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
644 Escapes const & get_regexp_escapes()
646 typedef std::pair<std::string, std::string> P;
648 static Escapes escape_map;
649 if (escape_map.empty()) {
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(".", "_x_."));
660 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
661 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
662 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
663 escape_map.push_back(P("_x_", "\\"));
669 /// A map of lyx escaped strings and their unescaped equivalent.
671 Escapes const & get_lyx_unescapes()
673 typedef std::pair<std::string, std::string> P;
675 static Escapes escape_map;
676 if (escape_map.empty()) {
677 escape_map.push_back(P("\\%", "%"));
678 escape_map.push_back(P("\\{", "{"));
679 escape_map.push_back(P("\\}", "}"));
680 escape_map.push_back(P("\\mathcircumflex ", "^"));
681 escape_map.push_back(P("\\mathcircumflex", "^"));
682 escape_map.push_back(P("\\backslash ", "\\"));
683 escape_map.push_back(P("\\backslash", "\\"));
684 escape_map.push_back(P("\\sim ", "~"));
685 escape_map.push_back(P("\\sim", "~"));
691 /// A map of escapes turning a regexp matching text to one matching latex.
693 Escapes const & get_regexp_latex_escapes()
695 typedef std::pair<std::string, std::string> P;
697 static Escapes escape_map;
698 if (escape_map.empty()) {
699 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
700 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
701 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
702 escape_map.push_back(P("\\[", "\\{\\[\\}"));
703 escape_map.push_back(P("\\]", "\\{\\]\\}"));
704 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
705 escape_map.push_back(P("%", "\\\\\\%"));
706 escape_map.push_back(P("#", "\\\\#"));
712 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
713 ** the found occurrence were escaped.
716 string apply_escapes(string s, Escapes const & escape_map)
718 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
719 Escapes::const_iterator it;
720 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
721 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
722 unsigned int pos = 0;
723 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
724 s.replace(pos, it->first.length(), it->second);
725 LYXERR(Debug::FIND, "After escape: " << s);
726 pos += it->second.length();
727 // LYXERR(Debug::FIND, "pos: " << pos);
730 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
736 string string2regex(string in)
738 static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\$\\])" };
739 string temp = std::regex_replace(in, specialChars, R"(\$&)" );
744 while (fl_pos < temp.size()) {
745 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
746 if (fl_pos == string::npos)
749 temp2 += temp.substr(lastpos, fl_pos - lastpos);
755 if (lastpos < temp.size()) {
756 temp2 += temp.substr(lastpos, temp.size() - lastpos);
761 string correctRegex(string t)
763 /* Convert \backslash => \
764 * and \{, \}, \[, \] => {, }, [, ]
767 regex wordre("(\\\\)*(\\\\((backslash|mathcircumflex) ?|[\\[\\]\\{\\}]))");
770 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
773 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
777 if (sub.str(4) == "backslash")
779 else if (sub.str(4) == "mathcircumflex")
782 replace = sub.str(3);
784 if (lastpos < (size_t) sub.position(2))
785 s += t.substr(lastpos, sub.position(2) - lastpos);
787 lastpos = sub.position(2) + sub.length(2);
791 else if (lastpos < t.length())
792 s += t.substr(lastpos, t.length() - lastpos);
796 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
797 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
798 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
799 string escape_for_regex(string s)
803 while (lastpos < s.size()) {
804 size_t regex_pos = s.find("\\regexp{", lastpos);
805 if (regex_pos == string::npos) {
806 regex_pos = s.size();
808 if (regex_pos > lastpos) {
809 result += string2regex(s.substr(lastpos, regex_pos-lastpos));
811 if (lastpos == s.size())
814 size_t end_pos = s.find("\\endregexp{}}", regex_pos + 8);
815 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)));
816 lastpos = end_pos + 13;
822 /// Wrapper for lyx::regex_replace with simpler interface
823 bool regex_replace(string const & s, string & t, string const & searchstr,
824 string const & replacestr)
826 regex e(searchstr, regex_constants::ECMAScript);
828 ostream_iterator<char, char> it(oss);
829 regex_replace(it, s.begin(), s.end(), e, replacestr);
830 // tolerate t and s be references to the same variable
831 bool rv = (s != oss.str());
837 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
839 ** Verify that closed braces exactly match open braces. This avoids that, for example,
840 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
843 ** Number of open braces that must remain open at the end for the verification to succeed.
846 bool braces_match(QString const & beg,
849 bool braces_match(string const & beg,
855 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg.toStdString() << "'");
857 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg << "'");
859 int lastidx = beg.size();
860 for (int i=0; i < lastidx; ++i) {
861 // Skip escaped braces in the count
871 } else if (c == '{') {
873 } else if (c == '}') {
874 if (open_pars == 0) {
875 LYXERR(Debug::FIND, "Found unmatched closed brace");
881 if (open_pars != unmatched) {
882 LYXERR(Debug::FIND, "Found " << open_pars
883 << " instead of " << unmatched
884 << " unmatched open braces at the end of count");
887 LYXERR(Debug::FIND, "Braces match as expected");
905 vector <string> result = vector <string>();
906 MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1) {};
909 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
911 int range = oldres.match_len;
912 if (range > 0) range--;
913 if (newres.match2end < oldres.match2end - oldres.match_len)
914 return MatchResult::newIsTooFar;
915 if (newres.match_len < oldres.match_len)
916 return MatchResult::newIsTooFar;
917 if ((newres.match_len == oldres.match_len) &&
918 (newres.match2end < oldres.match2end + range) &&
919 (newres.match2end > oldres.match2end - range)) {
920 return MatchResult::newIsBetter;
922 return MatchResult::newIsInvalid;
925 /** The class performing a match between a position in the document and the FindAdvOptions.
928 class MatchStringAdv {
930 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
932 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
933 ** constructor as opt.search, under the opt.* options settings.
936 ** If set, then match is searched only against beginning of text starting at cur.
937 ** If unset, then match is searched anywhere in text starting at cur.
940 ** The length of the matching text, or zero if no match was found.
942 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
951 /// first buffer on which search was started
952 lyx::Buffer * const p_first_buf;
954 FindAndReplaceOptions const & opt;
957 /// Auxiliary find method (does not account for opt.matchword)
958 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
959 void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
961 /** Normalize a stringified or latexified LyX paragraph.
965 ** <li>if search is not casesensitive, then lowercase the string;
966 ** <li>remove any newline at begin or end of the string;
967 ** <li>replace any newline in the middle of the string with a simple space;
968 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
971 ** @todo Normalization should also expand macros, if the corresponding
972 ** search option was checked.
974 string normalize(docstring const & s) const;
975 // normalized string to search
976 string par_as_string;
977 // regular expression to use for searching
978 // regexp2 is same as regexp, but prefixed with a ".*?"
980 QRegularExpression regexp;
981 QRegularExpression regexp2;
986 // leading format material as string
987 string lead_as_string;
988 // par_as_string after removal of lead_as_string
989 string par_as_string_nolead;
990 // unmatched open braces in the search string/regexp
992 // number of (.*?) subexpressions added at end of search regexp for closing
993 // environments, math mode, styles, etc...
996 // Are we searching with regular expressions ?
