for_search = true;
1017 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1018 Paragraph const & par = buffer.paragraphs().at(pit);
1019 LYXERR(Debug::FIND, "Adding to search string: '"
1020 << par.asString(pos_type(0), par.size(),
1021 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1024 str += par.asString(pos_type(0), par.size(),
1025 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1028 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1029 string t = to_utf8(str);
1030 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
1037 /// Return separation pos between the leading material and the rest
1038 static size_t identifyLeading(string const & s)
1041 // @TODO Support \item[text]
1042 // Kornel: Added textsl, textsf, textit, texttt and noun
1043 // + allow to search for colored text too
1044 while (regex_replace(t, t, "^\\\\(("
1045 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1046 "lyxaddress|lyxrightaddress|"
1047 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1048 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1049 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1050 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1051 || regex_replace(t, t, "^\\$", "")
1052 || regex_replace(t, t, "^\\\\\\[", "")
1053 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1054 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1056 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1061 * Given a latexified string, retrieve some handled features
1062 * The features of the regex will later be compared with the features
1063 * of the searched text. If the regex features are not a
1064 * subset of the analized, then, in not format ignoring search
1065 * we can early stop the search in the relevant inset.
1067 typedef map<string, bool> Features;
1069 static Features identifyFeatures(string const & s)
1071 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1072 static regex const valid("^("
1074 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1075 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1076 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1077 "(u|uu)line|(s|x)out|uwave|"
1078 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1079 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1084 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1087 if (sub.str(1).compare("regexp") == 0) {
1091 string token = sub.str(1);
1093 if (regex_match(token, sub2, valid)) {
1101 if (sub.str(1).compare("endregexp") == 0) {
1111 * defines values features of a key "\\[a-z]+{"
1116 /* Char type with content discarded
1117 * like \hspace{1cm} */
1119 /* Char, like \backslash */
1121 /* replace starting backslash with '#' */
1123 /* \part, \section*, ... */
1125 /* title, author etc */
1127 /* \foreignlanguage{ngerman}, ... */
1130 * to discard language in content */
1133 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1135 /* fonts, colors, markups, ... */
1137 /* footnotesize, ... large, ...
1138 * Ignore all of them */
1141 /* inputencoding, ...
1142 * Discard also content, because they do not help in search */
1145 * like remove, but also all arguments */
1147 /* item, listitem */
1149 /* tex, latex, ... like isChar */
1151 /* like \lettrine[lines=5]{}{} */
1153 // like isStandard, but always remove head
1155 /* End of arguments marker for lettrine,
1156 * so that they can be ignored */
1159 KeyInfo() = default;
1160 KeyInfo(KeyType type, int parcount, bool disable)
1162 parenthesiscount(parcount),
1163 disabled(disable) {}
1164 KeyType keytype = invalid;
1166 int _tokensize = -1;
1167 int _tokenstart = -1;
1168 int _dataStart = -1;
1170 int parenthesiscount = 1;
1171 bool disabled = false;
1172 bool used = false; /* by pattern */
1177 Border(int l=0, int u=0) : low(l), upper(u) {};
1182 #define MAXOPENED 30
1184 bool isPatternString_;
1186 explicit Intervall(bool isPattern, string const & p) :
1187 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1188 hasTitle(false), langcount(0)
1196 static vector<Border> borders;
1197 int depts[MAXOPENED];
1198 int closes[MAXOPENED];
1199 int actualdeptindex;
1200 int previousNotIgnored(int) const;
1201 int nextNotIgnored(int) const;
1202 void handleOpenP(int i);
1203 void handleCloseP(int i, bool closingAllowed);
1204 void resetOpenedP(int openPos);
1205 void addIntervall(int upper);
1206 void addIntervall(int low, int upper); /* if explicit */
1207 void removeAccents();
1208 void setForDefaultLang(KeyInfo const & defLang) const;
1209 int findclosing(int start, int end, char up, char down, int repeat);
1210 void handleParentheses(int lastpos, bool closingAllowed);
1212 int langcount; // Number of disabled language specs up to current position in actual interval
1213 int isOpeningPar(int pos) const;
1215 void output(ostringstream &os, int lastpos);
1216 // string show(int lastpos);
1219 vector<Border> Intervall::borders = vector<Border>(30);
1221 int Intervall::isOpeningPar(int pos) const
1223 if ((pos < 0) || (size_t(pos) >= par.size()))
1225 if (par[pos] != '{')
1227 if (size_t(pos) + 2 >= par.size())
1229 if (par[pos+2] != '}')
1231 if (par[pos+1] == '[' || par[pos+1] == ']')
1236 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1238 // Enable the use of first token again
1239 if (ignoreidx >= 0) {
1240 int value = defLang._tokenstart + defLang._tokensize;
1246 if (borders[borderidx].low < value)
1247 borders[borderidx].low = value;
1248 if (borders[borderidx].upper < value)
1249 borders[borderidx].upper = value;
1254 static void checkDepthIndex(int val)
1256 static int maxdepthidx = MAXOPENED-2;
1257 static int lastmaxdepth = 0;
1258 if (val > lastmaxdepth) {
1259 LYXERR(Debug::INFO, "Depth reached " << val);
1262 if (val > maxdepthidx) {
1264 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1269 // Not needed, because borders are now dynamically expanded
1270 static void checkIgnoreIdx(int val)
1272 static int lastmaxignore = -1;
1273 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1274 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1275 lastmaxignore = val;
1281 * Expand the region of ignored parts of the input latex string
1282 * The region is only relevant in output()
1284 void Intervall::addIntervall(int low, int upper)
1287 if (low == upper) return;
1288 for (idx = ignoreidx+1; idx > 0; --idx) {
1289 if (low > borders[idx-1].upper) {
1293 Border br(low, upper);
1294 if (idx > ignoreidx) {
1295 if (borders.size() <= size_t(idx)) {
1296 borders.push_back(br);
1302 // checkIgnoreIdx(ignoreidx);
1306 // Expand only if one of the new bound is inside the interwall
1307 // We know here that br.low > borders[idx-1].upper
1308 if (br.upper < borders[idx].low) {
1309 // We have to insert at this pos
1310 if (size_t(ignoreidx+1) >= borders.size()) {
1311 borders.push_back(borders[ignoreidx]);
1314 borders[ignoreidx+1] = borders[ignoreidx];
1316 for (int i = ignoreidx; i > idx; --i) {
1317 borders[i] = borders[i-1];
1321 // checkIgnoreIdx(ignoreidx);
1324 // Here we know, that we are overlapping
1325 if (br.low > borders[idx].low)
1326 br.low = borders[idx].low;
1327 // check what has to be concatenated
1329 for (int i = idx; i <= ignoreidx; i++) {
1330 if (br.upper >= borders[i].low) {
1332 if (br.upper < borders[i].upper)
1333 br.upper = borders[i].upper;
1339 // count should be >= 1 here
1342 for (int i = idx + count; i <= ignoreidx; i++) {
1343 borders[i-count+1] = borders[i];
1345 ignoreidx -= count - 1;
1351 typedef map<string, string> AccentsMap;
1352 static AccentsMap accents = map<string, string>();
1354 static void buildaccent(string n, string param, string values)
1358 const char delim = '|';
1359 while (getline(s, name, delim)) {
1361 for (char c : param) {
1362 string key = name + "{" + c + "}";
1363 // get the corresponding utf8-value
1364 if ((values[start] & 0xc0) != 0xc0) {
1365 // should not happen, utf8 encoding starts at least with 11xxxxxx
1366 // but value for '\dot{i}' is 'i', which is ascii
1367 if ((values[start] & 0x80) == 0) {
1369 accents[key] = values.substr(start, 1);
1370 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1375 for (int j = 1; ;j++) {
1376 if (start + j >= values.size()) {
1377 accents[key] = values.substr(start, j);
1378 start = values.size() - 1;
1381 else if ((values[start+j] & 0xc0) != 0x80) {
1382 // This is the first byte of following utf8 char
1383 accents[key] = values.substr(start, j);
1385 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1394 static string getutf8(unsigned uchar)
1397 string ret = string();
1399 if (uchar <= 0x7f) {
1400 c[maxc-1] = uchar & 0x7f;
1403 unsigned char rest = 0x40;
1404 unsigned char first = 0x80;
1406 for (int i = start; i >=0; --i) {
1408 c[i] = first + uchar;
1411 c[i] = 0x80 | (uchar & 0x3f);
1418 for (int i = 0; i < maxc; i++) {
1419 if (c[i] == 0) continue;
1425 static void buildAccentsMap()
1427 accents["imath"] = "ı";
1429 accents["jmath"] = "ȷ";
1430 accents["cdot"] = "·";
1431 accents["guillemotright"] = "»";
1432 accents["guillemotleft"] = "«";
1433 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1434 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1435 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1436 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1437 accents["negmedspace"] = getutf8(0xf0005);
1438 accents["thickspace"] = getutf8(0xf0006);
1439 accents["negthickspace"] = getutf8(0xf0007);
1440 accents["lyx"] = getutf8(0xf0010); // Used logos
1441 accents["LyX"] = getutf8(0xf0010);
1442 accents["tex"] = getutf8(0xf0011);
1443 accents["TeX"] = getutf8(0xf0011);
1444 accents["latex"] = getutf8(0xf0012);
1445 accents["LaTeX"] = getutf8(0xf0012);
1446 accents["latexe"] = getutf8(0xf0013);
1447 accents["LaTeXe"] = getutf8(0xf0013);
1448 accents["lyxarrow"] = getutf8(0xf0020);
1449 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1450 accents["backslash LyX"] = getutf8(0xf0010);
1451 accents["backslash tex"] = getutf8(0xf0011);
1452 accents["backslash TeX"] = getutf8(0xf0011);
1453 accents["backslash latex"] = getutf8(0xf0012);
1454 accents["backslash LaTeX"] = getutf8(0xf0012);
1455 accents["backslash latexe"] = getutf8(0xf0013);
1456 accents["backslash LaTeXe"] = getutf8(0xf0013);
1457 accents["backslash lyxarrow"] = getutf8(0xf0020);
1458 accents["ddot{\\imath}"] = "ï";
