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, true);
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, true);
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, false, true);
471 bool lyxreplace(BufferView * bv, FuncRequest const & ev)
473 if (!bv || ev.action() != LFUN_WORD_REPLACE)
476 // data is of the form
479 // <casesensitive> <matchword> <all> <forward> <findnext>"
482 docstring howto = split(ev.argument(), rplc, '\n');
483 howto = split(howto, search, '\n');
485 bool casesensitive = parse_bool(howto);
486 bool matchword = parse_bool(howto);
487 bool all = parse_bool(howto);
488 bool forward = parse_bool(howto);
489 bool findnext = howto.empty() ? true : parse_bool(howto);
493 int replace_count = 0;
495 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
496 update = replace_count > 0;
499 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
501 replace_count = rv.second;
504 Buffer const & buf = bv->buffer();
506 // emit message signal.
507 buf.message(_("String not found."));
509 if (replace_count == 0) {
510 buf.message(_("String found."));
511 } else if (replace_count == 1) {
512 buf.message(_("String has been replaced."));
514 docstring const str =
515 bformat(_("%1$d strings have been replaced."), replace_count);
523 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
525 for (; cur; cur.forwardPos())
526 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
530 DocIterator cur_orig(bv->cursor());
531 docstring q = _("End of file reached while searching forward.\n"
532 "Continue searching from the beginning?");
533 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
534 q, 0, 1, _("&Yes"), _("&No"));
535 if (wrap_answer == 0) {
536 bv->cursor().clear();
537 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
538 bv->clearSelection();
539 cur.setCursor(bv->cursor().selectionBegin());
540 if (findNextChange(bv, cur, false))
543 bv->cursor().setCursor(cur_orig);
550 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
552 for (cur.backwardPos(); cur; cur.backwardPos()) {
553 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
558 DocIterator cur_orig(bv->cursor());
559 docstring q = _("Beginning of file reached while searching backward.\n"
560 "Continue searching from the end?");
561 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
562 q, 0, 1, _("&Yes"), _("&No"));
563 if (wrap_answer == 0) {
564 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
565 bv->cursor().backwardPos();
566 bv->clearSelection();
567 cur.setCursor(bv->cursor().selectionBegin());
568 if (findPreviousChange(bv, cur, false))
571 bv->cursor().setCursor(cur_orig);
578 bool selectChange(Cursor & cur, bool forward)
580 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
582 Change ch = cur.paragraph().lookupChange(cur.pos());
584 CursorSlice tip1 = cur.top();
585 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
586 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
587 if (!ch2.isSimilarTo(ch))
590 CursorSlice tip2 = cur.top();
591 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
593 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
594 if (!ch2.isSimilarTo(ch)) {
595 // take a step forward to correctly set the selection
603 cur.bv().mouseSetCursor(cur, false);
605 cur.bv().mouseSetCursor(cur, true);
613 bool findChange(BufferView * bv, bool forward)
616 cur.setCursor(forward ? bv->cursor().selectionEnd()
617 : bv->cursor().selectionBegin());
618 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
619 return selectChange(cur, forward);
624 bool findNextChange(BufferView * bv)
626 return findChange(bv, true);
630 bool findPreviousChange(BufferView * bv)
632 return findChange(bv, false);
639 typedef vector<pair<string, string> > Escapes;
641 /// A map of symbols and their escaped equivalent needed within a regex.
642 /// @note Beware of order
643 Escapes const & get_regexp_escapes()
645 typedef std::pair<std::string, std::string> P;
647 static Escapes escape_map;
648 if (escape_map.empty()) {
649 escape_map.push_back(P("$", "_x_$"));
650 escape_map.push_back(P("{", "_x_{"));
651 escape_map.push_back(P("}", "_x_}"));
652 escape_map.push_back(P("[", "_x_["));
653 escape_map.push_back(P("]", "_x_]"));
654 escape_map.push_back(P("(", "_x_("));
655 escape_map.push_back(P(")", "_x_)"));
656 escape_map.push_back(P("+", "_x_+"));
657 escape_map.push_back(P("*", "_x_*"));
658 escape_map.push_back(P(".", "_x_."));
659 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
660 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
661 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
662 escape_map.push_back(P("_x_", "\\"));
667 /// A map of lyx escaped strings and their unescaped equivalent.
668 Escapes const & get_lyx_unescapes()
670 typedef std::pair<std::string, std::string> P;
672 static Escapes escape_map;
673 if (escape_map.empty()) {
674 escape_map.push_back(P("\\%", "%"));
675 escape_map.push_back(P("\\{", "{"));
676 escape_map.push_back(P("\\}", "}"));
677 escape_map.push_back(P("\\mathcircumflex ", "^"));
678 escape_map.push_back(P("\\mathcircumflex", "^"));
679 escape_map.push_back(P("\\backslash ", "\\"));
680 escape_map.push_back(P("\\backslash", "\\"));
681 escape_map.push_back(P("\\sim ", "~"));
682 escape_map.push_back(P("\\sim", "~"));
687 /// A map of escapes turning a regexp matching text to one matching latex.
688 Escapes const & get_regexp_latex_escapes()
690 typedef std::pair<std::string, std::string> P;
692 static Escapes escape_map;
693 if (escape_map.empty()) {
694 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
695 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
696 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
697 escape_map.push_back(P("\\[", "\\{\\[\\}"));
698 escape_map.push_back(P("\\]", "\\{\\]\\}"));
699 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
700 escape_map.push_back(P("%", "\\\\\\%"));
701 escape_map.push_back(P("#", "\\\\#"));
706 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
707 ** the found occurrence were escaped.
709 string apply_escapes(string s, Escapes const & escape_map)
711 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
712 Escapes::const_iterator it;
713 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
714 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
715 unsigned int pos = 0;
716 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
717 s.replace(pos, it->first.length(), it->second);
718 LYXERR(Debug::FIND, "After escape: " << s);
719 pos += it->second.length();
720 // LYXERR(Debug::FIND, "pos: " << pos);
723 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
728 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
729 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
730 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
731 string escape_for_regex(string s, bool match_latex)
734 while (pos < s.size()) {
735 size_t new_pos = s.find("\\regexp{", pos);
736 if (new_pos == string::npos)
741 LYXERR(Debug::FIND, "new_pos: " << new_pos);
742 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
743 LYXERR(Debug::FIND, "t [lyx]: " << t);
744 t = apply_escapes(t, get_regexp_escapes());
745 LYXERR(Debug::FIND, "t [rxp]: " << t);
746 s.replace(pos, new_pos - pos, t);
747 new_pos = pos + t.size();
748 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
749 LYXERR(Debug::FIND, "new_pos: " << new_pos);
750 if (new_pos == s.size())
753 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
754 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
755 LYXERR(Debug::FIND, "end_pos: " << end_pos);
756 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
757 LYXERR(Debug::FIND, "t in regexp : " << t);
758 t = apply_escapes(t, get_lyx_unescapes());
759 LYXERR(Debug::FIND, "t in regexp after unescapes [lyx]: " << t);
761 t = apply_escapes(t, get_regexp_latex_escapes());
762 LYXERR(Debug::FIND, "t in regexp after latex_escapes [ltx]: " << t);
764 if (end_pos == s.size()) {
765 s.replace(new_pos, end_pos - new_pos, t);
766 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
769 s.replace(new_pos, end_pos + 13 - new_pos, t);
770 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
771 pos = new_pos + t.size();
772 LYXERR(Debug::FIND, "pos: " << pos);
778 /// Wrapper for lyx::regex_replace with simpler interface
779 bool regex_replace(string const & s, string & t, string const & searchstr,
780 string const & replacestr)
782 regex e(searchstr, regex_constants::ECMAScript);
784 ostream_iterator<char, char> it(oss);
785 regex_replace(it, s.begin(), s.end(), e, replacestr);
786 // tolerate t and s be references to the same variable
787 bool rv = (s != oss.str());
793 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
795 ** Verify that closed braces exactly match open braces. This avoids that, for example,
796 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
799 ** Number of open braces that must remain open at the end for the verification to succeed.
802 bool braces_match(QString const & beg,
805 bool braces_match(string const & beg,
811 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg.toStdString() << "'");
813 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg << "'");
815 int lastidx = beg.size();
816 for (int i=0; i < lastidx; ++i) {
817 // Skip escaped braces in the count
827 } else if (c == '{') {
829 } else if (c == '}') {
830 if (open_pars == 0) {
831 LYXERR(Debug::FIND, "Found unmatched closed brace");
837 if (open_pars != unmatched) {
838 LYXERR(Debug::FIND, "Found " << open_pars
839 << " instead of " << unmatched
840 << " unmatched open braces at the end of count");
843 LYXERR(Debug::FIND, "Braces match as expected");
861 vector <string> result = vector <string>();
862 MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1) {};
865 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
867 int range = oldres.match_len;
868 if (range < 2) range = 2;
869 if (newres.match2end < oldres.match2end - oldres.match_len)
870 return MatchResult::newIsTooFar;
871 if (newres.match_len < oldres.match_len)
872 return MatchResult::newIsTooFar;
873 if ((newres.match_len == oldres.match_len) &&
874 (newres.match2end < oldres.match2end + range) &&
875 (newres.match2end > oldres.match2end - range)) {
876 return MatchResult::newIsBetter;
878 return MatchResult::newIsInvalid;
881 /** The class performing a match between a position in the document and the FindAdvOptions.
884 class MatchStringAdv {
886 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
888 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
889 ** constructor as opt.search, under the opt.* options settings.
892 ** If set, then match is searched only against beginning of text starting at cur.
893 ** If unset, then match is searched anywhere in text starting at cur.
896 ** The length of the matching text, or zero if no match was found.
898 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
907 /// first buffer on which search was started
908 lyx::Buffer * const p_first_buf;
910 FindAndReplaceOptions const & opt;
913 /// Auxiliary find method (does not account for opt.matchword)
914 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
915 void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
917 /** Normalize a stringified or latexified LyX paragraph.
921 ** <li>if search is not casesensitive, then lowercase the string;
922 ** <li>remove any newline at begin or end of the string;
923 ** <li>replace any newline in the middle of the string with a simple space;
924 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
927 ** @todo Normalization should also expand macros, if the corresponding
928 ** search option was checked.
930 string normalize(docstring const & s) const;
931 // normalized string to search
932 string par_as_string;
933 // regular expression to use for searching
934 // regexp2 is same as regexp, but prefixed with a ".*?"
936 QRegularExpression regexp;
937 QRegularExpression regexp2;
942 // leading format material as string
943 string lead_as_string;
944 // par_as_string after removal of lead_as_string
945 string par_as_string_nolead;
946 // unmatched open braces in the search string/regexp
948 // number of (.*?) subexpressions added at end of search regexp for closing
949 // environments, math mode, styles, etc...
952 // Are we searching with regular expressions ?