998 static int valid_matches;
999 static vector <string> matches;
1000 void FillResults(MatchResult &found_mr);
1003 int MatchStringAdv::valid_matches = 0;
1004 vector <string> MatchStringAdv::matches = vector <string>(10);
1006 void MatchStringAdv::FillResults(MatchResult &found_mr)
1008 if (found_mr.match_len > 0) {
1009 valid_matches = found_mr.result.size();
1010 for (size_t i = 0; i < found_mr.result.size(); i++)
1011 matches[i] = found_mr.result[i];
1017 static docstring buffer_to_latex(Buffer & buffer)
1019 //OutputParams runparams(&buffer.params().encoding());
1020 OutputParams runparams(encodings.fromLyXName("utf8"));
1021 odocstringstream ods;
1023 runparams.nice = true;
1024 runparams.flavor = Flavor::XeTeX;
1025 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1026 // No side effect of file copying and image conversion
1027 runparams.dryrun = true;
1028 runparams.for_search = true;
1029 pit_type const endpit = buffer.paragraphs().size();
1030 for (pit_type pit = 0; pit != endpit; ++pit) {
1031 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
1032 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
1038 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1041 if (!opt.ignoreformat) {
1042 str = buffer_to_latex(buffer);
1044 // OutputParams runparams(&buffer.params().encoding());
1045 OutputParams runparams(encodings.fromLyXName("utf8"));
1046 runparams.nice = true;
1047 runparams.flavor = Flavor::XeTeX;
1048 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1049 runparams.dryrun = true;
1050 runparams.for_search = true;
1051 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1052 Paragraph const & par = buffer.paragraphs().at(pit);
1053 LYXERR(Debug::FIND, "Adding to search string: '"
1054 << par.asString(pos_type(0), par.size(),
1055 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1058 str += par.asString(pos_type(0), par.size(),
1059 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1062 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1063 string t = to_utf8(str);
1064 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
1071 /// Return separation pos between the leading material and the rest
1072 static size_t identifyLeading(string const & s)
1075 // @TODO Support \item[text]
1076 // Kornel: Added textsl, textsf, textit, texttt and noun
1077 // + allow to search for colored text too
1078 while (regex_replace(t, t, "^\\\\(("
1079 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1080 "lyxaddress|lyxrightaddress|"
1081 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1082 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1083 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1084 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1085 || regex_replace(t, t, "^\\$", "")
1086 || regex_replace(t, t, "^\\\\\\[", "")
1087 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1088 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1090 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1095 * Given a latexified string, retrieve some handled features
1096 * The features of the regex will later be compared with the features
1097 * of the searched text. If the regex features are not a
1098 * subset of the analized, then, in not format ignoring search
1099 * we can early stop the search in the relevant inset.
1101 typedef map<string, bool> Features;
1103 static Features identifyFeatures(string const & s)
1105 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1106 static regex const valid("^("
1108 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1109 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1110 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1111 "(u|uu)line|(s|x)out|uwave|"
1112 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1113 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1118 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1121 if (sub.str(1).compare("regexp") == 0) {
1125 string token = sub.str(1);
1127 if (regex_match(token, sub2, valid)) {
1135 if (sub.str(1).compare("endregexp") == 0) {
1145 * defines values features of a key "\\[a-z]+{"
1150 /* Char type with content discarded
1151 * like \hspace{1cm} */
1153 /* Char, like \backslash */
1155 /* replace starting backslash with '#' */
1157 /* \part, \section*, ... */
1159 /* title, author etc */
1161 /* \foreignlanguage{ngerman}, ... */
1164 * to discard language in content */
1167 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1169 /* fonts, colors, markups, ... */
1171 /* footnotesize, ... large, ...
1172 * Ignore all of them */
1175 /* inputencoding, ...
1176 * Discard also content, because they do not help in search */
1179 * like remove, but also all arguments */
1181 /* item, listitem */
1183 /* tex, latex, ... like isChar */
1185 /* like \lettrine[lines=5]{}{} */
1187 // like isStandard, but always remove head
1189 /* End of arguments marker for lettrine,
1190 * so that they can be ignored */
1193 KeyInfo() = default;
1194 KeyInfo(KeyType type, int parcount, bool disable)
1196 parenthesiscount(parcount),
1197 disabled(disable) {}
1198 KeyType keytype = invalid;
1200 int _tokensize = -1;
1201 int _tokenstart = -1;
1202 int _dataStart = -1;
1204 int parenthesiscount = 1;
1205 bool disabled = false;
1206 bool used = false; /* by pattern */
1211 Border(int l=0, int u=0) : low(l), upper(u) {};
1216 #define MAXOPENED 30
1218 bool isPatternString_;
1220 explicit Intervall(bool isPattern, string const & p) :
1221 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1222 hasTitle(false), langcount(0)
1230 static vector<Border> borders;
1231 int depts[MAXOPENED];
1232 int closes[MAXOPENED];
1233 int actualdeptindex;
1234 int previousNotIgnored(int) const;
1235 int nextNotIgnored(int) const;
1236 void handleOpenP(int i);
1237 void handleCloseP(int i, bool closingAllowed);
1238 void resetOpenedP(int openPos);
1239 void addIntervall(int upper);
1240 void addIntervall(int low, int upper); /* if explicit */
1241 void removeAccents();
1242 void setForDefaultLang(KeyInfo const & defLang) const;
1243 int findclosing(int start, int end, char up, char down, int repeat);
1244 void handleParentheses(int lastpos, bool closingAllowed);
1246 int langcount; // Number of disabled language specs up to current position in actual interval
1247 int isOpeningPar(int pos) const;
1249 void output(ostringstream &os, int lastpos);
1250 // string show(int lastpos);
1253 vector<Border> Intervall::borders = vector<Border>(30);
1255 int Intervall::isOpeningPar(int pos) const
1257 if ((pos < 0) || (size_t(pos) >= par.size()))
1259 if (par[pos] != '{')
1261 if (size_t(pos) + 2 >= par.size())
1263 if (par[pos+2] != '}')
1265 if (par[pos+1] == '[' || par[pos+1] == ']')
1270 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1272 // Enable the use of first token again
1273 if (ignoreidx >= 0) {
1274 int value = defLang._tokenstart + defLang._tokensize;
1280 if (borders[borderidx].low < value)
1281 borders[borderidx].low = value;
1282 if (borders[borderidx].upper < value)
1283 borders[borderidx].upper = value;
1288 static void checkDepthIndex(int val)
1290 static int maxdepthidx = MAXOPENED-2;
1291 static int lastmaxdepth = 0;
1292 if (val > lastmaxdepth) {
1293 LYXERR(Debug::INFO, "Depth reached " << val);
1296 if (val > maxdepthidx) {
1298 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1303 // Not needed, because borders are now dynamically expanded
1304 static void checkIgnoreIdx(int val)
1306 static int lastmaxignore = -1;
1307 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1308 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1309 lastmaxignore = val;
1315 * Expand the region of ignored parts of the input latex string
1316 * The region is only relevant in output()
1318 void Intervall::addIntervall(int low, int upper)
1321 if (low == upper) return;
1322 for (idx = ignoreidx+1; idx > 0; --idx) {
1323 if (low > borders[idx-1].upper) {
1327 Border br(low, upper);
1328 if (idx > ignoreidx) {
1329 if (borders.size() <= size_t(idx)) {
1330 borders.push_back(br);
1336 // checkIgnoreIdx(ignoreidx);
1340 // Expand only if one of the new bound is inside the interwall
1341 // We know here that br.low > borders[idx-1].upper
1342 if (br.upper < borders[idx].low) {
1343 // We have to insert at this pos
1344 if (size_t(ignoreidx+1) >= borders.size()) {
1345 borders.push_back(borders[ignoreidx]);
1348 borders[ignoreidx+1] = borders[ignoreidx];
1350 for (int i = ignoreidx; i > idx; --i) {
1351 borders[i] = borders[i-1];
1355 // checkIgnoreIdx(ignoreidx);
1358 // Here we know, that we are overlapping
1359 if (br.low > borders[idx].low)
1360 br.low = borders[idx].low;
1361 // check what has to be concatenated
1363 for (int i = idx; i <= ignoreidx; i++) {
1364 if (br.upper >= borders[i].low) {
1366 if (br.upper < borders[i].upper)
1367 br.upper = borders[i].upper;
1373 // count should be >= 1 here
1376 for (int i = idx + count; i <= ignoreidx; i++) {
1377 borders[i-count+1] = borders[i];
1379 ignoreidx -= count - 1;
1385 typedef map<string, string> AccentsMap;
1386 static AccentsMap accents = map<string, string>();
1388 static void buildaccent(string n, string param, string values)
1392 const char delim = '|';
1393 while (getline(s, name, delim)) {
1395 for (char c : param) {
1396 string key = name + "{" + c + "}";
1397 // get the corresponding utf8-value
1398 if ((values[start] & 0xc0) != 0xc0) {
1399 // should not happen, utf8 encoding starts at least with 11xxxxxx
1400 // but value for '\dot{i}' is 'i', which is ascii
1401 if ((values[start] & 0x80) == 0) {
1403 accents[key] = values.substr(start, 1);