1459 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1460 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1461 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1462 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1463 accents["acute{\\imath}"] = "í";
1464 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1465 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1466 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1467 buildaccent("mathring|r", "aAuUwy",
1469 accents["check{\\imath}"] = "ǐ";
1470 accents["check{\\jmath}"] = "ǰ";
1471 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1472 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1473 accents["hat{\\imath}"] = "î";
1474 accents["hat{\\jmath}"] = "ĵ";
1475 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1476 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1477 accents["bar{\\imath}"] = "ī";
1478 buildaccent("bar|=", "aAeEiIoOuUyY",
1479 "āĀēĒīĪōŌūŪȳȲ"); // macron
1480 accents["tilde{\\imath}"] = "ĩ";
1481 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1482 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1483 accents["breve{\\imath}"] = "ĭ";
1484 buildaccent("breve|u", "aAeEgGiIoOuU",
1485 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1486 accents["grave{\\imath}"] = "ì";
1487 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1488 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1489 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1490 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1491 buildaccent("ogonek|k", "AaEeIiUuOo",
1492 "ĄąĘęĮįŲųǪǫ"); // ogonek
1493 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1494 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1495 buildaccent("subring|textsubring", "Aa",
1497 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1498 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1499 buildaccent("subtilde|textsubtilde", "EeIiUu",
1500 "ḚḛḬḭṴṵ"); // subtilde
1501 accents["dgrave{\\imath}"] = "ȉ";
1502 accents["textdoublegrave{\\i}"] = "ȉ";
1503 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1504 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1505 accents["rcap{\\imath}"] = "ȉ";
1506 accents["textroundcap{\\i}"] = "ȉ";
1507 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1508 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1509 buildaccent("slashed", "oO",
1514 * Created accents in math or regexp environment
1515 * are macros, but we need the utf8 equivalent
1517 void Intervall::removeAccents()
1519 if (accents.empty())
1521 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1522 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1523 "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))|guillemot(left|right))(?![a-zA-Z]))");
1525 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1527 string key = sub.str(1);
1528 if (accents.find(key) != accents.end()) {
1529 string val = accents[key];
1530 size_t pos = sub.position(size_t(0));
1531 for (size_t i = 0; i < val.size(); i++) {
1532 par[pos+i] = val[i];
1534 // Remove possibly following space too
1535 if (par[pos+sub.str(0).size()] == ' ')
1536 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1538 addIntervall(pos+val.size(), pos + sub.str(0).size());
1539 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1540 // remove traces of any remaining chars
1545 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1550 void Intervall::handleOpenP(int i)
1553 depts[actualdeptindex] = i+1;
1554 closes[actualdeptindex] = -1;
1555 checkDepthIndex(actualdeptindex);
1558 void Intervall::handleCloseP(int i, bool closingAllowed)
1560 if (actualdeptindex <= 0) {
1561 if (! closingAllowed)
1562 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1563 // if we are at the very end
1564 addIntervall(i, i+1);
1567 closes[actualdeptindex] = i+1;
1572 void Intervall::resetOpenedP(int openPos)
1574 // Used as initializer for foreignlanguage entry
1575 actualdeptindex = 1;
1576 depts[1] = openPos+1;
1580 int Intervall::previousNotIgnored(int start) const
1582 int idx = 0; /* int intervalls */
1583 for (idx = ignoreidx; idx >= 0; --idx) {
1584 if (start > borders[idx].upper)
1586 if (start >= borders[idx].low)
1587 start = borders[idx].low-1;
1592 int Intervall::nextNotIgnored(int start) const
1594 int idx = 0; /* int intervalls */
1595 for (idx = 0; idx <= ignoreidx; idx++) {
1596 if (start < borders[idx].low)
1598 if (start < borders[idx].upper)
1599 start = borders[idx].upper;
1604 typedef map<string, KeyInfo> KeysMap;
1605 typedef vector< KeyInfo> Entries;
1606 static KeysMap keys = map<string, KeyInfo>();
1612 Intervall interval_;
1613 void buildKeys(bool);
1614 void buildEntries(bool);
1615 void makeKey(const string &, KeyInfo, bool isPatternString);
1616 void processRegion(int start, int region_end); /* remove {} parts */
1617 void removeHead(KeyInfo const &, int count=0);
1620 LatexInfo(string const & par, bool isPatternString)
1621 : entidx_(-1), interval_(isPatternString, par)
1623 buildKeys(isPatternString);
1624 entries_ = vector<KeyInfo>();
1625 buildEntries(isPatternString);
1629 if (entries_.empty()) {
1632 if (entries_[0].keytype == KeyInfo::isTitle) {
1633 interval_.hasTitle = true;
1634 if (! entries_[0].disabled) {
1635 interval_.titleValue = entries_[0].head;
1638 interval_.titleValue = "";
1640 removeHead(entries_[0]);
1641 if (entries_.size() > 1)
1650 if (int(entries_.size()) > entidx_) {
1657 bool setNextKey(int idx) {
1658 if ((idx == entidx_) && (entidx_ >= 0)) {
1665 int find(int start, KeyInfo::KeyType keytype) const {
1669 while (tmpIdx < int(entries_.size())) {
1670 if (entries_[tmpIdx].keytype == keytype)
1676 int process(ostringstream & os, KeyInfo const & actual);
1677 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1678 // string show(int lastpos) { return interval.show(lastpos);};
1679 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1680 KeyInfo &getKeyInfo(int keyinfo) {
1681 static KeyInfo invalidInfo = KeyInfo();
1682 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1685 return entries_[keyinfo];
1687 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1688 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1692 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1696 for (int i = start; i < end; i += 1 + skip) {
1700 if (c == '\\') skip = 1;
1704 else if (c == down) {
1707 if ((repeat <= 0) || (par[i+1] != up))
1725 vector<MathEntry> entries_;
1730 void insert(string const & wait, size_t start, size_t end) {
1731 MathEntry m = MathEntry();
1733 m.mathStart = start;
1735 m.mathSize = end - start;
1736 entries_.push_back(m);
1738 bool empty() const { return entries_.empty(); };
1739 size_t getEndPos() const {
1740 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1743 return entries_[actualIdx_].mathEnd;
1745 size_t getStartPos() const {
1746 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1747 return 100000; /* definitely enough? */
1749 return entries_[actualIdx_].mathStart;
1751 size_t getFirstPos() {
1753 return getStartPos();
1755 size_t getSize() const {
1756 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1759 return entries_[actualIdx_].mathSize;
1761 void incrEntry() { actualIdx_++; };
1764 void LatexInfo::buildEntries(bool isPatternString)
1766 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\})");
1767 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1768 static bool disableLanguageOverride = false;
1769 smatch sub, submath;
1770 bool evaluatingRegexp = false;
1772 bool evaluatingMath = false;
1773 bool evaluatingCode = false;
1775 bool evaluatingOptional = false;
1776 size_t optionalEnd = 0;
1779 bool math_end_waiting = false;
1780 size_t math_pos = 10000;
1782 static vector<string> usedText = vector<string>();
1784 interval_.removeAccents();
1786 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1788 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1789 // prefixed by odd count of '\\'
1792 if (math_end_waiting) {
1793 size_t pos = submath.position(size_t(2));
1794 if ((math_end == "$") &&
1795 (submath.str(2) == "$")) {
1796 mi.insert("$", math_pos, pos + 1);
1797 math_end_waiting = false;
1799 else if ((math_end == "\\]") &&
1800 (submath.str(2) == "\\]")) {
1801 mi.insert("\\]", math_pos, pos + 2);
1802 math_end_waiting = false;
1804 else if ((submath.str(3).compare("end") == 0) &&
1805 (submath.str(4).compare(math_end) == 0)) {
1806 mi.insert(math_end, math_pos, pos + submath.str(2).length());
1807 math_end_waiting = false;
1813 if (submath.str(3).compare("begin") == 0) {
1814 math_end_waiting = true;
1815 math_end = submath.str(4);
1816 math_pos = submath.position(size_t(2));
1818 else if (submath.str(2).compare("\\[") == 0) {
1819 math_end_waiting = true;
1821 math_pos = submath.position(size_t(2));
1823 else if (submath.str(2) == "$") {
1824 size_t pos = submath.position(size_t(2));
1825 math_end_waiting = true;
1831 // Ignore language if there is math somewhere in pattern-string
1832 if (isPatternString) {
1833 for (auto s: usedText) {
1834 // Remove entries created in previous search runs
1837 usedText = vector<string>();
1840 keys["foreignlanguage"].disabled = true;
1841 disableLanguageOverride = true;
1844 disableLanguageOverride = false;
1847 if (disableLanguageOverride) {
1848 keys["foreignlanguage"].disabled = true;
1851 math_pos = mi.getFirstPos();
1852 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1854 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1855 // prefixed by odd count of '\\'
1858 string key = sub.str(5);
1860 if (sub.str(2)[0] == '\\')
1861 key = sub.str(2)[1];
1866 if (keys.find(key) != keys.end()) {
1867 if (keys[key].keytype == KeyInfo::headRemove) {
1868 KeyInfo found1 = keys[key];
1869 found1.disabled = true;
1870 found1.head = "\\" + key + "{";
1871 found1._tokenstart = sub.position(size_t(2));
1872 found1._tokensize = found1.head.length();
1873 found1._dataStart = found1._tokenstart + found1.head.length();
1874 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1875 found1._dataEnd = endpos;
1880 if (evaluatingRegexp) {
1881 if (sub.str(3).compare("endregexp") == 0) {
1882 evaluatingRegexp = false;
1883 // found._tokenstart already set
1884 found._dataEnd = sub.position(size_t(2)) + 13;
1885 found._dataStart = found._dataEnd;
1886 found._tokensize = found._dataEnd - found._tokenstart;