954 static int valid_matches;
955 static vector <string> matches;
956 void FillResults(MatchResult &found_mr);
959 int MatchStringAdv::valid_matches = 0;
960 vector <string> MatchStringAdv::matches = vector <string>(10);
962 void MatchStringAdv::FillResults(MatchResult &found_mr)
964 if (found_mr.match_len > 0) {
965 valid_matches = found_mr.result.size();
966 for (size_t i = 0; i < found_mr.result.size(); i++)
967 matches[i] = found_mr.result[i];
973 static docstring buffer_to_latex(Buffer & buffer)
975 //OutputParams runparams(&buffer.params().encoding());
976 OutputParams runparams(encodings.fromLyXName("utf8"));
977 odocstringstream ods;
979 runparams.nice = true;
980 runparams.flavor = Flavor::XeTeX;
981 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
982 // No side effect of file copying and image conversion
983 runparams.dryrun = true;
984 runparams.for_search = true;
985 pit_type const endpit = buffer.paragraphs().size();
986 for (pit_type pit = 0; pit != endpit; ++pit) {
987 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
988 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
994 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
997 if (!opt.ignoreformat) {
998 str = buffer_to_latex(buffer);
1000 // OutputParams runparams(&buffer.params().encoding());
1001 OutputParams runparams(encodings.fromLyXName("utf8"));
1002 runparams.nice = true;
1003 runparams.flavor = Flavor::XeTeX;
1004 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1005 runparams.dryrun = true;
1006 runparams.for_search = true;
1007 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1008 Paragraph const & par = buffer.paragraphs().at(pit);
1009 LYXERR(Debug::FIND, "Adding to search string: '"
1010 << par.asString(pos_type(0), par.size(),
1011 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1014 str += par.asString(pos_type(0), par.size(),
1015 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1018 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1019 string t = to_utf8(str);
1020 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
1027 /// Return separation pos between the leading material and the rest
1028 static size_t identifyLeading(string const & s)
1031 // @TODO Support \item[text]
1032 // Kornel: Added textsl, textsf, textit, texttt and noun
1033 // + allow to search for colored text too
1034 while (regex_replace(t, t, "^\\\\(("
1035 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1036 "lyxaddress|lyxrightaddress|"
1037 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1038 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1039 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1040 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1041 || regex_replace(t, t, "^\\$", "")
1042 || regex_replace(t, t, "^\\\\\\[", "")
1043 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1044 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1046 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1051 * Given a latexified string, retrieve some handled features
1052 * The features of the regex will later be compared with the features
1053 * of the searched text. If the regex features are not a
1054 * subset of the analized, then, in not format ignoring search
1055 * we can early stop the search in the relevant inset.
1057 typedef map<string, bool> Features;
1059 static Features identifyFeatures(string const & s)
1061 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1062 static regex const valid("^("
1064 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1065 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1066 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1067 "(u|uu)line|(s|x)out|uwave|"
1068 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1069 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1074 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1077 if (sub.str(1).compare("regexp") == 0) {
1081 string token = sub.str(1);
1083 if (regex_match(token, sub2, valid)) {
1091 if (sub.str(1).compare("endregexp") == 0) {
1101 * defines values features of a key "\\[a-z]+{"
1106 /* Char type with content discarded
1107 * like \hspace{1cm} */
1109 /* Char, like \backslash */
1111 /* replace starting backslash with '#' */
1113 /* \part, \section*, ... */
1115 /* title, author etc */
1117 /* \foreignlanguage{ngerman}, ... */
1120 * to discard language in content */
1123 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1125 /* fonts, colors, markups, ... */
1127 /* footnotesize, ... large, ...
1128 * Ignore all of them */
1131 /* inputencoding, ...
1132 * Discard also content, because they do not help in search */
1135 * like remove, but also all arguments */
1137 /* item, listitem */
1139 /* tex, latex, ... like isChar */
1141 /* like \lettrine[lines=5]{}{} */
1143 // like isStandard, but always remove head
1145 /* End of arguments marker for lettrine,
1146 * so that they can be ignored */
1149 KeyInfo() = default;
1150 KeyInfo(KeyType type, int parcount, bool disable)
1152 parenthesiscount(parcount),
1153 disabled(disable) {}
1154 KeyType keytype = invalid;
1156 int _tokensize = -1;
1157 int _tokenstart = -1;
1158 int _dataStart = -1;
1160 int parenthesiscount = 1;
1161 bool disabled = false;
1162 bool used = false; /* by pattern */
1167 Border(int l=0, int u=0) : low(l), upper(u) {};
1172 #define MAXOPENED 30
1174 bool isPatternString_;
1176 explicit Intervall(bool isPattern, string const & p) :
1177 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1178 hasTitle(false), langcount(0)
1186 static vector<Border> borders;
1187 int depts[MAXOPENED];
1188 int closes[MAXOPENED];
1189 int actualdeptindex;
1190 int previousNotIgnored(int) const;
1191 int nextNotIgnored(int) const;
1192 void handleOpenP(int i);
1193 void handleCloseP(int i, bool closingAllowed);
1194 void resetOpenedP(int openPos);
1195 void addIntervall(int upper);
1196 void addIntervall(int low, int upper); /* if explicit */
1197 void removeAccents();
1198 void setForDefaultLang(KeyInfo const & defLang) const;
1199 int findclosing(int start, int end, char up, char down, int repeat);
1200 void handleParentheses(int lastpos, bool closingAllowed);
1202 int langcount; // Number of disabled language specs up to current position in actual interval
1203 int isOpeningPar(int pos) const;
1205 void output(ostringstream &os, int lastpos);
1206 // string show(int lastpos);
1209 vector<Border> Intervall::borders = vector<Border>(30);
1211 int Intervall::isOpeningPar(int pos) const
1213 if ((pos < 0) || (size_t(pos) >= par.size()))
1215 if (par[pos] != '{')
1217 if (size_t(pos) + 2 >= par.size())
1219 if (par[pos+2] != '}')
1221 if (par[pos+1] == '[' || par[pos+1] == ']')
1226 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1228 // Enable the use of first token again
1229 if (ignoreidx >= 0) {
1230 int value = defLang._tokenstart + defLang._tokensize;
1236 if (borders[borderidx].low < value)
1237 borders[borderidx].low = value;
1238 if (borders[borderidx].upper < value)
1239 borders[borderidx].upper = value;
1244 static void checkDepthIndex(int val)
1246 static int maxdepthidx = MAXOPENED-2;
1247 static int lastmaxdepth = 0;
1248 if (val > lastmaxdepth) {
1249 LYXERR(Debug::INFO, "Depth reached " << val);
1252 if (val > maxdepthidx) {
1254 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1259 // Not needed, because borders are now dynamically expanded
1260 static void checkIgnoreIdx(int val)
1262 static int lastmaxignore = -1;
1263 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1264 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1265 lastmaxignore = val;
1271 * Expand the region of ignored parts of the input latex string
1272 * The region is only relevant in output()
1274 void Intervall::addIntervall(int low, int upper)
1277 if (low == upper) return;
1278 for (idx = ignoreidx+1; idx > 0; --idx) {
1279 if (low > borders[idx-1].upper) {
1283 Border br(low, upper);
1284 if (idx > ignoreidx) {
1285 if (borders.size() <= size_t(idx)) {
1286 borders.push_back(br);
1292 // checkIgnoreIdx(ignoreidx);
1296 // Expand only if one of the new bound is inside the interwall
1297 // We know here that br.low > borders[idx-1].upper
1298 if (br.upper < borders[idx].low) {
1299 // We have to insert at this pos
1300 if (size_t(ignoreidx+1) >= borders.size()) {
1301 borders.push_back(borders[ignoreidx]);
1304 borders[ignoreidx+1] = borders[ignoreidx];
1306 for (int i = ignoreidx; i > idx; --i) {
1307 borders[i] = borders[i-1];
1311 // checkIgnoreIdx(ignoreidx);
1314 // Here we know, that we are overlapping
1315 if (br.low > borders[idx].low)
1316 br.low = borders[idx].low;
1317 // check what has to be concatenated
1319 for (int i = idx; i <= ignoreidx; i++) {
1320 if (br.upper >= borders[i].low) {
1322 if (br.upper < borders[i].upper)
1323 br.upper = borders[i].upper;
1329 // count should be >= 1 here
1332 for (int i = idx + count; i <= ignoreidx; i++) {
1333 borders[i-count+1] = borders[i];
1335 ignoreidx -= count - 1;
1341 typedef map<string, string> AccentsMap;
1342 static AccentsMap accents = map<string, string>();
1344 static void buildaccent(string n, string param, string values)
1348 const char delim = '|';
1349 while (getline(s, name, delim)) {
1351 for (char c : param) {
1352 string key = name + "{" + c + "}";
1353 // get the corresponding utf8-value
1354 if ((values[start] & 0xc0) != 0xc0) {
1355 // should not happen, utf8 encoding starts at least with 11xxxxxx
1356 // but value for '\dot{i}' is 'i', which is ascii
1357 if ((values[start] & 0x80) == 0) {
1359 accents[key] = values.substr(start, 1);
1360 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1365 for (int j = 1; ;j++) {
1366 if (start + j >= values.size()) {
1367 accents[key] = values.substr(start, j);
1368 start = values.size() - 1;
1371 else if ((values[start+j] & 0xc0) != 0x80) {
1372 // This is the first byte of following utf8 char
1373 accents[key] = values.substr(start, j);
1375 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1384 static string getutf8(unsigned uchar)
1387 string ret = string();
1389 if (uchar <= 0x7f) {
1390 c[maxc-1] = uchar & 0x7f;
1393 unsigned char rest = 0x40;
1394 unsigned char first = 0x80;
1396 for (int i = start; i >=0; --i) {
1398 c[i] = first + uchar;
1401 c[i] = 0x80 | (uchar & 0x3f);
1408 for (int i = 0; i < maxc; i++) {
1409 if (c[i] == 0) continue;
1415 static void buildAccentsMap()
1417 accents["imath"] = "ı";
1419 accents["jmath"] = "ȷ";
1420 accents["cdot"] = "·";
1421 accents["guillemotright"] = "»";
1422 accents["guillemotleft"] = "«";
1423 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1424 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1425 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1426 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1427 accents["negmedspace"] = getutf8(0xf0005);
1428 accents["thickspace"] = getutf8(0xf0006);
1429 accents["negthickspace"] = getutf8(0xf0007);
1430 accents["lyx"] = getutf8(0xf0010); // Used logos
1431 accents["LyX"] = getutf8(0xf0010);
1432 accents["tex"] = getutf8(0xf0011);
1433 accents["TeX"] = getutf8(0xf0011);
1434 accents["latex"] = getutf8(0xf0012);
1435 accents["LaTeX"] = getutf8(0xf0012);
1436 accents["latexe"] = getutf8(0xf0013);
1437 accents["LaTeXe"] = getutf8(0xf0013);
1438 accents["lyxarrow"] = getutf8(0xf0020);
1439 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1440 accents["backslash LyX"] = getutf8(0xf0010);
1441 accents["backslash tex"] = getutf8(0xf0011);
1442 accents["backslash TeX"] = getutf8(0xf0011);
1443 accents["backslash latex"] = getutf8(0xf0012);
1444 accents["backslash LaTeX"] = getutf8(0xf0012);
1445 accents["backslash latexe"] = getutf8(0xf0013);
1446 accents["backslash LaTeXe"] = getutf8(0xf0013);
1447 accents["backslash lyxarrow"] = getutf8(0xf0020);
1448 accents["ddot{\\imath}"] = "ï";
1449 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1450 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1451 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1452 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1453 accents["acute{\\imath}"] = "í";