1404 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1409 for (int j = 1; ;j++) {
1410 if (start + j >= values.size()) {
1411 accents[key] = values.substr(start, j);
1412 start = values.size() - 1;
1415 else if ((values[start+j] & 0xc0) != 0x80) {
1416 // This is the first byte of following utf8 char
1417 accents[key] = values.substr(start, j);
1419 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1428 static string getutf8(unsigned uchar)
1431 string ret = string();
1433 if (uchar <= 0x7f) {
1434 c[maxc-1] = uchar & 0x7f;
1437 unsigned char rest = 0x40;
1438 unsigned char first = 0x80;
1440 for (int i = start; i >=0; --i) {
1442 c[i] = first + uchar;
1445 c[i] = 0x80 | (uchar & 0x3f);
1452 for (int i = 0; i < maxc; i++) {
1453 if (c[i] == 0) continue;
1459 static void buildAccentsMap()
1461 accents["imath"] = "ı";
1463 accents["jmath"] = "ȷ";
1464 accents["cdot"] = "·";
1465 accents["textasciicircum"] = "^";
1466 accents["mathcircumflex"] = "^";
1467 accents["sim"] = "~";
1468 accents["guillemotright"] = "»";
1469 accents["guillemotleft"] = "«";
1470 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1471 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1472 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1473 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1474 accents["negmedspace"] = getutf8(0xf0005);
1475 accents["thickspace"] = getutf8(0xf0006);
1476 accents["negthickspace"] = getutf8(0xf0007);
1477 accents["lyx"] = getutf8(0xf0010); // Used logos
1478 accents["LyX"] = getutf8(0xf0010);
1479 accents["tex"] = getutf8(0xf0011);
1480 accents["TeX"] = getutf8(0xf0011);
1481 accents["latex"] = getutf8(0xf0012);
1482 accents["LaTeX"] = getutf8(0xf0012);
1483 accents["latexe"] = getutf8(0xf0013);
1484 accents["LaTeXe"] = getutf8(0xf0013);
1485 accents["lyxarrow"] = getutf8(0xf0020);
1486 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1487 accents["backslash LyX"] = getutf8(0xf0010);
1488 accents["backslash tex"] = getutf8(0xf0011);
1489 accents["backslash TeX"] = getutf8(0xf0011);
1490 accents["backslash latex"] = getutf8(0xf0012);
1491 accents["backslash LaTeX"] = getutf8(0xf0012);
1492 accents["backslash latexe"] = getutf8(0xf0013);
1493 accents["backslash LaTeXe"] = getutf8(0xf0013);
1494 accents["backslash lyxarrow"] = getutf8(0xf0020);
1495 accents["ddot{\\imath}"] = "ï";
1496 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1497 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1498 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1499 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1500 accents["acute{\\imath}"] = "í";
1501 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1502 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1503 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1504 buildaccent("mathring|r", "aAuUwy",
1506 accents["check{\\imath}"] = "ǐ";
1507 accents["check{\\jmath}"] = "ǰ";
1508 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1509 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1510 accents["hat{\\imath}"] = "î";
1511 accents["hat{\\jmath}"] = "ĵ";
1512 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1513 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1514 accents["bar{\\imath}"] = "ī";
1515 buildaccent("bar|=", "aAeEiIoOuUyY",
1516 "āĀēĒīĪōŌūŪȳȲ"); // macron
1517 accents["tilde{\\imath}"] = "ĩ";
1518 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1519 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1520 accents["breve{\\imath}"] = "ĭ";
1521 buildaccent("breve|u", "aAeEgGiIoOuU",
1522 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1523 accents["grave{\\imath}"] = "ì";
1524 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1525 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1526 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1527 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1528 buildaccent("ogonek|k", "AaEeIiUuOo",
1529 "ĄąĘęĮįŲųǪǫ"); // ogonek
1530 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1531 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1532 buildaccent("subring|textsubring", "Aa",
1534 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1535 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1536 buildaccent("subtilde|textsubtilde", "EeIiUu",
1537 "ḚḛḬḭṴṵ"); // subtilde
1538 accents["dgrave{\\imath}"] = "ȉ";
1539 accents["textdoublegrave{\\i}"] = "ȉ";
1540 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1541 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1542 accents["rcap{\\imath}"] = "ȉ";
1543 accents["textroundcap{\\i}"] = "ȉ";
1544 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1545 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1546 buildaccent("slashed", "oO",
1551 * Created accents in math or regexp environment
1552 * are macros, but we need the utf8 equivalent
1554 void Intervall::removeAccents()
1556 if (accents.empty())
1558 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1559 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1560 "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))|guillemot(left|right)|textasciicircum|mathcircumflex|sim)(?![a-zA-Z]))");
1562 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1564 string key = sub.str(1);
1565 if (accents.find(key) != accents.end()) {
1566 string val = accents[key];
1567 size_t pos = sub.position(size_t(0));
1568 for (size_t i = 0; i < val.size(); i++) {
1569 par[pos+i] = val[i];
1571 // Remove possibly following space too
1572 if (par[pos+sub.str(0).size()] == ' ')
1573 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1575 addIntervall(pos+val.size(), pos + sub.str(0).size());
1576 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1577 // remove traces of any remaining chars
1582 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1587 void Intervall::handleOpenP(int i)
1590 depts[actualdeptindex] = i+1;
1591 closes[actualdeptindex] = -1;
1592 checkDepthIndex(actualdeptindex);
1595 void Intervall::handleCloseP(int i, bool closingAllowed)
1597 if (actualdeptindex <= 0) {
1598 if (! closingAllowed)
1599 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1600 // if we are at the very end
1601 addIntervall(i, i+1);
1604 closes[actualdeptindex] = i+1;
1609 void Intervall::resetOpenedP(int openPos)
1611 // Used as initializer for foreignlanguage entry
1612 actualdeptindex = 1;
1613 depts[1] = openPos+1;
1617 int Intervall::previousNotIgnored(int start) const
1619 int idx = 0; /* int intervalls */
1620 for (idx = ignoreidx; idx >= 0; --idx) {
1621 if (start > borders[idx].upper)
1623 if (start >= borders[idx].low)
1624 start = borders[idx].low-1;
1629 int Intervall::nextNotIgnored(int start) const
1631 int idx = 0; /* int intervalls */
1632 for (idx = 0; idx <= ignoreidx; idx++) {
1633 if (start < borders[idx].low)
1635 if (start < borders[idx].upper)
1636 start = borders[idx].upper;
1641 typedef map<string, KeyInfo> KeysMap;
1642 typedef vector< KeyInfo> Entries;
1643 static KeysMap keys = map<string, KeyInfo>();
1649 Intervall interval_;
1650 void buildKeys(bool);
1651 void buildEntries(bool);
1652 void makeKey(const string &, KeyInfo, bool isPatternString);
1653 void processRegion(int start, int region_end); /* remove {} parts */
1654 void removeHead(KeyInfo const &, int count=0);
1657 LatexInfo(string const & par, bool isPatternString)
1658 : entidx_(-1), interval_(isPatternString, par)
1660 buildKeys(isPatternString);
1661 entries_ = vector<KeyInfo>();
1662 buildEntries(isPatternString);
1666 if (entries_.empty()) {
1669 if (entries_[0].keytype == KeyInfo::isTitle) {
1670 interval_.hasTitle = true;
1671 if (! entries_[0].disabled) {
1672 interval_.titleValue = entries_[0].head;
1675 interval_.titleValue = "";
1677 removeHead(entries_[0]);
1678 if (entries_.size() > 1)
1687 if (int(entries_.size()) > entidx_) {
1694 bool setNextKey(int idx) {
1695 if ((idx == entidx_) && (entidx_ >= 0)) {
1702 int find(int start, KeyInfo::KeyType keytype) const {
1706 while (tmpIdx < int(entries_.size())) {
1707 if (entries_[tmpIdx].keytype == keytype)
1713 int process(ostringstream & os, KeyInfo const & actual);
1714 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1715 // string show(int lastpos) { return interval.show(lastpos);};
1716 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1717 KeyInfo &getKeyInfo(int keyinfo) {
1718 static KeyInfo invalidInfo = KeyInfo();
1719 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1722 return entries_[keyinfo];
1724 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1725 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1729 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1733 for (int i = start; i < end; i += 1 + skip) {
1737 if (c == '\\') skip = 1;
1741 else if (c == down) {
1744 if ((repeat <= 0) || (par[i+1] != up))
1762 vector<MathEntry> entries_;
1767 void insert(string const & wait, size_t start, size_t end) {
1768 MathEntry m = MathEntry();
1770 m.mathStart = start;
1772 m.mathSize = end - start;
1773 entries_.push_back(m);
1775 bool empty() const { return entries_.empty(); };
1776 size_t getEndPos() const {
1777 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1780 return entries_[actualIdx_].mathEnd;
1782 size_t getStartPos() const {
1783 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1784 return 100000; /* definitely enough? */
1786 return entries_[actualIdx_].mathStart;
1788 size_t getFirstPos() {
1790 return getStartPos();
1792 size_t getSize() const {
1793 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1796 return entries_[actualIdx_].mathSize;
1798 void incrEntry() { actualIdx_++; };
1801 void LatexInfo::buildEntries(bool isPatternString)