1887 found.parenthesiscount = 0;
1888 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1895 if (evaluatingMath) {
1896 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1898 evaluatingMath = false;
1900 math_pos = mi.getStartPos();
1902 if (keys.find(key) == keys.end()) {
1903 found = KeyInfo(KeyInfo::isStandard, 0, true);
1904 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1905 found = KeyInfo(KeyInfo::isText, 0, false);
1906 if (isPatternString) {
1907 found.keytype = KeyInfo::isChar;
1908 found.disabled = false;
1912 usedText.push_back(key);
1916 if (key.compare("regexp") == 0) {
1917 evaluatingRegexp = true;
1918 found._tokenstart = sub.position(size_t(2));
1919 found._tokensize = 0;
1923 // Handle the other params of key
1924 if (found.keytype == KeyInfo::isIgnored)
1926 else if (found.keytype == KeyInfo::isMath) {
1927 if (size_t(sub.position(size_t(2))) == math_pos) {
1929 found._tokenstart = sub.position(size_t(2));
1930 found._tokensize = mi.getSize();
1931 found._dataEnd = found._tokenstart + found._tokensize;
1932 found._dataStart = found._dataEnd;
1933 found.parenthesiscount = 0;
1934 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1935 evaluatingMath = true;
1938 // begin|end of unknown env, discard
1939 // First handle tables
1940 // longtable|tabular
1941 bool discardComment;
1943 found.keytype = KeyInfo::doRemove;
1944 if ((sub.str(7).compare("longtable") == 0) ||
1945 (sub.str(7).compare("tabular") == 0)) {
1946 discardComment = true; /* '%' */
1949 discardComment = false;
1950 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1952 string token = sub.str(7);
1953 if (regex_match(token, sub2, removeArgs)) {
1954 found.keytype = KeyInfo::removeWithArg;
1957 // discard spaces before pos(2)
1958 int pos = sub.position(size_t(2));
1960 for (count = 0; pos - count > 0; count++) {
1961 char c = interval_.par[pos-count-1];
1962 if (discardComment) {
1963 if ((c != ' ') && (c != '%'))
1969 found._tokenstart = pos - count;
1970 if (sub.str(3).compare(0, 5, "begin") == 0) {
1971 size_t pos1 = pos + sub.str(2).length();
1972 if (sub.str(7).compare("cjk") == 0) {
1973 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1974 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1976 found.keytype = KeyInfo::isMain;
1977 found._dataStart = pos1;
1978 found._dataEnd = interval_.par.length();
1979 found.disabled = keys["foreignlanguage"].disabled;
1980 found.used = keys["foreignlanguage"].used;
1981 found._tokensize = pos1 - found._tokenstart;
1982 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1985 // Swallow possible optional params
1986 while (interval_.par[pos1] == '[') {
1987 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1989 // Swallow also the eventual parameter
1990 if (interval_.par[pos1] == '{') {
1991 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1994 found._dataEnd = pos1;
1996 found._dataStart = found._dataEnd;
1997 found._tokensize = count + found._dataEnd - pos;
1998 found.parenthesiscount = 0;
1999 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2000 found.disabled = true;
2004 // Handle "\end{...}"
2005 found._dataStart = pos + sub.str(2).length();
2006 found._dataEnd = found._dataStart;
2007 found._tokensize = count + found._dataEnd - pos;
2008 found.parenthesiscount = 0;
2009 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2010 found.disabled = true;
2014 else if (found.keytype != KeyInfo::isRegex) {
2015 found._tokenstart = sub.position(size_t(2));
2016 if (found.parenthesiscount == 0) {
2017 // Probably to be discarded
2018 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2019 char following = interval_.par[following_pos];
2020 if (following == ' ')
2021 found.head = "\\" + sub.str(5) + " ";
2022 else if (following == '=') {
2023 // like \uldepth=1000pt
2024 found.head = sub.str(2);
2027 found.head = "\\" + key;
2028 found._tokensize = found.head.length();
2029 found._dataEnd = found._tokenstart + found._tokensize;
2030 found._dataStart = found._dataEnd;
2033 int params = found._tokenstart + key.length() + 1;
2034 if (evaluatingOptional) {
2035 if (size_t(found._tokenstart) > optionalEnd) {
2036 evaluatingOptional = false;
2039 found.disabled = true;
2042 int optend = params;
2043 while (interval_.par[optend] == '[') {
2044 // discard optional parameters
2045 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2047 if (optend > params) {
2048 key += interval_.par.substr(params, optend-params);
2049 evaluatingOptional = true;
2050 optionalEnd = optend;
2052 string token = sub.str(7);
2054 if (interval_.par[optend] != '{') {
2056 found.parenthesiscount = 0;
2057 found.head = "\\" + key;
2060 closings = found.parenthesiscount;
2061 if (found.parenthesiscount == 1) {
2062 found.head = "\\" + key + "{";
2064 else if (found.parenthesiscount > 1) {
2066 found.head = sub.str(2) + "{";
2067 closings = found.parenthesiscount - 1;
2070 found.head = "\\" + key + "{";
2073 found._tokensize = found.head.length();
2074 found._dataStart = found._tokenstart + found.head.length();
2075 if (found.keytype == KeyInfo::doRemove) {
2077 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2078 if (endpar >= interval_.par.length())
2079 found._dataStart = interval_.par.length();
2081 found._dataStart = endpar;
2082 found._tokensize = found._dataStart - found._tokenstart;
2085 found._dataStart = found._tokenstart + found._tokensize;
2089 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
2090 found._dataStart += 15;
2094 endpos = found._dataStart - 1;
2096 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2097 if (found.keytype == KeyInfo::isList) {
2098 // Check if it really is list env
2099 static regex const listre("^([a-z]+)$");
2101 if (!regex_match(token, sub2, listre)) {
2102 // Change the key of this entry. It is not in a list/item environment
2103 found.keytype = KeyInfo::endArguments;
2106 if (found.keytype == KeyInfo::noMain) {
2107 evaluatingCode = true;
2109 codeStart = found._dataStart;
2111 else if (evaluatingCode) {
2112 if (size_t(found._dataStart) > codeEnd)
2113 evaluatingCode = false;
2114 else if (found.keytype == KeyInfo::isMain) {
2115 // Disable this key, treate it as standard
2116 found.keytype = KeyInfo::isStandard;
2117 found.disabled = true;
2118 if ((codeEnd +1 >= interval_.par.length()) &&
2119 (found._tokenstart == codeStart)) {
2120 // trickery, because the code inset starts
2121 // with \selectlanguage ...
2123 if (entries_.size() > 1) {
2124 entries_[entries_.size()-1]._dataEnd = codeEnd;
2129 if ((endpos == interval_.par.length()) &&
2130 (found.keytype == KeyInfo::doRemove)) {
2131 // Missing closing => error in latex-input?
2132 // therefore do not delete remaining data
2133 found._dataStart -= 1;
2134 found._dataEnd = found._dataStart;
2137 found._dataEnd = endpos;
2139 if (isPatternString) {
2140 keys[key].used = true;
2143 entries_.push_back(found);
2147 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2149 stringstream s(keysstring);
2151 const char delim = '|';
2152 while (getline(s, key, delim)) {
2153 KeyInfo keyII(keyI);
2154 if (isPatternString) {
2157 else if ( !keys[key].used)
2158 keyII.disabled = true;
2163 void LatexInfo::buildKeys(bool isPatternString)
2166 static bool keysBuilt = false;
2167 if (keysBuilt && !isPatternString) return;
2169 // Keys to ignore in any case
2170 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2171 // Known standard keys with 1 parameter.
2172 // Split is done, if not at start of region
2173 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2174 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2175 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2176 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2177 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2178 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2180 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2181 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2182 makeKey("section*|subsection*|subsubsection*|paragraph*",
2183 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2184 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2185 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2187 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2189 // Split is done, if not at start of region
2190 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2191 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2193 // Split is done always.
2194 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2198 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2199 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2200 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2201 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2203 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2204 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2205 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2206 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2208 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2209 // Custom space/skip, remove the content (== length value)
2210 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2211 // Found in fr/UserGuide.lyx
2212 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2214 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2215 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2216 // Known macros to remove (including their parameter)
2218 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2219 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2220 // handle like standard keys with 1 parameter.
2221 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2223 // Ignore deleted text
2224 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2225 // but preserve added text
2226 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2228 // Macros to remove, but let the parameter survive
2230 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2232 // Remove language spec from content of these insets
2233 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2235 // Same effect as previous, parameter will survive (because there is no one anyway)
2237 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2238 // Remove table decorations
2239 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2240 // Discard shape-header.