1454 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1455 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1456 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1457 buildaccent("mathring|r", "aAuUwy",
1459 accents["check{\\imath}"] = "ǐ";
1460 accents["check{\\jmath}"] = "ǰ";
1461 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1462 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1463 accents["hat{\\imath}"] = "î";
1464 accents["hat{\\jmath}"] = "ĵ";
1465 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1466 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1467 accents["bar{\\imath}"] = "ī";
1468 buildaccent("bar|=", "aAeEiIoOuUyY",
1469 "āĀēĒīĪōŌūŪȳȲ"); // macron
1470 accents["tilde{\\imath}"] = "ĩ";
1471 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1472 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1473 accents["breve{\\imath}"] = "ĭ";
1474 buildaccent("breve|u", "aAeEgGiIoOuU",
1475 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1476 accents["grave{\\imath}"] = "ì";
1477 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1478 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1479 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1480 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1481 buildaccent("ogonek|k", "AaEeIiUuOo",
1482 "ĄąĘęĮįŲųǪǫ"); // ogonek
1483 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1484 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1485 buildaccent("subring|textsubring", "Aa",
1487 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1488 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1489 buildaccent("subtilde|textsubtilde", "EeIiUu",
1490 "ḚḛḬḭṴṵ"); // subtilde
1491 accents["dgrave{\\imath}"] = "ȉ";
1492 accents["textdoublegrave{\\i}"] = "ȉ";
1493 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1494 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1495 accents["rcap{\\imath}"] = "ȉ";
1496 accents["textroundcap{\\i}"] = "ȉ";
1497 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1498 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1499 buildaccent("slashed", "oO",
1504 * Created accents in math or regexp environment
1505 * are macros, but we need the utf8 equivalent
1507 void Intervall::removeAccents()
1509 if (accents.empty())
1511 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1512 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1513 "|((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]))");
1515 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1517 string key = sub.str(1);
1518 if (accents.find(key) != accents.end()) {
1519 string val = accents[key];
1520 size_t pos = sub.position(size_t(0));
1521 for (size_t i = 0; i < val.size(); i++) {
1522 par[pos+i] = val[i];
1524 // Remove possibly following space too
1525 if (par[pos+sub.str(0).size()] == ' ')
1526 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1528 addIntervall(pos+val.size(), pos + sub.str(0).size());
1529 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1530 // remove traces of any remaining chars
1535 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1540 void Intervall::handleOpenP(int i)
1543 depts[actualdeptindex] = i+1;
1544 closes[actualdeptindex] = -1;
1545 checkDepthIndex(actualdeptindex);
1548 void Intervall::handleCloseP(int i, bool closingAllowed)
1550 if (actualdeptindex <= 0) {
1551 if (! closingAllowed)
1552 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1553 // if we are at the very end
1554 addIntervall(i, i+1);
1557 closes[actualdeptindex] = i+1;
1562 void Intervall::resetOpenedP(int openPos)
1564 // Used as initializer for foreignlanguage entry
1565 actualdeptindex = 1;
1566 depts[1] = openPos+1;
1570 int Intervall::previousNotIgnored(int start) const
1572 int idx = 0; /* int intervalls */
1573 for (idx = ignoreidx; idx >= 0; --idx) {
1574 if (start > borders[idx].upper)
1576 if (start >= borders[idx].low)
1577 start = borders[idx].low-1;
1582 int Intervall::nextNotIgnored(int start) const
1584 int idx = 0; /* int intervalls */
1585 for (idx = 0; idx <= ignoreidx; idx++) {
1586 if (start < borders[idx].low)
1588 if (start < borders[idx].upper)
1589 start = borders[idx].upper;
1594 typedef map<string, KeyInfo> KeysMap;
1595 typedef vector< KeyInfo> Entries;
1596 static KeysMap keys = map<string, KeyInfo>();
1602 Intervall interval_;
1603 void buildKeys(bool);
1604 void buildEntries(bool);
1605 void makeKey(const string &, KeyInfo, bool isPatternString);
1606 void processRegion(int start, int region_end); /* remove {} parts */
1607 void removeHead(KeyInfo const &, int count=0);
1610 LatexInfo(string const & par, bool isPatternString)
1611 : entidx_(-1), interval_(isPatternString, par)
1613 buildKeys(isPatternString);
1614 entries_ = vector<KeyInfo>();
1615 buildEntries(isPatternString);
1619 if (entries_.empty()) {
1622 if (entries_[0].keytype == KeyInfo::isTitle) {
1623 interval_.hasTitle = true;
1624 if (! entries_[0].disabled) {
1625 interval_.titleValue = entries_[0].head;
1628 interval_.titleValue = "";
1630 removeHead(entries_[0]);
1631 if (entries_.size() > 1)
1640 if (int(entries_.size()) > entidx_) {
1647 bool setNextKey(int idx) {
1648 if ((idx == entidx_) && (entidx_ >= 0)) {
1655 int find(int start, KeyInfo::KeyType keytype) const {
1659 while (tmpIdx < int(entries_.size())) {
1660 if (entries_[tmpIdx].keytype == keytype)
1666 int process(ostringstream & os, KeyInfo const & actual);
1667 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1668 // string show(int lastpos) { return interval.show(lastpos);};
1669 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1670 KeyInfo &getKeyInfo(int keyinfo) {
1671 static KeyInfo invalidInfo = KeyInfo();
1672 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1675 return entries_[keyinfo];
1677 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1678 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1682 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1686 for (int i = start; i < end; i += 1 + skip) {
1690 if (c == '\\') skip = 1;
1694 else if (c == down) {
1697 if ((repeat <= 0) || (par[i+1] != up))
1715 vector<MathEntry> entries_;
1720 void insert(string const & wait, size_t start, size_t end) {
1721 MathEntry m = MathEntry();
1723 m.mathStart = start;
1725 m.mathSize = end - start;
1726 entries_.push_back(m);
1728 bool empty() const { return entries_.empty(); };
1729 size_t getEndPos() const {
1730 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1733 return entries_[actualIdx_].mathEnd;
1735 size_t getStartPos() const {
1736 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1737 return 100000; /* definitely enough? */
1739 return entries_[actualIdx_].mathStart;
1741 size_t getFirstPos() {
1743 return getStartPos();
1745 size_t getSize() const {
1746 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1749 return entries_[actualIdx_].mathSize;
1751 void incrEntry() { actualIdx_++; };
1754 void LatexInfo::buildEntries(bool isPatternString)
1756 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\})");
1757 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1758 static bool disableLanguageOverride = false;
1759 smatch sub, submath;
1760 bool evaluatingRegexp = false;
1762 bool evaluatingMath = false;
1763 bool evaluatingCode = false;
1765 bool evaluatingOptional = false;
1766 size_t optionalEnd = 0;
1769 bool math_end_waiting = false;
1770 size_t math_pos = 10000;
1772 static vector<string> usedText = vector<string>();
1774 interval_.removeAccents();
1776 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1778 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1779 // prefixed by odd count of '\\'
1782 if (math_end_waiting) {
1783 size_t pos = submath.position(size_t(2));
1784 if ((math_end == "$") &&
1785 (submath.str(2) == "$")) {
1786 mi.insert("$", math_pos, pos + 1);
1787 math_end_waiting = false;
1789 else if ((math_end == "\\]") &&
1790 (submath.str(2) == "\\]")) {
1791 mi.insert("\\]", math_pos, pos + 2);
1792 math_end_waiting = false;
1794 else if ((submath.str(3).compare("end") == 0) &&
1795 (submath.str(4).compare(math_end) == 0)) {
1796 mi.insert(math_end, math_pos, pos + submath.str(2).length());
1797 math_end_waiting = false;
1803 if (submath.str(3).compare("begin") == 0) {
1804 math_end_waiting = true;
1805 math_end = submath.str(4);
1806 math_pos = submath.position(size_t(2));
1808 else if (submath.str(2).compare("\\[") == 0) {
1809 math_end_waiting = true;
1811 math_pos = submath.position(size_t(2));
1813 else if (submath.str(2) == "$") {
1814 size_t pos = submath.position(size_t(2));
1815 math_end_waiting = true;
1821 // Ignore language if there is math somewhere in pattern-string
1822 if (isPatternString) {
1823 for (auto s: usedText) {
1824 // Remove entries created in previous search runs
1827 usedText = vector<string>();
1830 keys["foreignlanguage"].disabled = true;
1831 disableLanguageOverride = true;
1834 disableLanguageOverride = false;
1837 if (disableLanguageOverride) {
1838 keys["foreignlanguage"].disabled = true;
1841 math_pos = mi.getFirstPos();
1842 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1844 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1845 // prefixed by odd count of '\\'
1848 string key = sub.str(5);
1850 if (sub.str(2)[0] == '\\')
1851 key = sub.str(2)[1];
1856 if (keys.find(key) != keys.end()) {
1857 if (keys[key].keytype == KeyInfo::headRemove) {
1858 KeyInfo found1 = keys[key];
1859 found1.disabled = true;
1860 found1.head = "\\" + key + "{";
1861 found1._tokenstart = sub.position(size_t(2));
1862 found1._tokensize = found1.head.length();
1863 found1._dataStart = found1._tokenstart + found1.head.length();
1864 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1865 found1._dataEnd = endpos;
1870 if (evaluatingRegexp) {
1871 if (sub.str(3).compare("endregexp") == 0) {
1872 evaluatingRegexp = false;
1873 // found._tokenstart already set
1874 found._dataEnd = sub.position(size_t(2)) + 13;
1875 found._dataStart = found._dataEnd;
1876 found._tokensize = found._dataEnd - found._tokenstart;
1877 found.parenthesiscount = 0;
1878 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1885 if (evaluatingMath) {
1886 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1888 evaluatingMath = false;
1890 math_pos = mi.getStartPos();
1892 if (keys.find(key) == keys.end()) {
1893 found = KeyInfo(KeyInfo::isStandard, 0, true);
1894 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1895 found = KeyInfo(KeyInfo::isText, 0, false);
1896 if (isPatternString) {
1897 found.keytype = KeyInfo::isChar;
1898 found.disabled = false;
1902 usedText.push_back(key);
1906 if (key.compare("regexp") == 0) {
1907 evaluatingRegexp = true;
1908 found._tokenstart = sub.position(size_t(2));
1909 found._tokensize = 0;
1913 // Handle the other params of key
1914 if (found.keytype == KeyInfo::isIgnored)
1916 else if (found.keytype == KeyInfo::isMath) {
1917 if (size_t(sub.position(size_t(2))) == math_pos) {
1919 found._tokenstart = sub.position(size_t(2));
1920 found._tokensize = mi.getSize();
1921 found._dataEnd = found._tokenstart + found._tokensize;
1922 found._dataStart = found._dataEnd;
1923 found.parenthesiscount = 0;
1924 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1925 evaluatingMath = true;
1928 // begin|end of unknown env, discard
1929 // First handle tables
1930 // longtable|tabular
1931 bool discardComment;
1933 found.keytype = KeyInfo::doRemove;
1934 if ((sub.str(7).compare("longtable") == 0) ||
1935 (sub.str(7).compare("tabular") == 0)) {
1936 discardComment = true; /* '%' */
1939 discardComment = false;
1940 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1942 string token = sub.str(7);
1943 if (regex_match(token, sub2, removeArgs)) {
1944 found.keytype = KeyInfo::removeWithArg;
1947 // discard spaces before pos(2)
1948 int pos = sub.position(size_t(2));
1950 for (count = 0; pos - count > 0; count++) {