1803 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\})");
1804 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1805 static bool disableLanguageOverride = false;
1806 smatch sub, submath;
1807 bool evaluatingRegexp = false;
1809 bool evaluatingMath = false;
1810 bool evaluatingCode = false;
1812 bool evaluatingOptional = false;
1813 size_t optionalEnd = 0;
1816 bool math_end_waiting = false;
1817 size_t math_pos = 10000;
1819 static vector<string> usedText = vector<string>();
1821 interval_.removeAccents();
1823 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1825 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1826 // prefixed by odd count of '\\'
1829 if (math_end_waiting) {
1830 size_t pos = submath.position(size_t(2));
1831 if ((math_end == "$") &&
1832 (submath.str(2) == "$")) {
1833 mi.insert("$", math_pos, pos + 1);
1834 math_end_waiting = false;
1836 else if ((math_end == "\\]") &&
1837 (submath.str(2) == "\\]")) {
1838 mi.insert("\\]", math_pos, pos + 2);
1839 math_end_waiting = false;
1841 else if ((submath.str(3).compare("end") == 0) &&
1842 (submath.str(4).compare(math_end) == 0)) {
1843 mi.insert(math_end, math_pos, pos + submath.str(2).length());
1844 math_end_waiting = false;
1850 if (submath.str(3).compare("begin") == 0) {
1851 math_end_waiting = true;
1852 math_end = submath.str(4);
1853 math_pos = submath.position(size_t(2));
1855 else if (submath.str(2).compare("\\[") == 0) {
1856 math_end_waiting = true;
1858 math_pos = submath.position(size_t(2));
1860 else if (submath.str(2) == "$") {
1861 size_t pos = submath.position(size_t(2));
1862 math_end_waiting = true;
1868 // Ignore language if there is math somewhere in pattern-string
1869 if (isPatternString) {
1870 for (auto s: usedText) {
1871 // Remove entries created in previous search runs
1874 usedText = vector<string>();
1877 keys["foreignlanguage"].disabled = true;
1878 disableLanguageOverride = true;
1881 disableLanguageOverride = false;
1884 if (disableLanguageOverride) {
1885 keys["foreignlanguage"].disabled = true;
1888 math_pos = mi.getFirstPos();
1889 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1891 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1892 // prefixed by odd count of '\\'
1895 string key = sub.str(5);
1897 if (sub.str(2)[0] == '\\')
1898 key = sub.str(2)[1];
1903 if (keys.find(key) != keys.end()) {
1904 if (keys[key].keytype == KeyInfo::headRemove) {
1905 KeyInfo found1 = keys[key];
1906 found1.disabled = true;
1907 found1.head = "\\" + key + "{";
1908 found1._tokenstart = sub.position(size_t(2));
1909 found1._tokensize = found1.head.length();
1910 found1._dataStart = found1._tokenstart + found1.head.length();
1911 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1912 found1._dataEnd = endpos;
1917 if (evaluatingRegexp) {
1918 if (sub.str(3).compare("endregexp") == 0) {
1919 evaluatingRegexp = false;
1920 // found._tokenstart already set
1921 found._dataEnd = sub.position(size_t(2)) + 13;
1922 found._dataStart = found._dataEnd;
1923 found._tokensize = found._dataEnd - found._tokenstart;
1924 found.parenthesiscount = 0;
1925 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1932 if (evaluatingMath) {
1933 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1935 evaluatingMath = false;
1937 math_pos = mi.getStartPos();
1939 if (keys.find(key) == keys.end()) {
1940 found = KeyInfo(KeyInfo::isStandard, 0, true);
1941 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1942 found = KeyInfo(KeyInfo::isText, 0, false);
1943 if (isPatternString) {
1944 found.keytype = KeyInfo::isChar;
1945 found.disabled = false;
1949 usedText.push_back(key);
1953 if (key.compare("regexp") == 0) {
1954 evaluatingRegexp = true;
1955 found._tokenstart = sub.position(size_t(2));
1956 found._tokensize = 0;
1960 // Handle the other params of key
1961 if (found.keytype == KeyInfo::isIgnored)
1963 else if (found.keytype == KeyInfo::isMath) {
1964 if (size_t(sub.position(size_t(2))) == math_pos) {
1966 found._tokenstart = sub.position(size_t(2));
1967 found._tokensize = mi.getSize();
1968 found._dataEnd = found._tokenstart + found._tokensize;
1969 found._dataStart = found._dataEnd;
1970 found.parenthesiscount = 0;
1971 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1972 evaluatingMath = true;
1975 // begin|end of unknown env, discard
1976 // First handle tables
1977 // longtable|tabular
1978 bool discardComment;
1980 found.keytype = KeyInfo::doRemove;
1981 if ((sub.str(7).compare("longtable") == 0) ||
1982 (sub.str(7).compare("tabular") == 0)) {
1983 discardComment = true; /* '%' */
1986 discardComment = false;
1987 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1989 string token = sub.str(7);
1990 if (regex_match(token, sub2, removeArgs)) {
1991 found.keytype = KeyInfo::removeWithArg;
1994 // discard spaces before pos(2)
1995 int pos = sub.position(size_t(2));
1997 for (count = 0; pos - count > 0; count++) {
1998 char c = interval_.par[pos-count-1];
1999 if (discardComment) {
2000 if ((c != ' ') && (c != '%'))
2006 found._tokenstart = pos - count;
2007 if (sub.str(3).compare(0, 5, "begin") == 0) {
2008 size_t pos1 = pos + sub.str(2).length();
2009 if (sub.str(7).compare("cjk") == 0) {
2010 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2011 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2013 found.keytype = KeyInfo::isMain;
2014 found._dataStart = pos1;
2015 found._dataEnd = interval_.par.length();
2016 found.disabled = keys["foreignlanguage"].disabled;
2017 found.used = keys["foreignlanguage"].used;
2018 found._tokensize = pos1 - found._tokenstart;
2019 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2022 // Swallow possible optional params
2023 while (interval_.par[pos1] == '[') {
2024 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2026 // Swallow also the eventual parameter
2027 if (interval_.par[pos1] == '{') {
2028 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2031 found._dataEnd = pos1;
2033 found._dataStart = found._dataEnd;
2034 found._tokensize = count + found._dataEnd - pos;
2035 found.parenthesiscount = 0;
2036 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2037 found.disabled = true;
2041 // Handle "\end{...}"
2042 found._dataStart = pos + sub.str(2).length();
2043 found._dataEnd = found._dataStart;
2044 found._tokensize = count + found._dataEnd - pos;
2045 found.parenthesiscount = 0;
2046 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2047 found.disabled = true;
2051 else if (found.keytype != KeyInfo::isRegex) {
2052 found._tokenstart = sub.position(size_t(2));
2053 if (found.parenthesiscount == 0) {
2054 // Probably to be discarded
2055 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2056 char following = interval_.par[following_pos];
2057 if (following == ' ')
2058 found.head = "\\" + sub.str(5) + " ";
2059 else if (following == '=') {
2060 // like \uldepth=1000pt
2061 found.head = sub.str(2);
2064 found.head = "\\" + key;
2065 found._tokensize = found.head.length();
2066 found._dataEnd = found._tokenstart + found._tokensize;
2067 found._dataStart = found._dataEnd;
2070 int params = found._tokenstart + key.length() + 1;
2071 if (evaluatingOptional) {
2072 if (size_t(found._tokenstart) > optionalEnd) {
2073 evaluatingOptional = false;
2076 found.disabled = true;
2079 int optend = params;
2080 while (interval_.par[optend] == '[') {
2081 // discard optional parameters
2082 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2084 if (optend > params) {
2085 key += interval_.par.substr(params, optend-params);
2086 evaluatingOptional = true;
2087 optionalEnd = optend;
2088 if (found.keytype == KeyInfo::isSectioning) {
2089 // Remove optional values (but still keep in header)
2090 interval_.addIntervall(params, optend);
2093 string token = sub.str(7);
2095 if (interval_.par[optend] != '{') {
2097 found.parenthesiscount = 0;
2098 found.head = "\\" + key;
2101 closings = found.parenthesiscount;
2102 if (found.parenthesiscount == 1) {
2103 found.head = "\\" + key + "{";
2105 else if (found.parenthesiscount > 1) {
2107 found.head = sub.str(2) + "{";
2108 closings = found.parenthesiscount - 1;
2111 found.head = "\\" + key + "{";
2114 found._tokensize = found.head.length();
2115 found._dataStart = found._tokenstart + found.head.length();
2116 if (found.keytype == KeyInfo::doRemove) {
2118 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2119 if (endpar >= interval_.par.length())
2120 found._dataStart = interval_.par.length();
2122 found._dataStart = endpar;
2123 found._tokensize = found._dataStart - found._tokenstart;
2126 found._dataStart = found._tokenstart + found._tokensize;
2130 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2131 found._dataStart += 15;
2135 endpos = found._dataStart - 1;
2137 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2138 if (found.keytype == KeyInfo::isList) {
2139 // Check if it really is list env
2140 static regex const listre("^([a-z]+)$");
2142 if (!regex_match(token, sub2, listre)) {
2143 // Change the key of this entry. It is not in a list/item environment
2144 found.keytype = KeyInfo::endArguments;
2147 if (found.keytype == KeyInfo::noMain) {
2148 evaluatingCode = true;
2150 codeStart = found._dataStart;
2152 else if (evaluatingCode) {
2153 if (size_t(found._dataStart) > codeEnd)
2154 evaluatingCode = false;
2155 else if (found.keytype == KeyInfo::isMain) {
2156 // Disable this key, treate it as standard
2157 found.keytype = KeyInfo::isStandard;
2158 found.disabled = true;
2159 if ((codeEnd +1 >= interval_.par.length()) &&
2160 (found._tokenstart == codeStart)) {
2161 // trickery, because the code inset starts
2162 // with \selectlanguage ...