2241 // For footnote or shortcut too, because of lang settings
2242 // and wrong handling if used 'KeyInfo::noMain'
2243 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2244 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2245 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2246 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2247 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2248 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2249 // like ('tiny{}' or '\tiny ' ... )
2250 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2252 // Survives, like known character
2253 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2254 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2255 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2257 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2258 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2259 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2261 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2262 // Remove RTL/LTR marker
2263 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2264 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2265 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2266 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2267 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2268 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2269 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2270 if (isPatternString) {
2271 // Allow the first searched string to rebuild the keys too
2275 // no need to rebuild again
2281 * Keep the list of actual opened parentheses actual
2282 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2284 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2287 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2291 if (c == '\\') skip = 1;
2292 else if (c == '{') {
2295 else if (c == '}') {
2296 handleCloseP(i, closingAllowed);
2302 string Intervall::show(int lastpos)
2304 int idx = 0; /* int intervalls */
2307 for (idx = 0; idx <= ignoreidx; idx++) {
2308 while (i < lastpos) {
2310 if (i <= borders[idx].low) {
2311 if (borders[idx].low > lastpos)
2312 printsize = lastpos - i;
2314 printsize = borders[idx].low - i;
2315 s += par.substr(i, printsize);
2317 if (i >= borders[idx].low)
2318 i = borders[idx].upper;
2321 i = borders[idx].upper;
2327 s += par.substr(i, lastpos-i);
2333 void Intervall::output(ostringstream &os, int lastpos)
2335 // get number of chars to output
2336 int idx = 0; /* int intervalls */
2339 string startTitle = titleValue;
2340 for (idx = 0; idx <= ignoreidx; idx++) {
2342 if (i <= borders[idx].low) {
2344 if (borders[idx].low > lastpos)
2345 printsize = lastpos - i;
2347 printsize = borders[idx].low - i;
2348 if (printsize > 0) {
2349 os << startTitle << par.substr(i, printsize);
2351 printed += printsize;
2354 handleParentheses(i, false);
2355 if (i >= borders[idx].low)
2356 i = borders[idx].upper;
2359 i = borders[idx].upper;
2366 os << startTitle << par.substr(i, lastpos-i);
2367 printed += lastpos-i;
2369 handleParentheses(lastpos, false);
2371 if (keys["foreignlanguage"].disabled)
2372 startindex = actualdeptindex-langcount;
2374 startindex = actualdeptindex;
2375 for (int i = startindex; i > 0; --i) {
2378 if (hasTitle && (printed > 0))
2380 if (! isPatternString_)
2382 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2385 void LatexInfo::processRegion(int start, int region_end)
2387 while (start < region_end) { /* Let {[} and {]} survive */
2388 int cnt = interval_.isOpeningPar(start);
2390 // Closing is allowed past the region
2391 int closing = interval_.findclosing(start+1, interval_.par.length());
2392 interval_.addIntervall(start, start+1);
2393 interval_.addIntervall(closing, closing+1);
2397 start = interval_.nextNotIgnored(start+1);
2401 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2403 if (actual.parenthesiscount == 0) {
2404 // "{\tiny{} ...}" ==> "{{} ...}"
2405 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2408 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2409 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2410 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2414 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2417 switch (actual.keytype)
2419 case KeyInfo::isTitle: {
2421 nextKeyIdx = getNextKey();
2424 case KeyInfo::cleanToStart: {
2425 actual._dataEnd = actual._dataStart;
2426 nextKeyIdx = getNextKey();
2427 // Search for end of arguments
2428 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2430 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2431 entries_[i].disabled = true;
2433 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2435 while (interval_.par[actual._dataEnd] == ' ')
2437 interval_.addIntervall(0, actual._dataEnd+1);
2438 interval_.actualdeptindex = 0;
2439 interval_.depts[0] = actual._dataEnd+1;
2440 interval_.closes[0] = -1;
2443 case KeyInfo::isText:
2444 interval_.par[actual._tokenstart] = '#';
2445 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2446 nextKeyIdx = getNextKey();
2448 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2449 if (actual.disabled)
2450 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2452 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2455 case KeyInfo::isChar: {
2456 nextKeyIdx = getNextKey();
2459 case KeyInfo::isSize: {
2460 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2461 if (actual.parenthesiscount == 0)
2462 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2464 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2466 nextKeyIdx = getNextKey();
2468 // Here _dataStart points to '{', so correct it
2469 actual._dataStart += 1;
2470 actual._tokensize += 1;
2471 actual.parenthesiscount = 1;
2472 if (interval_.par[actual._dataStart] == '}') {
2473 // Determine the end if used like '{\tiny{}...}'
2474 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2475 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2478 // Determine the end if used like '\tiny{...}'
2479 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2481 // Split on this key if not at start
2482 int start = interval_.nextNotIgnored(previousStart);
2483 if (start < actual._tokenstart) {
2484 interval_.output(os, actual._tokenstart);
2485 interval_.addIntervall(start, actual._tokenstart);
2487 // discard entry if at end of actual
2488 nextKeyIdx = process(os, actual);
2492 case KeyInfo::endArguments: {
2493 // Remove trailing '{}' too
2494 actual._dataStart += 1;
2495 actual._dataEnd += 1;
2496 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2497 nextKeyIdx = getNextKey();
2500 case KeyInfo::noMain:
2502 case KeyInfo::isStandard: {
2503 if (actual.disabled) {
2505 processRegion(actual._dataStart, actual._dataStart+1);
2506 nextKeyIdx = getNextKey();
2508 // Split on this key if not at datastart of calling entry
2509 int start = interval_.nextNotIgnored(previousStart);
2510 if (start < actual._tokenstart) {
2511 interval_.output(os, actual._tokenstart);
2512 interval_.addIntervall(start, actual._tokenstart);
2514 // discard entry if at end of actual
2515 nextKeyIdx = process(os, actual);
2519 case KeyInfo::removeWithArg: {
2520 nextKeyIdx = getNextKey();
2521 // Search for end of arguments
2522 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2524 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2525 entries_[i].disabled = true;
2527 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2529 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2532 case KeyInfo::doRemove: {
2533 // Remove the key with all parameters and following spaces
2535 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2536 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2539 // Remove also enclosing parentheses [] and {}
2542 while (actual._tokenstart > numpars) {
2543 if (pos+numpars >= interval_.par.size())
2545 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2547 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2553 if (interval_.par[pos+numpars] == ' ')
2557 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2558 nextKeyIdx = getNextKey();
2561 case KeyInfo::isList: {
2562 // Discard space before _tokenstart
2564 for (count = 0; count < actual._tokenstart; count++) {
2565 if (interval_.par[actual._tokenstart-count-1] != ' ')
2568 nextKeyIdx = getNextKey();
2569 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2571 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2573 // How else can we catch this one?
2574 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2575 entries_[i].disabled = true;
2577 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2579 else if (nextKeyIdx > 0) {
2580 // Ignore any lang entries inside data region
2581 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2582 if (entries_[i].keytype == KeyInfo::isMain)
2583 entries_[i].disabled = true;
2586 if (actual.disabled) {
2587 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2590 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2592 if (interval_.par[actual._dataEnd+1] == '[') {
2593 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2594 if ((interval_.par[actual._dataEnd+2] == '{') &&
2595 (interval_.par[posdown-1] == '}')) {
2596 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2597 interval_.addIntervall(posdown-1, posdown+1);
2600 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2601 interval_.addIntervall(posdown, posdown+1);
2603 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2604 if (blk > posdown) {
2605 // Discard at most 1 space after empty item
2607 for (count = 0; count < 1; count++) {
2608 if (interval_.par[blk+count] != ' ')
2612 interval_.addIntervall(blk, blk+count);
2617 case KeyInfo::isSectioning: {
2618 // Discard spaces before _tokenstart
2620 int val = actual._tokenstart;
2621 for (count = 0; count < actual._tokenstart;) {
2622 val = interval_.previousNotIgnored(val-1);
2623 if (val < 0 || interval_.par[val] != ' ')
2626 count = actual._tokenstart - val;
2629 if (actual.disabled) {
2630 removeHead(actual, count);
2631 nextKeyIdx = getNextKey();
2633 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2634 nextKeyIdx = process(os, actual);
2638 case KeyInfo::isMath: {
2639 // Same as regex, use the content unchanged
2640 nextKeyIdx = getNextKey();
2643 case KeyInfo::isRegex: {
2644 // DO NOT SPLIT ON REGEX
2646 nextKeyIdx = getNextKey();
2649 case KeyInfo::isIgnored: {
2650 // Treat like a character for now
2651 nextKeyIdx = getNextKey();
2654 case KeyInfo::isMain: {
2655 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2656 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2657 if (actual._tokenstart > 0) {
2658 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2659 if ((prev >= 0) && interval_.par[prev] == '%')
2660 interval_.addIntervall(prev, prev+1);
2662 if (actual.disabled) {
2664 interval_.langcount++;
2665 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2666 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2667 // Discard also the space before math-equation
2668 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2670 nextKeyIdx = getNextKey();
2671 // interval.resetOpenedP(actual._dataStart-1);
2674 if (actual._tokenstart < 26) {
2675 // for the first (and maybe dummy) language
2676 interval_.setForDefaultLang(actual);
2678 interval_.resetOpenedP(actual._dataStart-1);
2682 case KeyInfo::invalid:
2683 case KeyInfo::headRemove:
2684 // These two cases cannot happen, already handled
2687 // LYXERR(Debug::INFO, "Unhandled keytype");
2688 nextKeyIdx = getNextKey();
2695 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2697 int end = interval_.nextNotIgnored(actual._dataEnd);
2698 int oldStart = actual._dataStart;
2699 int nextKeyIdx = getNextKey();
2701 if ((nextKeyIdx < 0) ||
2702 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2703 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2704 if (oldStart <= end) {
2705 processRegion(oldStart, end);
2710 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2712 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2713 (void) dispatch(os, actual._dataStart, nextKey);
2714 end = nextKey._tokenstart;
2717 processRegion(oldStart, nextKey._tokenstart);
2718 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2720 oldStart = nextKey._dataEnd+1;
2722 // now nextKey is either invalid or is outside of actual._dataEnd
2723 // output the remaining and discard myself
2724 if (oldStart <= end) {
2725 processRegion(oldStart, end);
2727 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2729 // This is the normal case.