1951 char c = interval_.par[pos-count-1];
1952 if (discardComment) {
1953 if ((c != ' ') && (c != '%'))
1959 found._tokenstart = pos - count;
1960 if (sub.str(3).compare(0, 5, "begin") == 0) {
1961 size_t pos1 = pos + sub.str(2).length();
1962 if (sub.str(7).compare("cjk") == 0) {
1963 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1964 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1966 found.keytype = KeyInfo::isMain;
1967 found._dataStart = pos1;
1968 found._dataEnd = interval_.par.length();
1969 found.disabled = keys["foreignlanguage"].disabled;
1970 found.used = keys["foreignlanguage"].used;
1971 found._tokensize = pos1 - found._tokenstart;
1972 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1975 // Swallow possible optional params
1976 while (interval_.par[pos1] == '[') {
1977 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1979 // Swallow also the eventual parameter
1980 if (interval_.par[pos1] == '{') {
1981 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1984 found._dataEnd = pos1;
1986 found._dataStart = found._dataEnd;
1987 found._tokensize = count + found._dataEnd - pos;
1988 found.parenthesiscount = 0;
1989 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1990 found.disabled = true;
1994 // Handle "\end{...}"
1995 found._dataStart = pos + sub.str(2).length();
1996 found._dataEnd = found._dataStart;
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 else if (found.keytype != KeyInfo::isRegex) {
2005 found._tokenstart = sub.position(size_t(2));
2006 if (found.parenthesiscount == 0) {
2007 // Probably to be discarded
2008 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2009 char following = interval_.par[following_pos];
2010 if (following == ' ')
2011 found.head = "\\" + sub.str(5) + " ";
2012 else if (following == '=') {
2013 // like \uldepth=1000pt
2014 found.head = sub.str(2);
2017 found.head = "\\" + key;
2018 found._tokensize = found.head.length();
2019 found._dataEnd = found._tokenstart + found._tokensize;
2020 found._dataStart = found._dataEnd;
2023 int params = found._tokenstart + key.length() + 1;
2024 if (evaluatingOptional) {
2025 if (size_t(found._tokenstart) > optionalEnd) {
2026 evaluatingOptional = false;
2029 found.disabled = true;
2032 int optend = params;
2033 while (interval_.par[optend] == '[') {
2034 // discard optional parameters
2035 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2037 if (optend > params) {
2038 key += interval_.par.substr(params, optend-params);
2039 evaluatingOptional = true;
2040 optionalEnd = optend;
2042 string token = sub.str(7);
2044 if (interval_.par[optend] != '{') {
2046 found.parenthesiscount = 0;
2047 found.head = "\\" + key;
2050 closings = found.parenthesiscount;
2051 if (found.parenthesiscount == 1) {
2052 found.head = "\\" + key + "{";
2054 else if (found.parenthesiscount > 1) {
2056 found.head = sub.str(2) + "{";
2057 closings = found.parenthesiscount - 1;
2060 found.head = "\\" + key + "{";
2063 found._tokensize = found.head.length();
2064 found._dataStart = found._tokenstart + found.head.length();
2065 if (found.keytype == KeyInfo::doRemove) {
2067 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2068 if (endpar >= interval_.par.length())
2069 found._dataStart = interval_.par.length();
2071 found._dataStart = endpar;
2072 found._tokensize = found._dataStart - found._tokenstart;
2075 found._dataStart = found._tokenstart + found._tokensize;
2079 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
2080 found._dataStart += 15;
2084 endpos = found._dataStart - 1;
2086 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2087 if (found.keytype == KeyInfo::isList) {
2088 // Check if it really is list env
2089 static regex const listre("^([a-z]+)$");
2091 if (!regex_match(token, sub2, listre)) {
2092 // Change the key of this entry. It is not in a list/item environment
2093 found.keytype = KeyInfo::endArguments;
2096 if (found.keytype == KeyInfo::noMain) {
2097 evaluatingCode = true;
2099 codeStart = found._dataStart;
2101 else if (evaluatingCode) {
2102 if (size_t(found._dataStart) > codeEnd)
2103 evaluatingCode = false;
2104 else if (found.keytype == KeyInfo::isMain) {
2105 // Disable this key, treate it as standard
2106 found.keytype = KeyInfo::isStandard;
2107 found.disabled = true;
2108 if ((codeEnd +1 >= interval_.par.length()) &&
2109 (found._tokenstart == codeStart)) {
2110 // trickery, because the code inset starts
2111 // with \selectlanguage ...
2113 if (entries_.size() > 1) {
2114 entries_[entries_.size()-1]._dataEnd = codeEnd;
2119 if ((endpos == interval_.par.length()) &&
2120 (found.keytype == KeyInfo::doRemove)) {
2121 // Missing closing => error in latex-input?
2122 // therefore do not delete remaining data
2123 found._dataStart -= 1;
2124 found._dataEnd = found._dataStart;
2127 found._dataEnd = endpos;
2129 if (isPatternString) {
2130 keys[key].used = true;
2133 entries_.push_back(found);
2137 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2139 stringstream s(keysstring);
2141 const char delim = '|';
2142 while (getline(s, key, delim)) {
2143 KeyInfo keyII(keyI);
2144 if (isPatternString) {
2147 else if ( !keys[key].used)
2148 keyII.disabled = true;
2153 void LatexInfo::buildKeys(bool isPatternString)
2156 static bool keysBuilt = false;
2157 if (keysBuilt && !isPatternString) return;
2159 // Keys to ignore in any case
2160 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2161 // Known standard keys with 1 parameter.
2162 // Split is done, if not at start of region
2163 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2164 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2165 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2166 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2167 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2168 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2170 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2171 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2172 makeKey("section*|subsection*|subsubsection*|paragraph*",
2173 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2174 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2175 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2177 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2179 // Split is done, if not at start of region
2180 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2181 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2183 // Split is done always.
2184 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2188 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2189 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2190 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2191 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2193 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2194 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2195 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2196 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2198 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2199 // Custom space/skip, remove the content (== length value)
2200 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2201 // Found in fr/UserGuide.lyx
2202 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2204 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2205 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2206 // Known macros to remove (including their parameter)
2208 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2209 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2210 // handle like standard keys with 1 parameter.
2211 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2213 // Ignore deleted text
2214 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2215 // but preserve added text
2216 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2218 // Macros to remove, but let the parameter survive
2220 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2222 // Remove language spec from content of these insets
2223 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2225 // Same effect as previous, parameter will survive (because there is no one anyway)
2227 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2228 // Remove table decorations
2229 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2230 // Discard shape-header.
2231 // For footnote or shortcut too, because of lang settings
2232 // and wrong handling if used 'KeyInfo::noMain'
2233 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2234 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2235 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2236 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2237 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2238 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2239 // like ('tiny{}' or '\tiny ' ... )
2240 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2242 // Survives, like known character
2243 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2244 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2245 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2247 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2248 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2249 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2251 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2252 // Remove RTL/LTR marker
2253 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2254 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2255 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2256 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2257 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2258 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2259 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2260 if (isPatternString) {
2261 // Allow the first searched string to rebuild the keys too
2265 // no need to rebuild again
2271 * Keep the list of actual opened parentheses actual
2272 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2274 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2277 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2281 if (c == '\\') skip = 1;
2282 else if (c == '{') {
2285 else if (c == '}') {
2286 handleCloseP(i, closingAllowed);
2292 string Intervall::show(int lastpos)
2294 int idx = 0; /* int intervalls */
2297 for (idx = 0; idx <= ignoreidx; idx++) {
2298 while (i < lastpos) {
2300 if (i <= borders[idx].low) {
2301 if (borders[idx].low > lastpos)
2302 printsize = lastpos - i;
2304 printsize = borders[idx].low - i;
2305 s += par.substr(i, printsize);
2307 if (i >= borders[idx].low)
2308 i = borders[idx].upper;
2311 i = borders[idx].upper;
2317 s += par.substr(i, lastpos-i);
2323 void Intervall::output(ostringstream &os, int lastpos)
2325 // get number of chars to output
2326 int idx = 0; /* int intervalls */
2329 string startTitle = titleValue;
2330 for (idx = 0; idx <= ignoreidx; idx++) {
2332 if (i <= borders[idx].low) {
2334 if (borders[idx].low > lastpos)
2335 printsize = lastpos - i;
2337 printsize = borders[idx].low - i;
2338 if (printsize > 0) {
2339 os << startTitle << par.substr(i, printsize);
2341 printed += printsize;
2344 handleParentheses(i, false);
2345 if (i >= borders[idx].low)
2346 i = borders[idx].upper;
2349 i = borders[idx].upper;
2356 os << startTitle << par.substr(i, lastpos-i);
2357 printed += lastpos-i;
2359 handleParentheses(lastpos, false);
2361 if (keys["foreignlanguage"].disabled)
2362 startindex = actualdeptindex-langcount;
2364 startindex = actualdeptindex;
2365 for (int i = startindex; i > 0; --i) {
2368 if (hasTitle && (printed > 0))
2370 if (! isPatternString_)
2372 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2375 void LatexInfo::processRegion(int start, int region_end)
2377 while (start < region_end) { /* Let {[} and {]} survive */
2378 int cnt = interval_.isOpeningPar(start);
2380 // Closing is allowed past the region
2381 int closing = interval_.findclosing(start+1, interval_.par.length());
2382 interval_.addIntervall(start, start+1);