2164 if (entries_.size() > 1) {
2165 entries_[entries_.size()-1]._dataEnd = codeEnd;
2170 if ((endpos == interval_.par.length()) &&
2171 (found.keytype == KeyInfo::doRemove)) {
2172 // Missing closing => error in latex-input?
2173 // therefore do not delete remaining data
2174 found._dataStart -= 1;
2175 found._dataEnd = found._dataStart;
2178 found._dataEnd = endpos;
2180 if (isPatternString) {
2181 keys[key].used = true;
2184 entries_.push_back(found);
2188 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2190 stringstream s(keysstring);
2192 const char delim = '|';
2193 while (getline(s, key, delim)) {
2194 KeyInfo keyII(keyI);
2195 if (isPatternString) {
2198 else if ( !keys[key].used)
2199 keyII.disabled = true;
2204 void LatexInfo::buildKeys(bool isPatternString)
2207 static bool keysBuilt = false;
2208 if (keysBuilt && !isPatternString) return;
2210 // Keys to ignore in any case
2211 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2212 // Known standard keys with 1 parameter.
2213 // Split is done, if not at start of region
2214 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2215 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2216 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2217 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2218 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2219 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2221 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2222 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2223 makeKey("section*|subsection*|subsubsection*|paragraph*",
2224 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2225 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2226 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2228 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2230 // Split is done, if not at start of region
2231 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2232 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2234 // Split is done always.
2235 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2239 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2240 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2241 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2242 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2244 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2245 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2246 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2247 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2249 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2250 // Custom space/skip, remove the content (== length value)
2251 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2252 // Found in fr/UserGuide.lyx
2253 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2255 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2256 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2257 // Known macros to remove (including their parameter)
2259 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2260 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2261 // handle like standard keys with 1 parameter.
2262 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2264 // Ignore deleted text
2265 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2266 // but preserve added text
2267 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2269 // Macros to remove, but let the parameter survive
2271 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2273 // Remove language spec from content of these insets
2274 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2276 // Same effect as previous, parameter will survive (because there is no one anyway)
2278 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2279 // Remove table decorations
2280 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2281 // Discard shape-header.
2282 // For footnote or shortcut too, because of lang settings
2283 // and wrong handling if used 'KeyInfo::noMain'
2284 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2285 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2286 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2287 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2288 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2289 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2290 // like ('tiny{}' or '\tiny ' ... )
2291 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2293 // Survives, like known character
2294 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2295 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2296 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2298 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2299 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2300 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2302 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2303 // Remove RTL/LTR marker
2304 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2305 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2306 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2307 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2308 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2309 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2310 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2311 if (isPatternString) {
2312 // Allow the first searched string to rebuild the keys too
2316 // no need to rebuild again
2322 * Keep the list of actual opened parentheses actual
2323 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2325 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2328 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2332 if (c == '\\') skip = 1;
2333 else if (c == '{') {
2336 else if (c == '}') {
2337 handleCloseP(i, closingAllowed);
2343 string Intervall::show(int lastpos)
2345 int idx = 0; /* int intervalls */
2348 for (idx = 0; idx <= ignoreidx; idx++) {
2349 while (i < lastpos) {
2351 if (i <= borders[idx].low) {
2352 if (borders[idx].low > lastpos)
2353 printsize = lastpos - i;
2355 printsize = borders[idx].low - i;
2356 s += par.substr(i, printsize);
2358 if (i >= borders[idx].low)
2359 i = borders[idx].upper;
2362 i = borders[idx].upper;
2368 s += par.substr(i, lastpos-i);
2374 void Intervall::output(ostringstream &os, int lastpos)
2376 // get number of chars to output
2377 int idx = 0; /* int intervalls */
2380 string startTitle = titleValue;
2381 for (idx = 0; idx <= ignoreidx; idx++) {
2383 if (i <= borders[idx].low) {
2385 if (borders[idx].low > lastpos)
2386 printsize = lastpos - i;
2388 printsize = borders[idx].low - i;
2389 if (printsize > 0) {
2390 os << startTitle << par.substr(i, printsize);
2392 printed += printsize;
2395 handleParentheses(i, false);
2396 if (i >= borders[idx].low)
2397 i = borders[idx].upper;
2400 i = borders[idx].upper;
2407 os << startTitle << par.substr(i, lastpos-i);
2408 printed += lastpos-i;
2410 handleParentheses(lastpos, false);
2412 if (keys["foreignlanguage"].disabled)
2413 startindex = actualdeptindex-langcount;
2415 startindex = actualdeptindex;
2416 for (int i = startindex; i > 0; --i) {
2419 if (hasTitle && (printed > 0))
2421 if (! isPatternString_)
2423 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2426 void LatexInfo::processRegion(int start, int region_end)
2428 while (start < region_end) { /* Let {[} and {]} survive */
2429 int cnt = interval_.isOpeningPar(start);
2431 // Closing is allowed past the region
2432 int closing = interval_.findclosing(start+1, interval_.par.length());
2433 interval_.addIntervall(start, start+1);
2434 interval_.addIntervall(closing, closing+1);
2438 start = interval_.nextNotIgnored(start+1);
2442 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2444 if (actual.parenthesiscount == 0) {
2445 // "{\tiny{} ...}" ==> "{{} ...}"
2446 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2449 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2450 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2451 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2455 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2458 switch (actual.keytype)
2460 case KeyInfo::isTitle: {
2462 nextKeyIdx = getNextKey();
2465 case KeyInfo::cleanToStart: {
2466 actual._dataEnd = actual._dataStart;
2467 nextKeyIdx = getNextKey();
2468 // Search for end of arguments
2469 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2471 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2472 entries_[i].disabled = true;
2474 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2476 while (interval_.par[actual._dataEnd] == ' ')
2478 interval_.addIntervall(0, actual._dataEnd+1);
2479 interval_.actualdeptindex = 0;
2480 interval_.depts[0] = actual._dataEnd+1;
2481 interval_.closes[0] = -1;
2484 case KeyInfo::isText:
2485 interval_.par[actual._tokenstart] = '#';
2486 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2487 nextKeyIdx = getNextKey();
2489 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2490 if (actual.disabled)
2491 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2493 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2496 case KeyInfo::isChar: {
2497 nextKeyIdx = getNextKey();
2500 case KeyInfo::isSize: {
2501 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2502 if (actual.parenthesiscount == 0)
2503 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2505 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2507 nextKeyIdx = getNextKey();
2509 // Here _dataStart points to '{', so correct it
2510 actual._dataStart += 1;
2511 actual._tokensize += 1;
2512 actual.parenthesiscount = 1;
2513 if (interval_.par[actual._dataStart] == '}') {
2514 // Determine the end if used like '{\tiny{}...}'
2515 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2516 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2519 // Determine the end if used like '\tiny{...}'
2520 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2522 // Split on this key if not at start
2523 int start = interval_.nextNotIgnored(previousStart);
2524 if (start < actual._tokenstart) {
2525 interval_.output(os, actual._tokenstart);
2526 interval_.addIntervall(start, actual._tokenstart);
2528 // discard entry if at end of actual
2529 nextKeyIdx = process(os, actual);
2533 case KeyInfo::endArguments: {
2534 // Remove trailing '{}' too
2535 actual._dataStart += 1;
2536 actual._dataEnd += 1;
2537 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2538 nextKeyIdx = getNextKey();
2541 case KeyInfo::noMain:
2543 case KeyInfo::isStandard: {
2544 if (actual.disabled) {
2546 processRegion(actual._dataStart, actual._dataStart+1);
2547 nextKeyIdx = getNextKey();
2549 // Split on this key if not at datastart of calling entry
2550 int start = interval_.nextNotIgnored(previousStart);
2551 if (start < actual._tokenstart) {
2552 interval_.output(os, actual._tokenstart);
2553 interval_.addIntervall(start, actual._tokenstart);
2555 // discard entry if at end of actual
2556 nextKeyIdx = process(os, actual);
2560 case KeyInfo::removeWithArg: {
2561 nextKeyIdx = getNextKey();
2562 // Search for end of arguments
2563 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2565 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2566 entries_[i].disabled = true;
2568 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2570 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2573 case KeyInfo::doRemove: {
2574 // Remove the key with all parameters and following spaces
2576 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2577 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2580 // Remove also enclosing parentheses [] and {}
2583 while (actual._tokenstart > numpars) {
2584 if (pos+numpars >= interval_.par.size())
2586 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2588 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2594 if (interval_.par[pos+numpars] == ' ')
2598 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2599 nextKeyIdx = getNextKey();
2602 case KeyInfo::isList: {
2603 // Discard space before _tokenstart
2605 for (count = 0; count < actual._tokenstart; count++) {
2606 if (interval_.par[actual._tokenstart-count-1] != ' ')
2609 nextKeyIdx = getNextKey();
2610 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2612 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2614 // How else can we catch this one?