2730 // But if using the firstlanguage, the closing may be missing
2732 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2734 if (actual._dataEnd < end)
2735 output_end = interval_.nextNotIgnored(actual._dataEnd);
2736 else if (interval_.par.size() > (size_t) end)
2737 output_end = interval_.nextNotIgnored(end);
2739 output_end = interval_.par.size();
2740 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2741 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2743 // Remove possible empty data
2744 int dstart = interval_.nextNotIgnored(actual._dataStart);
2745 while (interval_.isOpeningPar(dstart) == 1) {
2746 interval_.addIntervall(dstart, dstart+1);
2747 int dend = interval_.findclosing(dstart+1, output_end);
2748 interval_.addIntervall(dend, dend+1);
2749 dstart = interval_.nextNotIgnored(dstart+1);
2751 if (dstart < output_end)
2752 interval_.output(os, output_end);
2753 interval_.addIntervall(actual._tokenstart, end);
2757 string splitOnKnownMacros(string par, bool isPatternString)
2760 LatexInfo li(par, isPatternString);
2761 // LYXERR(Debug::INFO, "Berfore split: " << par);
2762 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2764 DummyKey._tokensize = 0;
2765 DummyKey._dataStart = 0;
2766 DummyKey._dataEnd = par.length();
2767 DummyKey.disabled = true;
2768 int firstkeyIdx = li.getFirstKey();
2770 if (firstkeyIdx >= 0) {
2771 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2772 DummyKey._tokenstart = firstKey._tokenstart;
2774 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2775 // Use dummy firstKey
2776 firstKey = DummyKey;
2777 (void) li.setNextKey(firstkeyIdx);
2780 if (par.substr(firstKey._dataStart, 2) == "% ")
2781 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2783 nextkeyIdx = li.process(os, firstKey);
2784 while (nextkeyIdx >= 0) {
2785 // Check for a possible gap between the last
2786 // entry and this one
2787 int datastart = li.nextNotIgnored(firstKey._dataStart);
2788 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2789 if ((nextKey._tokenstart > datastart)) {
2791 firstKey._dataStart = datastart;
2792 firstKey._dataEnd = par.length();
2793 (void) li.setNextKey(nextkeyIdx);
2794 // Fake the last opened parenthesis
2795 li.setForDefaultLang(firstKey);
2796 nextkeyIdx = li.process(os, firstKey);
2799 if (nextKey.keytype != KeyInfo::isMain) {
2800 firstKey._dataStart = datastart;
2801 firstKey._dataEnd = nextKey._dataEnd+1;
2802 (void) li.setNextKey(nextkeyIdx);
2803 li.setForDefaultLang(firstKey);
2804 nextkeyIdx = li.process(os, firstKey);
2807 nextkeyIdx = li.process(os, nextKey);
2811 // Handle the remaining
2812 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2813 firstKey._dataEnd = par.length();
2815 if ((firstKey._dataStart < firstKey._dataEnd) &&
2816 (par[firstKey._dataStart] != '}')) {
2817 li.setForDefaultLang(firstKey);
2818 (void) li.process(os, firstKey);
2822 // return string definitelly impossible to match
2823 s = "\\foreignlanguage{ignore}{ }";
2827 s = par; /* no known macros found */
2828 // LYXERR(Debug::INFO, "After split: " << s);
2833 * Try to unify the language specs in the latexified text.
2834 * Resulting modified string is set to "", if
2835 * the searched tex does not contain all the features in the search pattern
2837 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2839 static Features regex_f;
2840 static int missed = 0;
2841 static bool regex_with_format = false;
2843 int parlen = par.length();
2845 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2848 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2849 // Happens to be there in case of description or labeling environment
2854 // Split the latex input into pieces which
2855 // can be digested by our search engine
2856 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2857 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2858 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2861 result = par.substr(0, parlen);
2862 if (isPatternString) {
2865 regex_f = identifyFeatures(result);
2866 string features = "";
2867 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2868 string a = it->first;
2869 regex_with_format = true;
2870 features += " " + a;
2871 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2873 LYXERR(Debug::FIND, "Identified Features" << features);
2876 } else if (regex_with_format) {
2877 Features info = identifyFeatures(result);
2878 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2879 string a = it->first;
2880 bool b = it->second;
2881 if (b && ! info[a]) {
2883 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2889 // LYXERR(Debug::INFO, "No regex formats");
2895 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2896 static int identifyClosing(string & t)
2898 int open_braces = 0;
2900 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2901 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2903 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2905 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2907 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2916 static int num_replaced = 0;
2917 static bool previous_single_replace = true;
2919 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
2922 // Handle \w properly
2923 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
2924 if (! opt.casesensitive) {
2925 popts |= QRegularExpression::CaseInsensitiveOption;
2927 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
2928 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
2930 if (regexp.isValid() && regexp2.isValid()) {
2931 regexIsValid = true;
2932 // Check '{', '}' pairs inside the regex
2935 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
2936 char c = par_as_string[i];
2943 else if (c == '}') {
2950 if (balanced != 0) {
2951 regexIsValid = false;
2952 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
2956 regexIsValid = false;
2957 if (!regexp.isValid())
2958 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
2960 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
2963 if (opt.casesensitive) {
2964 regexp = regex(regexp_str);
2965 regexp2 = regex(regexp2_str);
2968 regexp = regex(regexp_str, std::regex_constants::icase);
2969 regexp2 = regex(regexp2_str, std::regex_constants::icase);
2974 static void modifyRegexForMatchWord(string &t)
2977 regex wordre("(\\\\)*((\\.|\\\\b))");
2980 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
2982 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
2985 else if (sub.str(2) == "\\\\b")
2987 if (lastpos < (size_t) sub.position(2))
2988 s += t.substr(lastpos, sub.position(2) - lastpos);
2990 lastpos = sub.position(2) + sub.length(2);
2994 else if (lastpos < t.length())
2995 s += t.substr(lastpos, t.length() - lastpos);
2996 t = "\\b" + s + "\\b";
2999 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3000 : p_buf(&buf), p_first_buf(&buf), opt(opt)
3002 static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\\])" };
3003 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3004 docstring const & ds = stringifySearchBuffer(find_buf, opt);
3005 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3006 if (opt.replace_all && previous_single_replace) {
3007 previous_single_replace = false;
3010 else if (!opt.replace_all) {
3011 num_replaced = 0; // count number of replaced strings
3012 previous_single_replace = true;
3014 // When using regexp, braces are hacked already by escape_for_regex()
3015 par_as_string = normalize(ds);
3017 close_wildcards = 0;
3019 size_t lead_size = 0;
3020 // correct the language settings
3021 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
3023 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
3024 if (opt.ignoreformat) {
3028 lead_size = identifyLeading(par_as_string);
3030 lead_as_string = par_as_string.substr(0, lead_size);
3031 string lead_as_regex_string = std::regex_replace(lead_as_string, specialChars, R"(\$&)" );
3032 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3033 string par_as_regex_string_nolead = std::regex_replace(par_as_string_nolead, specialChars, R"(\$&)" );
3034 /* Handle whole words too in this case
3036 if (opt.matchword) {
3037 par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3038 opt.matchword = false;
3040 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3041 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3042 CreateRegexp(opt, regexp_str, regexp2_str);
3047 if (!opt.ignoreformat) {
3048 lead_size = identifyLeading(par_as_string);
3049 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
3050 lead_as_string = par_as_string.substr(0, lead_size);
3051 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3054 // Here we are using regexp
3055 LASSERT(use_regexp, /**/);
3057 string lead_as_regexp;
3058 if (lead_size > 0) {
3059 lead_as_regexp = std::regex_replace(par_as_string.substr(0, lead_size), specialChars, R"(\$&)" );
3060 par_as_string = par_as_string_nolead;
3061 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
3062 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3064 LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3065 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3066 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3067 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3069 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
3070 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
3071 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
3072 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
3073 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
3074 || regex_replace(par_as_string, par_as_string,
3075 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
3076 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
3077 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
3081 if (!opt.ignoreformat) {
3082 // Remove extra '\}' at end if not part of \{\.\}
3083 size_t lng = par_as_string.size();
3085 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3087 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3096 if (lng < par_as_string.size())
3097 par_as_string = par_as_string.substr(0,lng);
3098 if ((lng > 0) && (par_as_string[0] == '^')) {
3099 par_as_string = par_as_string.substr(1);
3101 opt.matchstart = true;
3104 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3105 LYXERR(Debug::FIND, "Open braces: " << open_braces);
3106 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
3107 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3109 // If entered regexp must match at begin of searched string buffer
3110 // Kornel: Added parentheses to use $1 for size of the leading string
3114 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3115 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3116 // so the convert has no effect in that case
3117 for (int i = 7; i > 0; --i) {
3118 string orig = "\\\\" + std::to_string(i);
3119 string dest = "\\" + std::to_string(i+2);
3120 while (regex_replace(par_as_string, par_as_string, orig, dest));
3122 /* opt.matchword is ignored if using regex
3123 so expanding par_as_string with "\\b" seems appropriate here
3124 if regex contains for instance '.*' or '.+'
3125 1.) Nothing to do, if 'par_as_string' contains "\\b" already.