2383 interval_.addIntervall(closing, closing+1);
2387 start = interval_.nextNotIgnored(start+1);
2391 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2393 if (actual.parenthesiscount == 0) {
2394 // "{\tiny{} ...}" ==> "{{} ...}"
2395 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2398 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2399 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2400 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2404 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2407 switch (actual.keytype)
2409 case KeyInfo::isTitle: {
2411 nextKeyIdx = getNextKey();
2414 case KeyInfo::cleanToStart: {
2415 actual._dataEnd = actual._dataStart;
2416 nextKeyIdx = getNextKey();
2417 // Search for end of arguments
2418 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2420 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2421 entries_[i].disabled = true;
2423 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2425 while (interval_.par[actual._dataEnd] == ' ')
2427 interval_.addIntervall(0, actual._dataEnd+1);
2428 interval_.actualdeptindex = 0;
2429 interval_.depts[0] = actual._dataEnd+1;
2430 interval_.closes[0] = -1;
2433 case KeyInfo::isText:
2434 interval_.par[actual._tokenstart] = '#';
2435 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2436 nextKeyIdx = getNextKey();
2438 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2439 if (actual.disabled)
2440 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2442 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2445 case KeyInfo::isChar: {
2446 nextKeyIdx = getNextKey();
2449 case KeyInfo::isSize: {
2450 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2451 if (actual.parenthesiscount == 0)
2452 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2454 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2456 nextKeyIdx = getNextKey();
2458 // Here _dataStart points to '{', so correct it
2459 actual._dataStart += 1;
2460 actual._tokensize += 1;
2461 actual.parenthesiscount = 1;
2462 if (interval_.par[actual._dataStart] == '}') {
2463 // Determine the end if used like '{\tiny{}...}'
2464 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2465 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2468 // Determine the end if used like '\tiny{...}'
2469 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2471 // Split on this key if not at start
2472 int start = interval_.nextNotIgnored(previousStart);
2473 if (start < actual._tokenstart) {
2474 interval_.output(os, actual._tokenstart);
2475 interval_.addIntervall(start, actual._tokenstart);
2477 // discard entry if at end of actual
2478 nextKeyIdx = process(os, actual);
2482 case KeyInfo::endArguments: {
2483 // Remove trailing '{}' too
2484 actual._dataStart += 1;
2485 actual._dataEnd += 1;
2486 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2487 nextKeyIdx = getNextKey();
2490 case KeyInfo::noMain:
2492 case KeyInfo::isStandard: {
2493 if (actual.disabled) {
2495 processRegion(actual._dataStart, actual._dataStart+1);
2496 nextKeyIdx = getNextKey();
2498 // Split on this key if not at datastart of calling entry
2499 int start = interval_.nextNotIgnored(previousStart);
2500 if (start < actual._tokenstart) {
2501 interval_.output(os, actual._tokenstart);
2502 interval_.addIntervall(start, actual._tokenstart);
2504 // discard entry if at end of actual
2505 nextKeyIdx = process(os, actual);
2509 case KeyInfo::removeWithArg: {
2510 nextKeyIdx = getNextKey();
2511 // Search for end of arguments
2512 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2514 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2515 entries_[i].disabled = true;
2517 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2519 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2522 case KeyInfo::doRemove: {
2523 // Remove the key with all parameters and following spaces
2525 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2526 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2529 // Remove also enclosing parentheses [] and {}
2532 while (actual._tokenstart > numpars) {
2533 if (pos+numpars >= interval_.par.size())
2535 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2537 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2543 if (interval_.par[pos+numpars] == ' ')
2547 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2548 nextKeyIdx = getNextKey();
2551 case KeyInfo::isList: {
2552 // Discard space before _tokenstart
2554 for (count = 0; count < actual._tokenstart; count++) {
2555 if (interval_.par[actual._tokenstart-count-1] != ' ')
2558 nextKeyIdx = getNextKey();
2559 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2561 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2563 // How else can we catch this one?
2564 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2565 entries_[i].disabled = true;
2567 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2569 else if (nextKeyIdx > 0) {
2570 // Ignore any lang entries inside data region
2571 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2572 if (entries_[i].keytype == KeyInfo::isMain)
2573 entries_[i].disabled = true;
2576 if (actual.disabled) {
2577 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2580 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2582 if (interval_.par[actual._dataEnd+1] == '[') {
2583 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2584 if ((interval_.par[actual._dataEnd+2] == '{') &&
2585 (interval_.par[posdown-1] == '}')) {
2586 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2587 interval_.addIntervall(posdown-1, posdown+1);
2590 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2591 interval_.addIntervall(posdown, posdown+1);
2593 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2594 if (blk > posdown) {
2595 // Discard at most 1 space after empty item
2597 for (count = 0; count < 1; count++) {
2598 if (interval_.par[blk+count] != ' ')
2602 interval_.addIntervall(blk, blk+count);
2607 case KeyInfo::isSectioning: {
2608 // Discard spaces before _tokenstart
2610 int val = actual._tokenstart;
2611 for (count = 0; count < actual._tokenstart;) {
2612 val = interval_.previousNotIgnored(val-1);
2613 if (val < 0 || interval_.par[val] != ' ')
2616 count = actual._tokenstart - val;
2619 if (actual.disabled) {
2620 removeHead(actual, count);
2621 nextKeyIdx = getNextKey();
2623 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2624 nextKeyIdx = process(os, actual);
2628 case KeyInfo::isMath: {
2629 // Same as regex, use the content unchanged
2630 nextKeyIdx = getNextKey();
2633 case KeyInfo::isRegex: {
2634 // DO NOT SPLIT ON REGEX
2636 nextKeyIdx = getNextKey();
2639 case KeyInfo::isIgnored: {
2640 // Treat like a character for now
2641 nextKeyIdx = getNextKey();
2644 case KeyInfo::isMain: {
2645 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2646 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2647 if (actual._tokenstart > 0) {
2648 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2649 if ((prev >= 0) && interval_.par[prev] == '%')
2650 interval_.addIntervall(prev, prev+1);
2652 if (actual.disabled) {
2654 interval_.langcount++;
2655 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2656 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2657 // Discard also the space before math-equation
2658 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2660 nextKeyIdx = getNextKey();
2661 // interval.resetOpenedP(actual._dataStart-1);
2664 if (actual._tokenstart < 26) {
2665 // for the first (and maybe dummy) language
2666 interval_.setForDefaultLang(actual);
2668 interval_.resetOpenedP(actual._dataStart-1);
2672 case KeyInfo::invalid:
2673 case KeyInfo::headRemove:
2674 // These two cases cannot happen, already handled
2677 // LYXERR(Debug::INFO, "Unhandled keytype");
2678 nextKeyIdx = getNextKey();
2685 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2687 int end = interval_.nextNotIgnored(actual._dataEnd);
2688 int oldStart = actual._dataStart;
2689 int nextKeyIdx = getNextKey();
2691 if ((nextKeyIdx < 0) ||
2692 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2693 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2694 if (oldStart <= end) {
2695 processRegion(oldStart, end);
2700 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2702 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2703 (void) dispatch(os, actual._dataStart, nextKey);
2704 end = nextKey._tokenstart;
2707 processRegion(oldStart, nextKey._tokenstart);
2708 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2710 oldStart = nextKey._dataEnd+1;
2712 // now nextKey is either invalid or is outside of actual._dataEnd
2713 // output the remaining and discard myself
2714 if (oldStart <= end) {
2715 processRegion(oldStart, end);
2717 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2719 // This is the normal case.
2720 // But if using the firstlanguage, the closing may be missing
2722 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2724 if (actual._dataEnd < end)
2725 output_end = interval_.nextNotIgnored(actual._dataEnd);
2726 else if (interval_.par.size() > (size_t) end)
2727 output_end = interval_.nextNotIgnored(end);
2729 output_end = interval_.par.size();
2730 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2731 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2733 // Remove possible empty data
2734 int dstart = interval_.nextNotIgnored(actual._dataStart);
2735 while (interval_.isOpeningPar(dstart) == 1) {
2736 interval_.addIntervall(dstart, dstart+1);
2737 int dend = interval_.findclosing(dstart+1, output_end);
2738 interval_.addIntervall(dend, dend+1);
2739 dstart = interval_.nextNotIgnored(dstart+1);
2741 if (dstart < output_end)
2742 interval_.output(os, output_end);
2743 interval_.addIntervall(actual._tokenstart, end);
2747 string splitOnKnownMacros(string par, bool isPatternString)
2750 LatexInfo li(par, isPatternString);
2751 // LYXERR(Debug::INFO, "Berfore split: " << par);
2752 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2754 DummyKey._tokensize = 0;
2755 DummyKey._dataStart = 0;
2756 DummyKey._dataEnd = par.length();
2757 DummyKey.disabled = true;
2758 int firstkeyIdx = li.getFirstKey();
2760 if (firstkeyIdx >= 0) {
2761 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2762 DummyKey._tokenstart = firstKey._tokenstart;
2764 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2765 // Use dummy firstKey
2766 firstKey = DummyKey;
2767 (void) li.setNextKey(firstkeyIdx);
2770 if (par.substr(firstKey._dataStart, 2) == "% ")
2771 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2773 nextkeyIdx = li.process(os, firstKey);
2774 while (nextkeyIdx >= 0) {
2775 // Check for a possible gap between the last
2776 // entry and this one
2777 int datastart = li.nextNotIgnored(firstKey._dataStart);
2778 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2779 if ((nextKey._tokenstart > datastart)) {
2781 firstKey._dataStart = datastart;
2782 firstKey._dataEnd = par.length();
2783 (void) li.setNextKey(nextkeyIdx);
2784 // Fake the last opened parenthesis
2785 li.setForDefaultLang(firstKey);
2786 nextkeyIdx = li.process(os, firstKey);
2789 if (nextKey.keytype != KeyInfo::isMain) {
2790 firstKey._dataStart = datastart;
2791 firstKey._dataEnd = nextKey._dataEnd+1;
2792 (void) li.setNextKey(nextkeyIdx);
2793 li.setForDefaultLang(firstKey);
2794 nextkeyIdx = li.process(os, firstKey);
2797 nextkeyIdx = li.process(os, nextKey);
2801 // Handle the remaining
2802 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2803 firstKey._dataEnd = par.length();
2805 if ((firstKey._dataStart < firstKey._dataEnd) &&
2806 (par[firstKey._dataStart] != '}')) {
2807 li.setForDefaultLang(firstKey);
2808 (void) li.process(os, firstKey);
2812 // return string definitelly impossible to match
2813 s = "\\foreignlanguage{ignore}{ }";
2817 s = par; /* no known macros found */
2818 // LYXERR(Debug::INFO, "After split: " << s);
2823 * Try to unify the language specs in the latexified text.