2615 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2616 entries_[i].disabled = true;
2618 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2620 else if (nextKeyIdx > 0) {
2621 // Ignore any lang entries inside data region
2622 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2623 if (entries_[i].keytype == KeyInfo::isMain)
2624 entries_[i].disabled = true;
2627 if (actual.disabled) {
2628 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2631 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2633 if (interval_.par[actual._dataEnd+1] == '[') {
2634 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2635 if ((interval_.par[actual._dataEnd+2] == '{') &&
2636 (interval_.par[posdown-1] == '}')) {
2637 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2638 interval_.addIntervall(posdown-1, posdown+1);
2641 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2642 interval_.addIntervall(posdown, posdown+1);
2644 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2645 if (blk > posdown) {
2646 // Discard at most 1 space after empty item
2648 for (count = 0; count < 1; count++) {
2649 if (interval_.par[blk+count] != ' ')
2653 interval_.addIntervall(blk, blk+count);
2658 case KeyInfo::isSectioning: {
2659 // Discard spaces before _tokenstart
2661 int val = actual._tokenstart;
2662 for (count = 0; count < actual._tokenstart;) {
2663 val = interval_.previousNotIgnored(val-1);
2664 if (val < 0 || interval_.par[val] != ' ')
2667 count = actual._tokenstart - val;
2670 if (actual.disabled) {
2671 removeHead(actual, count);
2672 nextKeyIdx = getNextKey();
2674 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2675 nextKeyIdx = process(os, actual);
2679 case KeyInfo::isMath: {
2680 // Same as regex, use the content unchanged
2681 nextKeyIdx = getNextKey();
2684 case KeyInfo::isRegex: {
2685 // DO NOT SPLIT ON REGEX
2687 nextKeyIdx = getNextKey();
2690 case KeyInfo::isIgnored: {
2691 // Treat like a character for now
2692 nextKeyIdx = getNextKey();
2695 case KeyInfo::isMain: {
2696 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2697 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2698 if (actual._tokenstart > 0) {
2699 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2700 if ((prev >= 0) && interval_.par[prev] == '%')
2701 interval_.addIntervall(prev, prev+1);
2703 if (actual.disabled) {
2705 interval_.langcount++;
2706 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2707 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2708 // Discard also the space before math-equation
2709 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2711 nextKeyIdx = getNextKey();
2712 // interval.resetOpenedP(actual._dataStart-1);
2715 if (actual._tokenstart < 26) {
2716 // for the first (and maybe dummy) language
2717 interval_.setForDefaultLang(actual);
2719 interval_.resetOpenedP(actual._dataStart-1);
2723 case KeyInfo::invalid:
2724 case KeyInfo::headRemove:
2725 // These two cases cannot happen, already handled
2728 // LYXERR(Debug::INFO, "Unhandled keytype");
2729 nextKeyIdx = getNextKey();
2736 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2738 int end = interval_.nextNotIgnored(actual._dataEnd);
2739 int oldStart = actual._dataStart;
2740 int nextKeyIdx = getNextKey();
2742 if ((nextKeyIdx < 0) ||
2743 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2744 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2745 if (oldStart <= end) {
2746 processRegion(oldStart, end);
2751 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2753 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2754 (void) dispatch(os, actual._dataStart, nextKey);
2755 end = nextKey._tokenstart;
2758 processRegion(oldStart, nextKey._tokenstart);
2759 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2761 oldStart = nextKey._dataEnd+1;
2763 // now nextKey is either invalid or is outside of actual._dataEnd
2764 // output the remaining and discard myself
2765 if (oldStart <= end) {
2766 processRegion(oldStart, end);
2768 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2770 // This is the normal case.
2771 // But if using the firstlanguage, the closing may be missing
2773 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2775 if (actual._dataEnd < end)
2776 output_end = interval_.nextNotIgnored(actual._dataEnd);
2777 else if (interval_.par.size() > (size_t) end)
2778 output_end = interval_.nextNotIgnored(end);
2780 output_end = interval_.par.size();
2781 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2782 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2784 // Remove possible empty data
2785 int dstart = interval_.nextNotIgnored(actual._dataStart);
2786 while (interval_.isOpeningPar(dstart) == 1) {
2787 interval_.addIntervall(dstart, dstart+1);
2788 int dend = interval_.findclosing(dstart+1, output_end);
2789 interval_.addIntervall(dend, dend+1);
2790 dstart = interval_.nextNotIgnored(dstart+1);
2792 if (dstart < output_end)
2793 interval_.output(os, output_end);
2794 interval_.addIntervall(actual._tokenstart, end);
2798 string splitOnKnownMacros(string par, bool isPatternString)
2801 LatexInfo li(par, isPatternString);
2802 // LYXERR(Debug::INFO, "Berfore split: " << par);
2803 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2805 DummyKey._tokensize = 0;
2806 DummyKey._dataStart = 0;
2807 DummyKey._dataEnd = par.length();
2808 DummyKey.disabled = true;
2809 int firstkeyIdx = li.getFirstKey();
2811 if (firstkeyIdx >= 0) {
2812 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2813 DummyKey._tokenstart = firstKey._tokenstart;
2815 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2816 // Use dummy firstKey
2817 firstKey = DummyKey;
2818 (void) li.setNextKey(firstkeyIdx);
2821 if (par.substr(firstKey._dataStart, 2) == "% ")
2822 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2824 nextkeyIdx = li.process(os, firstKey);
2825 while (nextkeyIdx >= 0) {
2826 // Check for a possible gap between the last
2827 // entry and this one
2828 int datastart = li.nextNotIgnored(firstKey._dataStart);
2829 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2830 if ((nextKey._tokenstart > datastart)) {
2832 firstKey._dataStart = datastart;
2833 firstKey._dataEnd = par.length();
2834 (void) li.setNextKey(nextkeyIdx);
2835 // Fake the last opened parenthesis
2836 li.setForDefaultLang(firstKey);
2837 nextkeyIdx = li.process(os, firstKey);
2840 if (nextKey.keytype != KeyInfo::isMain) {
2841 firstKey._dataStart = datastart;
2842 firstKey._dataEnd = nextKey._dataEnd+1;
2843 (void) li.setNextKey(nextkeyIdx);
2844 li.setForDefaultLang(firstKey);
2845 nextkeyIdx = li.process(os, firstKey);
2848 nextkeyIdx = li.process(os, nextKey);
2852 // Handle the remaining
2853 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2854 firstKey._dataEnd = par.length();
2856 if ((firstKey._dataStart < firstKey._dataEnd) &&
2857 (par[firstKey._dataStart] != '}')) {
2858 li.setForDefaultLang(firstKey);
2859 (void) li.process(os, firstKey);
2863 // return string definitelly impossible to match
2864 s = "\\foreignlanguage{ignore}{ }";
2868 s = par; /* no known macros found */
2869 // LYXERR(Debug::INFO, "After split: " << s);
2874 * Try to unify the language specs in the latexified text.