3126 (Means, that the user knows how to handle whole words
3127 2.) else replace '.' with "\\S" and wrap the regex with "\\b"
3129 if (opt.matchword) {
3130 modifyRegexForMatchWord(par_as_string);
3131 opt.matchword = false;
3133 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3134 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3136 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3137 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3138 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3143 // Count number of characters in string
3147 // \\[a-zA-Z]+ ==> 1
3149 static int computeSize(QStringRef s, int len)
3150 #define isLyxAlpha(arg) arg.isLetter()
3152 static int computeSize(string s, int len)
3153 #define isLyxAlpha(arg) isalpha(arg)
3160 for (int i = 0; i < len; i += skip, count++) {
3161 if (s.at(i) == '\\') {
3163 if (i + 1 < len && isLyxAlpha(s.at(i+1))) {
3164 for (int j = 2; i+j < len; j++) {
3165 if (! isLyxAlpha(s.at(i+j))) {
3166 if (s.at(i+j) == ' ')
3168 else if (s.at(i+j) == '{') {
3169 if (i+j+1 < len && s.at(i+j+1) == '}')
3171 else if (i + j + 1 >= len)
3180 else if (s.at(i) == '{') {
3181 if (i + 1 < len && s.at(i+1) == '}')
3186 else if (s.at(i) == '-') {
3187 if (i+1 < len && s.at(i+1) == '-') {
3188 if (i + 2 < len && s.at(i+2) == '-')
3204 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3209 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3212 docstring docstr = stringifyFromForSearch(opt, cur, len);
3214 if (use_regexp || opt.casesensitive)
3215 str = normalize(docstr);
3217 str = normalize(lowercase(docstr));
3218 if (!opt.ignoreformat) {
3219 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3222 mres.match_len = -1;
3225 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
3226 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
3229 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3231 QString qstr = QString::fromStdString(str);
3232 QRegularExpression const *p_regexp;
3233 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3237 p_regexp = ®exp2;
3239 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3240 if (!match.hasMatch())
3243 regex const *p_regexp;
3244 regex_constants::match_flag_type flags;
3246 flags = regex_constants::match_continuous;
3249 flags = regex_constants::match_default;
3250 p_regexp = ®exp2;
3252 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3253 if (re_it == sregex_iterator())
3255 match_results<string::const_iterator> const & m = *re_it;
3257 // Whole found string, including the leading
3258 // std: m[0].second - m[0].first
3259 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3261 // Size of the leading string
3262 // std: m[1].second - m[1].first
3263 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3264 int leadingsize = 0;
3266 if (match.lastCapturedIndex() > 0) {
3267 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3272 leadingsize = m[1].second - m[1].first;
3276 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3277 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3278 // because of different number of closing at end of string
3279 // we have to 'unify' the length of the post-match.
3280 // Done by ignoring closing parenthesis and linefeeds at string end
3281 int matchend = match.capturedEnd(0);
3282 while (mres.match_len > 0) {
3283 QChar c = qstr.at(matchend - 1);
3284 if ((c == '\n') || (c == '}') || (c == '{')) {
3291 size_t strsize = qstr.size();
3292 while (strsize > (size_t) match.capturedEnd(0)) {
3293 QChar c = qstr.at(strsize-1);
3294 if ((c == '\n') || (c == '}')) {
3300 // LYXERR0(qstr.toStdString());
3301 mres.match2end = strsize - matchend;
3302 mres.pos = match.capturedStart(2);
3304 mres.match_prefix = m[2].second - m[2].first;
3305 mres.match_len = m[0].second - m[2].second;
3306 // ignore closing parenthesis and linefeeds at string end
3307 size_t strend = m[0].second - m[0].first;
3308 int matchend = strend;
3309 while (mres.match_len > 0) {
3310 char c = str.at(matchend - 1);
3311 if ((c == '\n') || (c == '}') || (c == '{')) {
3318 size_t strsize = str.size();
3319 while (strsize > strend) {
3320 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3327 mres.match2end = strsize - matchend;
3328 mres.pos = m[2].first - m[0].first;;
3330 if (mres.match2end < 0)
3332 mres.leadsize = leadingsize;
3334 if (mres.match_len > 0) {
3335 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3336 mres.result.push_back(a0);
3337 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3338 mres.result.push_back(match.captured(i).toStdString());
3342 if (mres.match_len > 0) {
3343 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3344 mres.result.push_back(a0);
3345 for (size_t i = 3; i < m.size(); i++) {
3346 mres.result.push_back(m[i]);
3353 // else !use_regexp: but all code paths above return
3354 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
3355 << par_as_string << "', str='" << str << "'");
3356 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
3357 << lead_as_string << "', par_as_string_nolead='"
3358 << par_as_string_nolead << "'");
3361 LYXERR(Debug::FIND, "size=" << par_as_string.size()
3362 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
3363 if (str.substr(0, par_as_string.size()) == par_as_string) {
3364 mres.match_len = par_as_string.size();
3365 mres.match2end = str.size();
3370 // Start the search _after_ the leading part
3371 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3372 if (pos != string::npos) {
3373 mres.match_len = par_as_string.size();
3374 mres.match2end = str.size() - pos;
3383 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3385 MatchResult mres = findAux(cur, len, at_begin);
3386 int res = mres.match_len;
3388 "res=" << res << ", at_begin=" << at_begin
3389 << ", matchstart=" << opt.matchstart
3390 << ", inTexted=" << cur.inTexted());
3391 if (opt.matchstart) {
3394 else if (mres.match_prefix > 0)
3400 /* DEAD CODE follows
3401 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3403 if ((len > 0) && (res < len)) {
3407 Paragraph const & par = cur.paragraph();
3408 bool ws_left = (cur.pos() > 0)
3409 ? par.isWordSeparator(cur.pos() - 1)
3415 ws_right = (cur.pos() + len < par.size())
3416 ? par.isWordSeparator(cur.pos() + len)
3420 "cur.pos()=" << cur.pos() << ", res=" << res
3421 << ", separ: " << ws_left << ", " << ws_right
3424 if (ws_left && ws_right) {
3425 // Check for word separators inside the found 'word'
3426 for (int i = 0; i < len; i++) {
3427 if (par.isWordSeparator(cur.pos() + i)) {
3440 static bool simple_replace(string &t, string from, string to)
3442 regex repl("(\\\\)*(" + from + ")");
3446 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3448 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3450 if (lastpos < (size_t) sub.position(2))
3451 s += t.substr(lastpos, sub.position(2) - lastpos);
3453 lastpos = sub.position(2) + sub.length(2);
3457 else if (lastpos < t.length())
3458 s += t.substr(lastpos, t.length() - lastpos);
3464 string MatchStringAdv::normalize(docstring const & s) const
3467 t = lyx::to_utf8(s);
3468 // Remove \n at begin
3469 while (!t.empty() && t[0] == '\n')
3472 while (!t.empty() && t[t.size() - 1] == '\n')
3473 t = t.substr(0, t.size() - 1);
3475 // Handle all other '\n'
3476 while ((pos = t.find("\n")) != string::npos) {
3477 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3479 if (isAlnumASCII(t[pos+1])) {
3480 t.replace(pos-2, 3, " ");
3483 t.replace(pos-2, 3, "");
3486 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3487 // '\n' adjacent to non-alpha-numerics, discard
3488 t.replace(pos, 1, "");
3491 // Replace all other \n with spaces
3492 t.replace(pos, 1, " ");
3495 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3496 // Kornel: Added textsl, textsf, textit, texttt and noun
3497 // + allow to seach for colored text too
3498 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3499 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3500 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3501 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3502 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3503 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3509 docstring stringifyFromCursor(DocIterator const & cur, int len)
3511 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3512 if (cur.inTexted()) {
3513 Paragraph const & par = cur.paragraph();
3514 // TODO what about searching beyond/across paragraph breaks ?
3515 // TODO Try adding a AS_STR_INSERTS as last arg
3516 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3517 int(par.size()) : cur.pos() + len;
3518 // OutputParams runparams(&cur.buffer()->params().encoding());
3519 OutputParams runparams(encodings.fromLyXName("utf8"));
3520 runparams.nice = true;
3521 runparams.flavor = Flavor::XeTeX;
3522 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3523 // No side effect of file copying and image conversion
3524 runparams.dryrun = true;
3525 runparams.for_search = true;
3526 LYXERR(Debug::FIND, "Stringifying with cur: "
3527 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3528 return par.asString(cur.pos(), end,
3529 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3531 } else if (cur.inMathed()) {
3532 CursorSlice cs = cur.top();
3533 MathData md = cs.cell();
3534 MathData::const_iterator it_end =
3535 (( len == -1 || cs.pos() + len > int(md.size()))
3537 : md.begin() + cs.pos() + len );
3539 for (MathData::const_iterator it = md.begin() + cs.pos();
3542 docstring s = asString(md2);
3543 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3546 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3551 /** Computes the LaTeX export of buf starting from cur and ending len positions
3552 * after cur, if len is positive, or at the paragraph or innermost inset end
3555 docstring latexifyFromCursor(DocIterator const & cur, int len)
3557 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3558 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3559 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3560 Buffer const & buf = *cur.buffer();
3562 odocstringstream ods;
3564 //OutputParams runparams(&buf.params().encoding());
3565 OutputParams runparams(encodings.fromLyXName("utf8"));
3566 runparams.nice = false;
3567 runparams.flavor = Flavor::XeTeX;
3568 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3569 // No side effect of file copying and image conversion
3570 runparams.dryrun = true;
3571 runparams.for_search = true;
3573 if (cur.inTexted()) {
3574 // @TODO what about searching beyond/across paragraph breaks ?