2824 * Resulting modified string is set to "", if
2825 * the searched tex does not contain all the features in the search pattern
2827 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2829 static Features regex_f;
2830 static int missed = 0;
2831 static bool regex_with_format = false;
2833 int parlen = par.length();
2835 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2838 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2839 // Happens to be there in case of description or labeling environment
2844 // Split the latex input into pieces which
2845 // can be digested by our search engine
2846 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2847 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2848 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2851 result = par.substr(0, parlen);
2852 if (isPatternString) {
2855 regex_f = identifyFeatures(result);
2856 string features = "";
2857 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2858 string a = it->first;
2859 regex_with_format = true;
2860 features += " " + a;
2861 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2863 LYXERR(Debug::FIND, "Identified Features" << features);
2866 } else if (regex_with_format) {
2867 Features info = identifyFeatures(result);
2868 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2869 string a = it->first;
2870 bool b = it->second;
2871 if (b && ! info[a]) {
2873 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2879 // LYXERR(Debug::INFO, "No regex formats");
2885 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2886 static int identifyClosing(string & t)
2888 int open_braces = 0;
2890 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2891 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2893 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2895 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2897 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2906 static int num_replaced = 0;
2907 static bool previous_single_replace = true;
2909 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
2912 // Handle \w properly
2913 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
2914 if (! opt.casesensitive) {
2915 popts |= QRegularExpression::CaseInsensitiveOption;
2917 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
2918 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
2920 if (regexp.isValid() && regexp2.isValid()) {
2921 regexIsValid = true;
2922 // Check '{', '}' pairs inside the regex
2925 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
2926 char c = par_as_string[i];
2933 else if (c == '}') {
2940 if (balanced != 0) {
2941 regexIsValid = false;
2942 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
2946 regexIsValid = false;
2947 if (!regexp.isValid())
2948 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
2950 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
2953 if (opt.casesensitive) {
2954 regexp = regex(regexp_str);
2955 regexp2 = regex(regexp2_str);
2958 regexp = regex(regexp_str, std::regex_constants::icase);
2959 regexp2 = regex(regexp2_str, std::regex_constants::icase);
2964 static void modifyRegexForMatchWord(string &t)
2967 regex wordre("(\\\\)*((\\.|\\\\b))");
2970 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
2972 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
2975 else if (sub.str(2) == "\\\\b")
2977 if (lastpos < (size_t) sub.position(2))
2978 s += t.substr(lastpos, sub.position(2) - lastpos);
2980 lastpos = sub.position(2) + sub.length(2);
2984 else if (lastpos < t.length())
2985 s += t.substr(lastpos, t.length() - lastpos);
2986 t = "\\b" + s + "\\b";
2989 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
2990 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2992 static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\\])" };
2993 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2994 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2995 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2996 if (opt.replace_all && previous_single_replace) {
2997 previous_single_replace = false;
3000 else if (!opt.replace_all) {
3001 num_replaced = 0; // count number of replaced strings
3002 previous_single_replace = true;
3004 // When using regexp, braces are hacked already by escape_for_regex()
3005 par_as_string = normalize(ds);
3007 close_wildcards = 0;
3009 size_t lead_size = 0;
3010 // correct the language settings
3011 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
3013 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
3014 if (opt.ignoreformat) {
3018 lead_size = identifyLeading(par_as_string);
3020 lead_as_string = par_as_string.substr(0, lead_size);
3021 string lead_as_regex_string = std::regex_replace(lead_as_string, specialChars, R"(\$&)" );
3022 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3023 string par_as_regex_string_nolead = std::regex_replace(par_as_string_nolead, specialChars, R"(\$&)" );
3024 /* Handle whole words too in this case
3026 if (opt.matchword) {
3027 par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3028 opt.matchword = false;
3030 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3031 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3032 CreateRegexp(opt, regexp_str, regexp2_str);
3037 if (!opt.ignoreformat) {
3038 lead_size = identifyLeading(par_as_string);
3039 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
3040 lead_as_string = par_as_string.substr(0, lead_size);
3041 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3044 // Here we are using regexp
3045 LASSERT(use_regexp, /**/);
3047 string lead_as_regexp;
3048 if (lead_size > 0) {
3049 lead_as_regexp = std::regex_replace(par_as_string.substr(0, lead_size), specialChars, R"(\$&)" );
3050 par_as_string = par_as_string_nolead;
3051 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
3052 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3054 LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3055 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3056 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3057 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3059 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
3060 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
3061 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
3062 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
3063 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
3064 || regex_replace(par_as_string, par_as_string,
3065 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
3066 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
3067 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
3071 if (!opt.ignoreformat) {
3072 // Remove extra '\}' at end if not part of \{\.\}
3073 size_t lng = par_as_string.size();
3075 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3077 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3086 if (lng < par_as_string.size())
3087 par_as_string = par_as_string.substr(0,lng);
3088 if ((lng > 0) && (par_as_string[0] == '^')) {
3089 par_as_string = par_as_string.substr(1);
3091 opt.matchstart = true;
3094 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3095 LYXERR(Debug::FIND, "Open braces: " << open_braces);
3096 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
3097 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3099 // If entered regexp must match at begin of searched string buffer
3100 // Kornel: Added parentheses to use $1 for size of the leading string
3104 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3105 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3106 // so the convert has no effect in that case
3107 for (int i = 7; i > 0; --i) {
3108 string orig = "\\\\" + std::to_string(i);
3109 string dest = "\\" + std::to_string(i+2);
3110 while (regex_replace(par_as_string, par_as_string, orig, dest));
3112 /* opt.matchword is ignored if using regex
3113 so expanding par_as_string with "\\b" seems appropriate here
3114 if regex contains for instance '.*' or '.+'
3115 1.) Nothing to do, if 'par_as_string' contains "\\b" already.
3116 (Means, that the user knows how to handle whole words
3117 2.) else replace '.' with "\\S" and wrap the regex with "\\b"
3119 if (opt.matchword) {
3120 modifyRegexForMatchWord(par_as_string);
3121 opt.matchword = false;
3123 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3124 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3126 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3127 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3128 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3133 // Count number of characters in string
3137 // \\[a-zA-Z]+ ==> 1
3139 static int computeSize(QStringRef s, int len)
3140 #define isLyxAlpha(arg) arg.isLetter()
3142 static int computeSize(string s, int len)
3143 #define isLyxAlpha(arg) isalpha(arg)
3150 for (int i = 0; i < len; i += skip, count++) {
3151 if (s.at(i) == '\\') {
3153 if (i + 1 < len && isLyxAlpha(s.at(i+1))) {
3154 for (int j = 2; i+j < len; j++) {
3155 if (! isLyxAlpha(s.at(i+j))) {
3156 if (s.at(i+j) == ' ')
3158 else if (s.at(i+j) == '{') {
3159 if (i+j+1 < len && s.at(i+j+1) == '}')
3161 else if (i + j + 1 >= len)
3170 else if (s.at(i) == '{') {
3171 if (i + 1 < len && s.at(i+1) == '}')
3176 else if (s.at(i) == '-') {
3177 if (i+1 < len && s.at(i+1) == '-') {
3178 if (i + 2 < len && s.at(i+2) == '-')
3194 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3199 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3202 docstring docstr = stringifyFromForSearch(opt, cur, len);
3204 if (use_regexp || opt.casesensitive)
3205 str = normalize(docstr);
3207 str = normalize(lowercase(docstr));
3208 if (!opt.ignoreformat) {
3209 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3212 mres.match_len = -1;
3215 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
3216 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
3219 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3221 QString qstr = QString::fromStdString(str);
3222 QRegularExpression const *p_regexp;
3223 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3227 p_regexp = ®exp2;
3229 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3230 if (!match.hasMatch())
3233 regex const *p_regexp;
3234 regex_constants::match_flag_type flags;
3236 flags = regex_constants::match_continuous;
3239 flags = regex_constants::match_default;
3240 p_regexp = ®exp2;
3242 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3243 if (re_it == sregex_iterator())
3245 match_results<string::const_iterator> const & m = *re_it;
3247 // Whole found string, including the leading
3248 // std: m[0].second - m[0].first
3249 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3251 // Size of the leading string
3252 // std: m[1].second - m[1].first
3253 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3254 int leadingsize = 0;
3256 if (match.lastCapturedIndex() > 0) {
3257 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3262 leadingsize = m[1].second - m[1].first;
3266 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3267 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3268 // because of different number of closing at end of string
3269 // we have to 'unify' the length of the post-match.
3270 // Done by ignoring closing parenthesis and linefeeds at string end
3271 int matchend = match.capturedEnd(0);
3272 while (mres.match_len > 0) {
3273 QChar c = qstr.at(matchend - 1);
3274 if ((c == '\n') || (c == '}') || (c == '{')) {
3281 size_t strsize = qstr.size();
3282 while (strsize > (size_t) match.capturedEnd(0)) {
3283 QChar c = qstr.at(strsize-1);
3284 if ((c == '\n') || (c == '}')) {
3290 // LYXERR0(qstr.toStdString());
3291 mres.match2end = strsize - matchend;
3292 mres.pos = match.capturedStart(2);
3294 mres.match_prefix = m[2].second - m[2].first;
3295 mres.match_len = m[0].second - m[2].second;
3296 // ignore closing parenthesis and linefeeds at string end
3297 size_t strend = m[0].second - m[0].first;
3298 int matchend = strend;
3299 while (mres.match_len > 0) {
3300 char c = str.at(matchend - 1);
3301 if ((c == '\n') || (c == '}') || (c == '{')) {
3308 size_t strsize = str.size();
3309 while (strsize > strend) {
3310 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3317 mres.match2end = strsize - matchend;
3318 mres.pos = m[2].first - m[0].first;;
3320 if (mres.match2end < 0)
3322 mres.leadsize = leadingsize;
3324 if (mres.match_len > 0) {
3325 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3326 mres.result.push_back(a0);
3327 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3328 mres.result.push_back(match.captured(i).toStdString());
3332 if (mres.match_len > 0) {
3333 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3334 mres.result.push_back(a0);
3335 for (size_t i = 3; i < m.size(); i++) {
3336 mres.result.push_back(m[i]);
3343 // else !use_regexp: but all code paths above return
3344 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
3345 << par_as_string << "', str='" << str << "'");
3346 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
3347 << lead_as_string << "', par_as_string_nolead='"
3348 << par_as_string_nolead << "'");
3351 LYXERR(Debug::FIND, "size=" << par_as_string.size()
3352 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
3353 if (str.substr(0, par_as_string.size()) == par_as_string) {
3354 mres.match_len = par_as_string.size();
3355 mres.match2end = str.size();
3360 // Start the search _after_ the leading part
3361 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3362 if (pos != string::npos) {
3363 mres.match_len = par_as_string.size();
3364 mres.match2end = str.size() - pos;
3373 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3375 MatchResult mres = findAux(cur, len, at_begin);
3376 int res = mres.match_len;
3378 "res=" << res << ", at_begin=" << at_begin
3379 << ", matchstart=" << opt.matchstart
3380 << ", inTexted=" << cur.inTexted());
3381 if (opt.matchstart) {
3384 else if (mres.match_prefix > 0)
3390 /* DEAD CODE follows
3391 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3393 if ((len > 0) && (res < len)) {
3397 Paragraph const & par = cur.paragraph();
3398 bool ws_left = (cur.pos() > 0)
3399 ? par.isWordSeparator(cur.pos() - 1)
3405 ws_right = (cur.pos() + len < par.size())
3406 ? par.isWordSeparator(cur.pos() + len)
3410 "cur.pos()=" << cur.pos() << ", res=" << res
3411 << ", separ: " << ws_left << ", " << ws_right
3414 if (ws_left && ws_right) {
3415 // Check for word separators inside the found 'word'
3416 for (int i = 0; i < len; i++) {
3417 if (par.isWordSeparator(cur.pos() + i)) {
3430 static bool simple_replace(string &t, string from, string to)
3432 regex repl("(\\\\)*(" + from + ")");
3436 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3438 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3440 if (lastpos < (size_t) sub.position(2))
3441 s += t.substr(lastpos, sub.position(2) - lastpos);
3443 lastpos = sub.position(2) + sub.length(2);
3447 else if (lastpos < t.length())
3448 s += t.substr(lastpos, t.length() - lastpos);
3454 string MatchStringAdv::normalize(docstring const & s) const
3457 t = lyx::to_utf8(s);
3458 // Remove \n at begin
3459 while (!t.empty() && t[0] == '\n')
3462 while (!t.empty() && t[t.size() - 1] == '\n')
3463 t = t.substr(0, t.size() - 1);
3465 // Handle all other '\n'
3466 while ((pos = t.find("\n")) != string::npos) {
3467 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3469 if (isAlnumASCII(t[pos+1])) {
3470 t.replace(pos-2, 3, " ");
3473 t.replace(pos-2, 3, "");
3476 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3477 // '\n' adjacent to non-alpha-numerics, discard
3478 t.replace(pos, 1, "");
3481 // Replace all other \n with spaces
3482 t.replace(pos, 1, " ");
3485 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3486 // Kornel: Added textsl, textsf, textit, texttt and noun
3487 // + allow to seach for colored text too
3488 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3489 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3490 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3491 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3492 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3493 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3499 docstring stringifyFromCursor(DocIterator const & cur, int len)
3501 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3502 if (cur.inTexted()) {
3503 Paragraph const & par = cur.paragraph();
3504 // TODO what about searching beyond/across paragraph breaks ?