2875 * Resulting modified string is set to "", if
2876 * the searched tex does not contain all the features in the search pattern
2878 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2880 static Features regex_f;
2881 static int missed = 0;
2882 static bool regex_with_format = false;
2884 int parlen = par.length();
2886 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2889 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2890 // Happens to be there in case of description or labeling environment
2895 // Split the latex input into pieces which
2896 // can be digested by our search engine
2897 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2898 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2899 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2902 result = par.substr(0, parlen);
2903 if (isPatternString) {
2906 regex_f = identifyFeatures(result);
2907 string features = "";
2908 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2909 string a = it->first;
2910 regex_with_format = true;
2911 features += " " + a;
2912 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2914 LYXERR(Debug::FIND, "Identified Features" << features);
2917 } else if (regex_with_format) {
2918 Features info = identifyFeatures(result);
2919 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2920 string a = it->first;
2921 bool b = it->second;
2922 if (b && ! info[a]) {
2924 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2930 // LYXERR(Debug::INFO, "No regex formats");
2936 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2937 static int identifyClosing(string & t)
2939 int open_braces = 0;
2941 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2942 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2944 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2946 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2948 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2957 static int num_replaced = 0;
2958 static bool previous_single_replace = true;
2960 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
2963 // Handle \w properly
2964 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
2965 if (! opt.casesensitive) {
2966 popts |= QRegularExpression::CaseInsensitiveOption;
2968 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
2969 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
2971 if (regexp.isValid() && regexp2.isValid()) {
2972 regexIsValid = true;
2973 // Check '{', '}' pairs inside the regex
2976 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
2977 char c = par_as_string[i];
2984 else if (c == '}') {
2991 if (balanced != 0) {
2992 regexIsValid = false;
2993 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
2997 regexIsValid = false;
2998 if (!regexp.isValid())
2999 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3001 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3004 if (opt.casesensitive) {
3005 regexp = regex(regexp_str);
3006 regexp2 = regex(regexp2_str);
3009 regexp = regex(regexp_str, std::regex_constants::icase);
3010 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3015 static void modifyRegexForMatchWord(string &t)
3018 regex wordre("(\\\\)*((\\.|\\\\b))");
3021 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3023 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3026 else if (sub.str(2) == "\\\\b")
3028 if (lastpos < (size_t) sub.position(2))
3029 s += t.substr(lastpos, sub.position(2) - lastpos);
3031 lastpos = sub.position(2) + sub.length(2);
3035 else if (lastpos < t.length())
3036 s += t.substr(lastpos, t.length() - lastpos);
3037 t = "\\b" + s + "\\b";
3040 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3041 : p_buf(&buf), p_first_buf(&buf), opt(opt)
3043 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3044 docstring const & ds = stringifySearchBuffer(find_buf, opt);
3045 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3046 if (opt.replace_all && previous_single_replace) {
3047 previous_single_replace = false;
3050 else if (!opt.replace_all) {
3051 num_replaced = 0; // count number of replaced strings
3052 previous_single_replace = true;
3054 // When using regexp, braces are hacked already by escape_for_regex()
3055 par_as_string = normalize(ds);
3057 close_wildcards = 0;
3059 size_t lead_size = 0;
3060 // correct the language settings
3061 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
3062 opt.matchstart = false;
3064 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
3065 if (opt.ignoreformat) {
3069 lead_size = identifyLeading(par_as_string);
3071 lead_as_string = par_as_string.substr(0, lead_size);
3072 string lead_as_regex_string = string2regex(lead_as_string);
3073 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3074 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3075 /* Handle whole words too in this case
3077 if (opt.matchword) {
3078 par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3079 opt.matchword = false;
3081 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3082 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3083 CreateRegexp(opt, regexp_str, regexp2_str);
3085 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3086 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3090 if (!opt.ignoreformat) {
3091 lead_size = identifyLeading(par_as_string);
3092 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
3093 lead_as_string = par_as_string.substr(0, lead_size);
3094 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3097 // Here we are using regexp
3098 LASSERT(use_regexp, /**/);
3100 string lead_as_regexp;
3101 if (lead_size > 0) {
3102 lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3103 regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3104 par_as_string = par_as_string_nolead;
3105 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
3106 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3108 LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3109 par_as_string = escape_for_regex(par_as_string);
3110 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3111 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3112 LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
3116 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
3117 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
3118 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
3119 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
3120 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
3121 || regex_replace(par_as_string, par_as_string,
3122 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
3123 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
3124 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
3129 size_t lng = par_as_string.size();
3130 if (!opt.ignoreformat) {
3131 // Remove extra '\}' at end if not part of \{\.\}
3133 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3135 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3144 if (lng < par_as_string.size())
3145 par_as_string = par_as_string.substr(0,lng);
3147 if ((lng > 0) && (par_as_string[0] == '^')) {
3148 par_as_string = par_as_string.substr(1);
3150 opt.matchstart = true;
3152 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3153 LYXERR(Debug::FIND, "Open braces: " << open_braces);
3154 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3156 // If entered regexp must match at begin of searched string buffer
3157 // Kornel: Added parentheses to use $1 for size of the leading string
3161 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3162 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3163 // so the convert has no effect in that case
3164 for (int i = 7; i > 0; --i) {
3165 string orig = "\\\\" + std::to_string(i);
3166 string dest = "\\" + std::to_string(i+2);
3167 while (regex_replace(par_as_string, par_as_string, orig, dest));
3169 /* opt.matchword is ignored if using regex
3170 so expanding par_as_string with "\\b" seems appropriate here
3171 if regex contains for instance '.*' or '.+'
3172 1.) Nothing to do, if 'par_as_string' contains "\\b" already.
3173 (Means, that the user knows how to handle whole words
3174 2.) else replace '.' with "\\S" and wrap the regex with "\\b"
3176 if (opt.matchword) {
3177 modifyRegexForMatchWord(par_as_string);
3178 opt.matchword = false;
3180 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3181 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3183 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3184 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3185 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3190 // Count number of characters in string
3194 // \\[a-zA-Z]+ ==> 1
3196 static int computeSize(QStringRef s, int len)
3197 #define isLyxAlpha(arg) arg.isLetter()
3199 static int computeSize(string s, int len)
3200 #define isLyxAlpha(arg) isalpha(arg)
3207 for (int i = 0; i < len; i += skip, count++) {
3208 if (s.at(i) == '\\') {
3210 if (i + 1 < len && isLyxAlpha(s.at(i+1))) {
3211 for (int j = 2; i+j < len; j++) {
3212 if (! isLyxAlpha(s.at(i+j))) {
3213 if (s.at(i+j) == ' ')
3215 else if (s.at(i+j) == '{') {
3216 if (i+j+1 < len && s.at(i+j+1) == '}')
3218 else if (i + j + 1 >= len)
3227 else if (s.at(i) == '{') {
3228 if (i + 1 < len && s.at(i+1) == '}')
3233 else if (s.at(i) == '-') {
3234 if (i+1 < len && s.at(i+1) == '-') {
3235 if (i + 2 < len && s.at(i+2) == '-')
3251 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3256 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3259 docstring docstr = stringifyFromForSearch(opt, cur, len);
3261 if (use_regexp || opt.casesensitive)
3262 str = normalize(docstr);
3264 str = normalize(lowercase(docstr));
3265 if (!opt.ignoreformat) {
3266 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3269 mres.match_len = -1;
3272 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
3273 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
3276 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3278 QString qstr = QString::fromStdString(str);
3279 QRegularExpression const *p_regexp;
3280 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3284 p_regexp = ®exp2;
3286 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3287 if (!match.hasMatch())
3290 regex const *p_regexp;
3291 regex_constants::match_flag_type flags;
3293 flags = regex_constants::match_continuous;
3296 flags = regex_constants::match_default;
3297 p_regexp = ®exp2;
3299 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3300 if (re_it == sregex_iterator())
3302 match_results<string::const_iterator> const & m = *re_it;
3304 // Whole found string, including the leading
3305 // std: m[0].second - m[0].first
3306 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3308 // Size of the leading string
3309 // std: m[1].second - m[1].first
3310 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3311 int leadingsize = 0;
3313 if (match.lastCapturedIndex() > 0) {
3314 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3319 leadingsize = m[1].second - m[1].first;
3323 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3324 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3325 // because of different number of closing at end of string
3326 // we have to 'unify' the length of the post-match.