3575 pos_type endpos = cur.paragraph().size();
3576 if (len != -1 && endpos > cur.pos() + len)
3577 endpos = cur.pos() + len;
3578 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3579 string(), cur.pos(), endpos);
3580 string s = lyx::to_utf8(ods.str());
3581 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3582 return(lyx::from_utf8(s));
3583 } else if (cur.inMathed()) {
3584 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3585 for (int s = cur.depth() - 1; s >= 0; --s) {
3586 CursorSlice const & cs = cur[s];
3587 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3588 TeXMathStream ws(os);
3589 cs.asInsetMath()->asHullInset()->header_write(ws);
3594 CursorSlice const & cs = cur.top();
3595 MathData md = cs.cell();
3596 MathData::const_iterator it_end =
3597 ((len == -1 || cs.pos() + len > int(md.size()))
3599 : md.begin() + cs.pos() + len);
3601 for (MathData::const_iterator it = md.begin() + cs.pos();
3605 ods << asString(md2);
3606 // Retrieve the math environment type, and add '$' or '$]'
3607 // or others (\end{equation}) accordingly
3608 for (int s = cur.depth() - 1; s >= 0; --s) {
3609 CursorSlice const & cs2 = cur[s];
3610 InsetMath * inset = cs2.asInsetMath();
3611 if (inset && inset->asHullInset()) {
3612 TeXMathStream ws(os);
3613 inset->asHullInset()->footer_write(ws);
3617 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3619 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3624 #if defined(ResultsDebug)
3626 static void displayMResult(MatchResult &mres, int increment)
3628 LYXERR0( "pos: " << mres.pos << " increment " << increment);
3629 LYXERR0( "leadsize: " << mres.leadsize);
3630 LYXERR0( "match_len: " << mres.match_len);
3631 LYXERR0( "match_prefix: " << mres.match_prefix);
3632 LYXERR0( "match2end: " << mres.match2end);
3633 LYXERR0( "pos_len: " << mres.pos_len); // Set in finalize
3634 for (size_t i = 0; i < mres.result.size(); i++)
3635 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3637 #define displayMres(s,i) displayMResult(s,i);
3639 #define displayMres(s,i)
3642 static bool findAdvForwardInnermost(DocIterator & cur)
3645 DocIterator old_cur(cur.buffer());
3646 int forwardCount = 0;
3654 if (cur.depth() > d) {
3658 if (cur.depth() == d)
3662 if (forwardCount > 0) {
3663 LYXERR(Debug::FIND, "Forwarded " << forwardCount << " step(s) (searching for innermost match)");
3670 /** Finalize an advanced find operation, advancing the cursor to the innermost
3671 ** position that matches, plus computing the length of the matching text to
3673 ** Return the cur.pos() difference between start and end of found match
3675 MatchResult &findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
3677 // Search the foremost position that matches (avoids find of entire math
3678 // inset when match at start of it)
3679 DocIterator old_cur(cur.buffer());
3681 static MatchResult fail = MatchResult();
3682 static MatchResult max_match;
3683 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3684 // Happens with e.g. hyperlinks
3685 // either one sees "http://www.bla.bla" or nothing
3686 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3687 // and although we search for only 3 chars, we find the whole hyperlink inset
3688 bool at_begin = (expected.match_prefix == 0);
3689 if (findAdvForwardInnermost(cur)) {
3690 mres = match(cur, -1, at_begin);
3691 displayMres(mres, 0);
3692 if (expected.match_len > 0) {
3693 if (mres.match_len < expected.match_len)
3697 if (mres.match_len <= 0)
3700 max_match = mres.match_len;
3702 else if (expected.match_len < 0) {
3703 mres = match(cur); /* match valid only if not searching whole words */
3704 displayMres(mres, 0);
3708 max_match = expected;
3710 if (max_match.match_len <= 0) return fail;
3711 LYXERR(Debug::FIND, "Ok");
3713 // Compute the match length
3715 if (cur.pos() + len > cur.lastpos())
3717 // regexp should use \w+, \S+, or \b(some string)\b
3718 // to search for whole words
3719 if (match.opt.matchword && !match.use_regexp) {
3720 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3721 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3723 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3725 // Length of matched text (different from len param)
3726 static MatchResult old_match = match(cur, len, at_begin);
3727 if (old_match.match_len < 0)
3729 MatchResult new_match;
3730 // Greedy behaviour while matching regexps
3731 while ((new_match = match(cur, len + 1, at_begin)).match_len > old_match.match_len) {
3733 old_match = new_match;
3734 LYXERR(Debug::FIND, "verifying match with len = " << len);
3740 int maxl = cur.lastpos() - cur.pos();
3741 // Greedy behaviour while matching regexps
3742 while (maxl > minl) {
3744 mres2 = match(cur, len, at_begin);
3745 displayMres(mres2, len);
3746 int actual_match = mres2.match_len;
3747 if (actual_match >= max_match.match_len) {
3748 // actual_match > max_match _can_ happen,
3749 // if the search area splits
3750 // some following word so that the regex
3751 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3753 // This means, the len value is too big
3755 if (maxl - minl < 4)
3756 len = (int)((maxl + minl)/2);
3758 len = (int)(minl + (maxl - minl + 3)/4);
3761 // (actual_match < max_match)
3763 len = (int)((maxl + minl)/2);
3767 // Search for real start of matched characters
3769 MatchResult actual_match;
3772 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3773 if (cur.depth() < old_cur.depth()) {
3775 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3778 if (cur.pos() != old_cur.pos()) {
3779 // OK, forwarded 1 pos in actual inset
3780 actual_match = match(cur, len-1, at_begin);
3781 if (actual_match.match_len == max_match.match_len) {
3782 // Ha, got it! The shorter selection has the same match length
3787 // OK, the shorter selection matches less chars, revert to previous value
3793 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3794 actual_match = match(cur, len, at_begin);
3795 if (actual_match.match_len == max_match.match_len) {
3797 max_match = actual_match;
3804 max_match.pos_len = len;
3811 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3815 while (!theApp()->longOperationCancelled() && cur) {
3816 (void) findAdvForwardInnermost(cur);
3817 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3818 MatchResult mres = match(cur, -1, false);
3819 displayMres(mres,-1)
3820 int match_len = mres.match_len;
3821 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3822 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3825 if (match_len > 0) {
3826 // Try to find the begin of searched string
3828 int firstInvalid = 100000;
3829 if (mres.match_prefix + mres.pos - mres.leadsize > 1)
3830 increment = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3833 LYXERR(Debug::FIND, "Set increment to " << increment);
3834 while (increment > 0) {
3835 DocIterator old_cur = cur;
3836 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3838 if (! cur || (cur.pit() > old_cur.pit())) {
3839 // Are we outside of the paragraph?
3840 // This can happen if moving past some UTF8-encoded chars
3845 MatchResult mres2 = match(cur, -1, false);
3846 displayMres(mres2,increment)
3847 switch (interpretMatch(mres, mres2)) {
3848 case MatchResult::newIsTooFar:
3849 // behind the expected match
3850 firstInvalid = increment;
3854 case MatchResult::newIsBetter:
3855 // not reached ye, but cur.pos()+increment is bettert
3857 firstInvalid -= increment;
3858 if (increment > firstInvalid*3/4)
3859 increment = firstInvalid*3/4;
3860 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3861 if (increment >= mres2.match_prefix)
3862 increment = (mres2.match_prefix+1)*3/4;
3867 // Handle not like MatchResult::newIsTooFar
3868 // LYXERR0( "Something is wrong: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3870 increment = increment*3/4;
3876 // LYXERR0("Leaving first loop");
3878 LYXERR(Debug::FIND, "Finalizing 1");
3879 MatchResult found_match = findAdvFinalize(cur, match, mres);
3880 if (found_match.match_len > 0) {
3881 LASSERT(found_match.pos_len > 0, /**/);
3882 match.FillResults(found_match);
3883 return found_match.pos_len;
3886 // try next possible match
3891 // The following code is newer reached
3892 // but parts of it may be needed in future
3893 int match_len_zero_count = 0;
3895 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3897 mres3 = match(cur, -1, false);
3898 displayMres(mres3, 1)
3899 int remaining_len = mres3.match_len;
3900 if (remaining_len <= 0) {
3901 // Apparently the searched string is not in the remaining part
3908 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3909 mres3 = match(cur, 1);
3910 displayMres(mres3, 1)
3911 int match_len3 = mres3.match_len;
3915 displayMres(mres3, 1)
3916 int match_len2 = mres3.match_len;
3917 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3918 if (match_len2 > 0) {
3919 // Sometimes in finalize we understand it wasn't a match
3920 // and we need to continue the outest loop
3921 LYXERR(Debug::FIND, "Finalizing 2");
3922 MatchResult mres4 = findAdvFinalize(cur, match, mres.match_len);
3923 if (mres4.match_len > 0) {
3924 match.FillResults(mres4);
3925 LASSERT(mres4.pos_len > 0, /**/);
3926 return mres4.pos_len;
3930 match_len_zero_count = 0;
3931 else if (match_len2 == 0)
3932 match_len_zero_count++;
3933 if (match_len2 < 0) {
3934 if (++match_len_zero_count > 3) {
3935 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3943 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3944 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3947 // This should exit nested insets, if any, or otherwise undefine the currsor.