3505 // TODO Try adding a AS_STR_INSERTS as last arg
3506 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3507 int(par.size()) : cur.pos() + len;
3508 // OutputParams runparams(&cur.buffer()->params().encoding());
3509 OutputParams runparams(encodings.fromLyXName("utf8"));
3510 runparams.nice = true;
3511 runparams.flavor = Flavor::XeTeX;
3512 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3513 // No side effect of file copying and image conversion
3514 runparams.dryrun = true;
3515 runparams.for_search = true;
3516 LYXERR(Debug::FIND, "Stringifying with cur: "
3517 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3518 return par.asString(cur.pos(), end,
3519 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3521 } else if (cur.inMathed()) {
3522 CursorSlice cs = cur.top();
3523 MathData md = cs.cell();
3524 MathData::const_iterator it_end =
3525 (( len == -1 || cs.pos() + len > int(md.size()))
3527 : md.begin() + cs.pos() + len );
3529 for (MathData::const_iterator it = md.begin() + cs.pos();
3532 docstring s = asString(md2);
3533 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3536 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3541 /** Computes the LaTeX export of buf starting from cur and ending len positions
3542 * after cur, if len is positive, or at the paragraph or innermost inset end
3545 docstring latexifyFromCursor(DocIterator const & cur, int len)
3547 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3548 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3549 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3550 Buffer const & buf = *cur.buffer();
3552 odocstringstream ods;
3554 //OutputParams runparams(&buf.params().encoding());
3555 OutputParams runparams(encodings.fromLyXName("utf8"));
3556 runparams.nice = false;
3557 runparams.flavor = Flavor::XeTeX;
3558 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3559 // No side effect of file copying and image conversion
3560 runparams.dryrun = true;
3561 runparams.for_search = true;
3563 if (cur.inTexted()) {
3564 // @TODO what about searching beyond/across paragraph breaks ?
3565 pos_type endpos = cur.paragraph().size();
3566 if (len != -1 && endpos > cur.pos() + len)
3567 endpos = cur.pos() + len;
3568 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3569 string(), cur.pos(), endpos);
3570 string s = lyx::to_utf8(ods.str());
3571 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3572 return(lyx::from_utf8(s));
3573 } else if (cur.inMathed()) {
3574 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3575 for (int s = cur.depth() - 1; s >= 0; --s) {
3576 CursorSlice const & cs = cur[s];
3577 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3578 TeXMathStream ws(os);
3579 cs.asInsetMath()->asHullInset()->header_write(ws);
3584 CursorSlice const & cs = cur.top();
3585 MathData md = cs.cell();
3586 MathData::const_iterator it_end =
3587 ((len == -1 || cs.pos() + len > int(md.size()))
3589 : md.begin() + cs.pos() + len);
3591 for (MathData::const_iterator it = md.begin() + cs.pos();
3595 ods << asString(md2);
3596 // Retrieve the math environment type, and add '$' or '$]'
3597 // or others (\end{equation}) accordingly
3598 for (int s = cur.depth() - 1; s >= 0; --s) {
3599 CursorSlice const & cs2 = cur[s];
3600 InsetMath * inset = cs2.asInsetMath();
3601 if (inset && inset->asHullInset()) {
3602 TeXMathStream ws(os);
3603 inset->asHullInset()->footer_write(ws);
3607 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3609 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3614 #if defined(ResultsDebug)
3616 static void displayMResult(MatchResult &mres, int increment)
3618 LYXERR0( "pos: " << mres.pos << " increment " << increment);
3619 LYXERR0( "leadsize: " << mres.leadsize);
3620 LYXERR0( "match_len: " << mres.match_len);
3621 LYXERR0( "match_prefix: " << mres.match_prefix);
3622 LYXERR0( "match2end: " << mres.match2end);
3623 LYXERR0( "pos_len: " << mres.pos_len); // Set in finalize
3624 for (size_t i = 0; i < mres.result.size(); i++)
3625 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3627 #define displayMres(s,i) displayMResult(s,i);
3629 #define displayMres(s,i)
3632 static bool findAdvForwardInnermost(DocIterator & cur)
3635 DocIterator old_cur(cur.buffer());
3636 int forwardCount = 0;
3644 if (cur.depth() > d) {
3648 if (cur.depth() == d)
3652 if (forwardCount > 0) {
3653 LYXERR(Debug::FIND, "Forwarded " << forwardCount << " step(s) (searching for innermost match)");
3660 /** Finalize an advanced find operation, advancing the cursor to the innermost
3661 ** position that matches, plus computing the length of the matching text to
3663 ** Return the cur.pos() difference between start and end of found match
3665 MatchResult &findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
3667 // Search the foremost position that matches (avoids find of entire math
3668 // inset when match at start of it)
3669 DocIterator old_cur(cur.buffer());
3671 static MatchResult fail = MatchResult();
3672 static MatchResult max_match;
3673 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3674 // Happens with e.g. hyperlinks
3675 // either one sees "http://www.bla.bla" or nothing
3676 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3677 // and although we search for only 3 chars, we find the whole hyperlink inset
3678 bool at_begin = (expected.match_prefix == 0);
3679 if (findAdvForwardInnermost(cur)) {
3680 mres = match(cur, -1, at_begin);
3681 displayMres(mres, 0);
3682 if (expected.match_len > 0) {
3683 if (mres.match_len < expected.match_len)
3687 if (mres.match_len <= 0)
3690 max_match = mres.match_len;
3692 else if (expected.match_len < 0) {
3693 mres = match(cur); /* match valid only if not searching whole words */
3694 displayMres(mres, 0);
3698 max_match = expected;
3700 if (max_match.match_len <= 0) return fail;
3701 LYXERR(Debug::FIND, "Ok");
3703 // Compute the match length
3705 if (cur.pos() + len > cur.lastpos())
3707 // regexp should use \w+, \S+, or \b(some string)\b
3708 // to search for whole words
3709 if (match.opt.matchword && !match.use_regexp) {
3710 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3711 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3713 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3715 // Length of matched text (different from len param)
3716 static MatchResult old_match = match(cur, len, at_begin);
3717 if (old_match.match_len < 0)
3719 MatchResult new_match;
3720 // Greedy behaviour while matching regexps
3721 while ((new_match = match(cur, len + 1, at_begin)).match_len > old_match.match_len) {
3723 old_match = new_match;
3724 LYXERR(Debug::FIND, "verifying match with len = " << len);
3730 int maxl = cur.lastpos() - cur.pos();
3731 // Greedy behaviour while matching regexps
3732 while (maxl > minl) {
3734 mres2 = match(cur, len, at_begin);
3735 displayMres(mres2, len);
3736 int actual_match = mres2.match_len;
3737 if (actual_match >= max_match.match_len) {
3738 // actual_match > max_match _can_ happen,
3739 // if the search area splits
3740 // some following word so that the regex
3741 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3743 // This means, the len value is too big
3745 if (maxl - minl < 4)
3746 len = (int)((maxl + minl)/2);
3748 len = (int)(minl + (maxl - minl + 3)/4);
3751 // (actual_match < max_match)
3753 len = (int)((maxl + minl)/2);
3757 // Search for real start of matched characters
3759 MatchResult actual_match;
3762 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3763 if (cur.depth() < old_cur.depth()) {
3765 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3768 if (cur.pos() != old_cur.pos()) {
3769 // OK, forwarded 1 pos in actual inset
3770 actual_match = match(cur, len-1, at_begin);
3771 if (actual_match.match_len == max_match.match_len) {
3772 // Ha, got it! The shorter selection has the same match length
3777 // OK, the shorter selection matches less chars, revert to previous value
3783 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3784 actual_match = match(cur, len, at_begin);
3785 if (actual_match.match_len == max_match.match_len) {
3787 max_match = actual_match;
3794 max_match.pos_len = len;
3801 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3805 while (!theApp()->longOperationCancelled() && cur) {
3806 (void) findAdvForwardInnermost(cur);
3807 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3808 MatchResult mres = match(cur, -1, false);
3809 displayMres(mres,-1)
3810 int match_len = mres.match_len;
3811 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3812 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3815 if (match_len > 0) {
3816 // Try to find the begin of searched string
3818 int firstInvalid = 100000;
3819 if (mres.match_prefix + mres.pos - mres.leadsize > 1)
3820 increment = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3823 LYXERR(Debug::FIND, "Set increment to " << increment);
3824 while (increment > 0) {
3825 DocIterator old_cur = cur;
3826 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3828 if (! cur || (cur.pit() > old_cur.pit())) {
3829 // Are we outside of the paragraph?