3948 cur.pos() = cur.lastpos();
3949 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3957 /// Find the most backward consecutive match within same paragraph while searching backwards.
3958 MatchResult &findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3960 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3961 DocIterator tmp_cur = cur;
3962 static MatchResult mr = findAdvFinalize(tmp_cur, match, MatchResult(-1));
3963 Inset & inset = cur.inset();
3964 for (; cur != cur_begin; cur.backwardPos()) {
3965 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3966 DocIterator new_cur = cur;
3967 new_cur.backwardPos();
3968 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3970 MatchResult new_mr = findAdvFinalize(new_cur, match, MatchResult(-1));
3971 if (new_mr.match_len == mr.match_len)
3975 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3981 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3985 // Backup of original position
3986 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3987 if (cur == cur_begin)
3990 DocIterator cur_orig(cur);
3991 bool pit_changed = false;
3994 bool found_match = (match(cur, -1, false).match_len > 0);
3998 cur.pos() = cur.lastpos();
4000 cur.pos() = cur_orig.pos();
4001 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
4002 DocIterator cur_prev_iter;
4004 found_match = (match(cur).match_len > 0);
4005 LYXERR(Debug::FIND, "findBackAdv3: found_match="
4006 << found_match << ", cur: " << cur);
4008 MatchResult found_mr = findMostBackwards(cur, match);
4009 match.FillResults(found_mr);
4010 LASSERT(found_mr.pos_len > 0, /**/);
4011 return found_mr.pos_len;
4014 // Stop if begin of document reached
4015 if (cur == cur_begin)
4017 cur_prev_iter = cur;
4021 if (cur == cur_begin)
4028 } while (!theApp()->longOperationCancelled());
4036 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4037 DocIterator const & cur, int len)
4039 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4041 if (!opt.ignoreformat)
4042 return latexifyFromCursor(cur, len);
4044 return stringifyFromCursor(cur, len);
4048 FindAndReplaceOptions::FindAndReplaceOptions(
4049 docstring const & _find_buf_name, bool _casesensitive,
4050 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4051 docstring const & _repl_buf_name, bool _keep_case,
4052 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4053 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4054 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4055 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4063 /** Check if 'len' letters following cursor are all non-lowercase */
4064 static bool allNonLowercase(Cursor const & cur, int len)
4066 pos_type beg_pos = cur.selectionBegin().pos();
4067 pos_type end_pos = cur.selectionBegin().pos() + len;
4068 if (len > cur.lastpos() + 1 - beg_pos) {
4069 LYXERR(Debug::FIND, "This should not happen, more debug needed");
4070 len = cur.lastpos() + 1 - beg_pos;
4071 end_pos = beg_pos + len;
4073 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4074 if (isLowerCase(cur.paragraph().getChar(pos)))
4080 /** Check if first letter is upper case and second one is lower case */
4081 static bool firstUppercase(Cursor const & cur)
4084 pos_type pos = cur.selectionBegin().pos();
4085 if (pos >= cur.lastpos() - 1) {
4086 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
4089 ch1 = cur.paragraph().getChar(pos);
4090 ch2 = cur.paragraph().getChar(pos + 1);
4091 bool result = isUpperCase(ch1) && isLowerCase(ch2);
4092 LYXERR(Debug::FIND, "firstUppercase(): "
4093 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4094 << ch2 << "(" << char(ch2) << ")"
4095 << ", result=" << result << ", cur=" << cur);
4100 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4102 ** \fixme What to do with possible further paragraphs in replace buffer ?
4104 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4106 ParagraphList::iterator pit = buffer.paragraphs().begin();
4107 LASSERT(!pit->empty(), /**/);
4108 pos_type right = pos_type(1);
4109 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4110 right = pit->size();
4111 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4116 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4118 // Should replace the string "$" + std::to_string(matchnum) with replacement
4119 // if the char '$' is not prefixed with odd number of char '\\'
4120 static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4124 for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4126 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4128 int num = stoi(sub.str(3), nullptr, 10);
4129 if (num >= maxmatchnum)
4131 if (lastpos < (size_t) sub.position(2))
4132 s += t.substr(lastpos, sub.position(2) - lastpos);
4133 s += replacements[num];
4134 lastpos = sub.position(2) + sub.length(2);
4138 else if (lastpos < t.length())
4139 s += t.substr(lastpos, t.length() - lastpos);
4146 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4148 Cursor & cur = bv->cursor();
4149 if (opt.repl_buf_name.empty()
4150 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4151 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4154 DocIterator sel_beg = cur.selectionBegin();
4155 DocIterator sel_end = cur.selectionEnd();
4156 if (&sel_beg.inset() != &sel_end.inset()
4157 || sel_beg.pit() != sel_end.pit()
4158 || sel_beg.idx() != sel_end.idx())
4160 int sel_len = sel_end.pos() - sel_beg.pos();
4161 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4162 << ", sel_len: " << sel_len << endl);
4165 LASSERT(sel_len > 0, return 0);
4167 if (!matchAdv(sel_beg, sel_len).match_len)
4170 // Build a copy of the replace buffer, adapted to the KeepCase option
4171 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4173 repl_buffer_orig.write(oss);
4174 string lyx = oss.str();
4175 if (matchAdv.valid_matches > 0) {
4176 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4178 Buffer repl_buffer("", false);
4179 repl_buffer.setUnnamed(true);
4180 LASSERT(repl_buffer.readString(lyx), return 0);
4181 if (opt.keep_case && sel_len >= 2) {
4182 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4183 if (cur.inTexted()) {
4184 if (firstUppercase(cur))
4185 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4186 else if (allNonLowercase(cur, sel_len))
4187 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4190 cap::cutSelection(cur, false);
4191 if (cur.inTexted()) {
4192 repl_buffer.changeLanguage(
4193 repl_buffer.language(),
4194 cur.getFont().language());
4195 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4196 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4197 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4198 repl_buffer.params().documentClassPtr(),
4199 bv->buffer().errorList("Paste"));
4200 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4201 sel_len = repl_buffer.paragraphs().begin()->size();
4202 } else if (cur.inMathed()) {
4203 odocstringstream ods;
4205 // OutputParams runparams(&repl_buffer.params().encoding());
4206 OutputParams runparams(encodings.fromLyXName("utf8"));
4207 runparams.nice = false;
4208 runparams.flavor = Flavor::XeTeX;
4209 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4210 runparams.dryrun = true;
4211 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4212 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4213 docstring repl_latex = ods.str();
4214 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4216 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4217 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4218 repl_latex = from_utf8(s);
4219 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4220 MathData ar(cur.buffer());
4221 asArray(repl_latex, ar, Parse::NORMAL);
4223 sel_len = ar.size();
4224 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4226 if (cur.pos() >= sel_len)
4227 cur.pos() -= sel_len;
4230 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4231 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4232 bv->processUpdateFlags(Update::Force);
4237 /// Perform a FindAdv operation.
4238 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4243 // e.g., when invoking word-findadv from mini-buffer wither with
4244 // wrong options syntax or before ever opening advanced F&R pane
4245 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4249 MatchStringAdv matchAdv(bv->buffer(), opt);
4251 if (!matchAdv.regexIsValid) {
4252 bv->message(lyx::from_utf8(matchAdv.regexError));
4256 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4258 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4259 num_replaced += findAdvReplace(bv, opt, matchAdv);
4262 pos_len = findForwardAdv(cur, matchAdv);
4264 pos_len = findBackwardsAdv(cur, matchAdv);
4265 } catch (exception & ex) {
4266 bv->message(from_utf8(ex.what()));
4271 if (num_replaced > 0) {
4272 switch (num_replaced)
4275 bv->message(_("One match has been replaced."));
4278 bv->message(_("Two matches have been replaced."));
4281 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4287 bv->message(_("Match not found."));
4292 if (num_replaced > 0)
4293 bv->message(_("Match has been replaced."));
4295 bv->message(_("Match found."));
4297 if (cur.pos() + pos_len > cur.lastpos()) {
4298 // Prevent crash in bv->putSelectionAt()
4299 // Should never happen, maybe LASSERT() here?
4300 pos_len = cur.lastpos() - cur.pos();
4302 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4303 bv->putSelectionAt(cur, pos_len, !opt.forward);
4309 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4311 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4312 << opt.casesensitive << ' '
4313 << opt.matchword << ' '
4314 << opt.forward << ' '
4315 << opt.expandmacros << ' '
4316 << opt.ignoreformat << ' '
4317 << opt.replace_all << ' '
4318 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4319 << opt.keep_case << ' '
4320 << int(opt.scope) << ' '
4323 LYXERR(Debug::FIND, "built: " << os.str());
4329 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4331 LYXERR(Debug::FIND, "parsing");
4335 while (line != "EOSS") {
4339 if (is.eof()) // Tolerate malformed request
4343 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4344 opt.find_buf_name = from_utf8(s);
4345 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4346 is.get(); // Waste space before replace string
4349 while (line != "EOSS") {
4353 if (is.eof()) // Tolerate malformed request
4357 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4358 opt.repl_buf_name = from_utf8(s);
4359 is >> opt.keep_case;
4362 opt.scope = FindAndReplaceOptions::SearchScope(i);
4364 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4366 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4367 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4368 << opt.scope << ' ' << opt.restr);