3830 // This can happen if moving past some UTF8-encoded chars
3835 MatchResult mres2 = match(cur, -1, false);
3836 displayMres(mres2,increment)
3837 switch (interpretMatch(mres, mres2)) {
3838 case MatchResult::newIsTooFar:
3839 // behind the expected match
3840 firstInvalid = increment;
3844 case MatchResult::newIsBetter:
3845 // not reached ye, but cur.pos()+increment is bettert
3847 firstInvalid -= increment;
3848 if (increment > firstInvalid*3/4)
3849 increment = firstInvalid*3/4;
3850 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3851 if (increment >= mres2.match_prefix)
3852 increment = (mres2.match_prefix+1)*3/4;
3857 // Handle not like MatchResult::newIsTooFar
3858 // LYXERR0( "Something is wrong: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3860 increment = increment*3/4;
3866 // LYXERR0("Leaving first loop");
3868 LYXERR(Debug::FIND, "Finalizing 1");
3869 MatchResult found_match = findAdvFinalize(cur, match, mres);
3870 if (found_match.match_len > 0) {
3871 LASSERT(found_match.pos_len > 0, /**/);
3872 match.FillResults(found_match);
3873 return found_match.pos_len;
3876 // try next possible match
3881 // The following code is newer reached
3882 // but parts of it may be needed in future
3883 int match_len_zero_count = 0;
3885 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3887 mres3 = match(cur, -1, false);
3888 displayMres(mres3, 1)
3889 int remaining_len = mres3.match_len;
3890 if (remaining_len <= 0) {
3891 // Apparently the searched string is not in the remaining part
3898 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3899 mres3 = match(cur, 1);
3900 displayMres(mres3, 1)
3901 int match_len3 = mres3.match_len;
3905 displayMres(mres3, 1)
3906 int match_len2 = mres3.match_len;
3907 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3908 if (match_len2 > 0) {
3909 // Sometimes in finalize we understand it wasn't a match
3910 // and we need to continue the outest loop
3911 LYXERR(Debug::FIND, "Finalizing 2");
3912 MatchResult mres4 = findAdvFinalize(cur, match, mres.match_len);
3913 if (mres4.match_len > 0) {
3914 match.FillResults(mres4);
3915 LASSERT(mres4.pos_len > 0, /**/);
3916 return mres4.pos_len;
3920 match_len_zero_count = 0;
3921 else if (match_len2 == 0)
3922 match_len_zero_count++;
3923 if (match_len2 < 0) {
3924 if (++match_len_zero_count > 3) {
3925 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3933 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3934 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3937 // This should exit nested insets, if any, or otherwise undefine the currsor.
3938 cur.pos() = cur.lastpos();
3939 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3947 /// Find the most backward consecutive match within same paragraph while searching backwards.
3948 MatchResult &findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3950 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3951 DocIterator tmp_cur = cur;
3952 static MatchResult mr = findAdvFinalize(tmp_cur, match, MatchResult(-1));
3953 Inset & inset = cur.inset();
3954 for (; cur != cur_begin; cur.backwardPos()) {
3955 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3956 DocIterator new_cur = cur;
3957 new_cur.backwardPos();
3958 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3960 MatchResult new_mr = findAdvFinalize(new_cur, match, MatchResult(-1));
3961 if (new_mr.match_len == mr.match_len)
3965 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3971 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3975 // Backup of original position
3976 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3977 if (cur == cur_begin)
3980 DocIterator cur_orig(cur);
3981 bool pit_changed = false;
3984 bool found_match = (match(cur, -1, false).match_len > 0);
3988 cur.pos() = cur.lastpos();
3990 cur.pos() = cur_orig.pos();
3991 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3992 DocIterator cur_prev_iter;
3994 found_match = (match(cur).match_len > 0);
3995 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3996 << found_match << ", cur: " << cur);
3998 MatchResult found_mr = findMostBackwards(cur, match);
3999 match.FillResults(found_mr);
4000 LASSERT(found_mr.pos_len > 0, /**/);
4001 return found_mr.pos_len;
4004 // Stop if begin of document reached
4005 if (cur == cur_begin)
4007 cur_prev_iter = cur;
4011 if (cur == cur_begin)
4018 } while (!theApp()->longOperationCancelled());
4026 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4027 DocIterator const & cur, int len)
4029 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4031 if (!opt.ignoreformat)
4032 return latexifyFromCursor(cur, len);
4034 return stringifyFromCursor(cur, len);
4038 FindAndReplaceOptions::FindAndReplaceOptions(
4039 docstring const & _find_buf_name, bool _casesensitive,
4040 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4041 docstring const & _repl_buf_name, bool _keep_case,
4042 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4043 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4044 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4045 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4053 /** Check if 'len' letters following cursor are all non-lowercase */
4054 static bool allNonLowercase(Cursor const & cur, int len)
4056 pos_type beg_pos = cur.selectionBegin().pos();
4057 pos_type end_pos = cur.selectionBegin().pos() + len;
4058 if (len > cur.lastpos() + 1 - beg_pos) {
4059 LYXERR(Debug::FIND, "This should not happen, more debug needed");
4060 len = cur.lastpos() + 1 - beg_pos;
4061 end_pos = beg_pos + len;
4063 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4064 if (isLowerCase(cur.paragraph().getChar(pos)))
4070 /** Check if first letter is upper case and second one is lower case */
4071 static bool firstUppercase(Cursor const & cur)
4074 pos_type pos = cur.selectionBegin().pos();
4075 if (pos >= cur.lastpos() - 1) {
4076 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
4079 ch1 = cur.paragraph().getChar(pos);
4080 ch2 = cur.paragraph().getChar(pos + 1);
4081 bool result = isUpperCase(ch1) && isLowerCase(ch2);
4082 LYXERR(Debug::FIND, "firstUppercase(): "
4083 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4084 << ch2 << "(" << char(ch2) << ")"
4085 << ", result=" << result << ", cur=" << cur);
4090 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4092 ** \fixme What to do with possible further paragraphs in replace buffer ?
4094 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4096 ParagraphList::iterator pit = buffer.paragraphs().begin();
4097 LASSERT(!pit->empty(), /**/);
4098 pos_type right = pos_type(1);
4099 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4100 right = pit->size();
4101 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4106 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4108 // Should replace the string "$" + std::to_string(matchnum) with replacement
4109 // if the char '$' is not prefixed with odd number of char '\\'
4110 static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4114 for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4116 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4118 int num = stoi(sub.str(3), nullptr, 10);
4119 if (num >= maxmatchnum)
4121 if (lastpos < (size_t) sub.position(2))
4122 s += t.substr(lastpos, sub.position(2) - lastpos);
4123 s += replacements[num];
4124 lastpos = sub.position(2) + sub.length(2);
4128 else if (lastpos < t.length())
4129 s += t.substr(lastpos, t.length() - lastpos);
4136 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4138 Cursor & cur = bv->cursor();
4139 if (opt.repl_buf_name.empty()
4140 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4141 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4144 DocIterator sel_beg = cur.selectionBegin();
4145 DocIterator sel_end = cur.selectionEnd();
4146 if (&sel_beg.inset() != &sel_end.inset()
4147 || sel_beg.pit() != sel_end.pit()
4148 || sel_beg.idx() != sel_end.idx())
4150 int sel_len = sel_end.pos() - sel_beg.pos();
4151 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4152 << ", sel_len: " << sel_len << endl);
4155 LASSERT(sel_len > 0, return 0);
4157 if (!matchAdv(sel_beg, sel_len).match_len)
4160 // Build a copy of the replace buffer, adapted to the KeepCase option
4161 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4163 repl_buffer_orig.write(oss);
4164 string lyx = oss.str();
4165 if (matchAdv.valid_matches > 0) {
4166 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4168 Buffer repl_buffer("", false);
4169 repl_buffer.setUnnamed(true);
4170 LASSERT(repl_buffer.readString(lyx), return 0);
4171 if (opt.keep_case && sel_len >= 2) {
4172 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4173 if (cur.inTexted()) {
4174 if (firstUppercase(cur))
4175 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4176 else if (allNonLowercase(cur, sel_len))
4177 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4180 cap::cutSelection(cur, false);
4181 if (cur.inTexted()) {
4182 repl_buffer.changeLanguage(
4183 repl_buffer.language(),
4184 cur.getFont().language());
4185 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4186 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4187 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4188 repl_buffer.params().documentClassPtr(),
4189 bv->buffer().errorList("Paste"));
4190 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4191 sel_len = repl_buffer.paragraphs().begin()->size();
4192 } else if (cur.inMathed()) {
4193 odocstringstream ods;
4195 // OutputParams runparams(&repl_buffer.params().encoding());
4196 OutputParams runparams(encodings.fromLyXName("utf8"));
4197 runparams.nice = false;
4198 runparams.flavor = Flavor::XeTeX;
4199 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4200 runparams.dryrun = true;
4201 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4202 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4203 docstring repl_latex = ods.str();
4204 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4206 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4207 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4208 repl_latex = from_utf8(s);
4209 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4210 MathData ar(cur.buffer());
4211 asArray(repl_latex, ar, Parse::NORMAL);
4213 sel_len = ar.size();
4214 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4216 if (cur.pos() >= sel_len)
4217 cur.pos() -= sel_len;
4220 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4221 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4222 bv->processUpdateFlags(Update::Force);
4227 /// Perform a FindAdv operation.
4228 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4233 // e.g., when invoking word-findadv from mini-buffer wither with
4234 // wrong options syntax or before ever opening advanced F&R pane
4235 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4239 MatchStringAdv matchAdv(bv->buffer(), opt);
4241 if (!matchAdv.regexIsValid) {
4242 bv->message(lyx::from_utf8(matchAdv.regexError));
4246 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4248 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4249 num_replaced += findAdvReplace(bv, opt, matchAdv);
4252 pos_len = findForwardAdv(cur, matchAdv);
4254 pos_len = findBackwardsAdv(cur, matchAdv);
4255 } catch (exception & ex) {
4256 bv->message(from_utf8(ex.what()));
4261 if (num_replaced > 0) {
4262 switch (num_replaced)
4265 bv->message(_("One match has been replaced."));
4268 bv->message(_("Two matches have been replaced."));
4271 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4277 bv->message(_("Match not found."));
4282 if (num_replaced > 0)
4283 bv->message(_("Match has been replaced."));
4285 bv->message(_("Match found."));
4287 if (cur.pos() + pos_len > cur.lastpos()) {
4288 // Prevent crash in bv->putSelectionAt()
4289 // Should never happen, maybe LASSERT() here?
4290 pos_len = cur.lastpos() - cur.pos();
4292 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4293 bv->putSelectionAt(cur, pos_len, !opt.forward);
4299 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4301 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4302 << opt.casesensitive << ' '
4303 << opt.matchword << ' '
4304 << opt.forward << ' '
4305 << opt.expandmacros << ' '
4306 << opt.ignoreformat << ' '
4307 << opt.replace_all << ' '
4308 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4309 << opt.keep_case << ' '
4310 << int(opt.scope) << ' '
4313 LYXERR(Debug::FIND, "built: " << os.str());
4319 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4321 LYXERR(Debug::FIND, "parsing");
4325 while (line != "EOSS") {
4329 if (is.eof()) // Tolerate malformed request
4333 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4334 opt.find_buf_name = from_utf8(s);
4335 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4336 is.get(); // Waste space before replace string
4339 while (line != "EOSS") {
4343 if (is.eof()) // Tolerate malformed request
4347 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4348 opt.repl_buf_name = from_utf8(s);
4349 is >> opt.keep_case;
4352 opt.scope = FindAndReplaceOptions::SearchScope(i);
4354 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4356 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4357 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4358 << opt.scope << ' ' << opt.restr);