3327 // Done by ignoring closing parenthesis and linefeeds at string end
3328 int matchend = match.capturedEnd(0);
3329 while (mres.match_len > 0) {
3330 QChar c = qstr.at(matchend - 1);
3331 if ((c == '\n') || (c == '}') || (c == '{')) {
3338 size_t strsize = qstr.size();
3339 while (strsize > (size_t) match.capturedEnd(0)) {
3340 QChar c = qstr.at(strsize-1);
3341 if ((c == '\n') || (c == '}')) {
3347 // LYXERR0(qstr.toStdString());
3348 mres.match2end = strsize - matchend;
3349 mres.pos = match.capturedStart(2);
3351 mres.match_prefix = m[2].second - m[2].first;
3352 mres.match_len = m[0].second - m[2].second;
3353 // ignore closing parenthesis and linefeeds at string end
3354 size_t strend = m[0].second - m[0].first;
3355 int matchend = strend;
3356 while (mres.match_len > 0) {
3357 char c = str.at(matchend - 1);
3358 if ((c == '\n') || (c == '}') || (c == '{')) {
3365 size_t strsize = str.size();
3366 while (strsize > strend) {
3367 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3374 mres.match2end = strsize - matchend;
3375 mres.pos = m[2].first - m[0].first;;
3377 if (mres.match2end < 0)
3379 mres.leadsize = leadingsize;
3381 if (mres.match_len > 0) {
3382 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3383 mres.result.push_back(a0);
3384 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3385 mres.result.push_back(match.captured(i).toStdString());
3389 if (mres.match_len > 0) {
3390 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3391 mres.result.push_back(a0);
3392 for (size_t i = 3; i < m.size(); i++) {
3393 mres.result.push_back(m[i]);
3400 // else !use_regexp: but all code paths above return
3401 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
3402 << par_as_string << "', str='" << str << "'");
3403 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
3404 << lead_as_string << "', par_as_string_nolead='"
3405 << par_as_string_nolead << "'");
3408 LYXERR(Debug::FIND, "size=" << par_as_string.size()
3409 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
3410 if (str.substr(0, par_as_string.size()) == par_as_string) {
3411 mres.match_len = par_as_string.size();
3412 mres.match2end = str.size();
3417 // Start the search _after_ the leading part
3418 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3419 if (pos != string::npos) {
3420 mres.match_len = par_as_string.size();
3421 mres.match2end = str.size() - pos;
3430 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3432 MatchResult mres = findAux(cur, len, at_begin);
3433 int res = mres.match_len;
3435 "res=" << res << ", at_begin=" << at_begin
3436 << ", matchstart=" << opt.matchstart
3437 << ", inTexted=" << cur.inTexted());
3438 if (opt.matchstart) {
3441 else if (mres.match_prefix > 0)
3447 /* DEAD CODE follows
3448 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3450 if ((len > 0) && (res < len)) {
3454 Paragraph const & par = cur.paragraph();
3455 bool ws_left = (cur.pos() > 0)
3456 ? par.isWordSeparator(cur.pos() - 1)
3462 ws_right = (cur.pos() + len < par.size())
3463 ? par.isWordSeparator(cur.pos() + len)
3467 "cur.pos()=" << cur.pos() << ", res=" << res
3468 << ", separ: " << ws_left << ", " << ws_right
3471 if (ws_left && ws_right) {
3472 // Check for word separators inside the found 'word'
3473 for (int i = 0; i < len; i++) {
3474 if (par.isWordSeparator(cur.pos() + i)) {
3487 static bool simple_replace(string &t, string from, string to)
3489 regex repl("(\\\\)*(" + from + ")");
3493 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3495 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3497 if (lastpos < (size_t) sub.position(2))
3498 s += t.substr(lastpos, sub.position(2) - lastpos);
3500 lastpos = sub.position(2) + sub.length(2);
3504 else if (lastpos < t.length())
3505 s += t.substr(lastpos, t.length() - lastpos);
3511 string MatchStringAdv::normalize(docstring const & s) const
3514 t = lyx::to_utf8(s);
3515 // Remove \n at begin
3516 while (!t.empty() && t[0] == '\n')
3519 while (!t.empty() && t[t.size() - 1] == '\n')
3520 t = t.substr(0, t.size() - 1);
3522 // Handle all other '\n'
3523 while ((pos = t.find("\n")) != string::npos) {
3524 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3526 if (isAlnumASCII(t[pos+1])) {
3527 t.replace(pos-2, 3, " ");
3530 t.replace(pos-2, 3, "");
3533 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3534 // '\n' adjacent to non-alpha-numerics, discard
3535 t.replace(pos, 1, "");
3538 // Replace all other \n with spaces
3539 t.replace(pos, 1, " ");
3542 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3543 // Kornel: Added textsl, textsf, textit, texttt and noun
3544 // + allow to seach for colored text too
3545 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3546 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3547 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3548 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3549 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3550 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3556 docstring stringifyFromCursor(DocIterator const & cur, int len)
3558 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3559 if (cur.inTexted()) {
3560 Paragraph const & par = cur.paragraph();
3561 // TODO what about searching beyond/across paragraph breaks ?
3562 // TODO Try adding a AS_STR_INSERTS as last arg
3563 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3564 int(par.size()) : cur.pos() + len;
3565 // OutputParams runparams(&cur.buffer()->params().encoding());
3566 OutputParams runparams(encodings.fromLyXName("utf8"));
3567 runparams.nice = true;
3568 runparams.flavor = Flavor::XeTeX;
3569 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3570 // No side effect of file copying and image conversion
3571 runparams.dryrun = true;
3572 runparams.for_search = true;
3573 LYXERR(Debug::FIND, "Stringifying with cur: "
3574 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3575 return par.asString(cur.pos(), end,
3576 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3578 } else if (cur.inMathed()) {
3579 CursorSlice cs = cur.top();
3580 MathData md = cs.cell();
3581 MathData::const_iterator it_end =
3582 (( len == -1 || cs.pos() + len > int(md.size()))
3584 : md.begin() + cs.pos() + len );
3586 for (MathData::const_iterator it = md.begin() + cs.pos();
3589 docstring s = asString(md2);
3590 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3593 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3598 /** Computes the LaTeX export of buf starting from cur and ending len positions
3599 * after cur, if len is positive, or at the paragraph or innermost inset end
3602 docstring latexifyFromCursor(DocIterator const & cur, int len)
3604 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3605 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3606 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3607 Buffer const & buf = *cur.buffer();
3609 odocstringstream ods;
3611 //OutputParams runparams(&buf.params().encoding());
3612 OutputParams runparams(encodings.fromLyXName("utf8"));
3613 runparams.nice = false;
3614 runparams.flavor = Flavor::XeTeX;
3615 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3616 // No side effect of file copying and image conversion
3617 runparams.dryrun = true;
3618 runparams.for_search = true;
3620 if (cur.inTexted()) {
3621 // @TODO what about searching beyond/across paragraph breaks ?
3622 pos_type endpos = cur.paragraph().size();
3623 if (len != -1 && endpos > cur.pos() + len)
3624 endpos = cur.pos() + len;
3625 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3626 string(), cur.pos(), endpos);
3627 string s = lyx::to_utf8(ods.str());
3628 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3629 return(lyx::from_utf8(s));
3630 } else if (cur.inMathed()) {
3631 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3632 for (int s = cur.depth() - 1; s >= 0; --s) {
3633 CursorSlice const & cs = cur[s];
3634 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3635 TeXMathStream ws(os);
3636 cs.asInsetMath()->asHullInset()->header_write(ws);
3641 CursorSlice const & cs = cur.top();
3642 MathData md = cs.cell();
3643 MathData::const_iterator it_end =
3644 ((len == -1 || cs.pos() + len > int(md.size()))
3646 : md.begin() + cs.pos() + len);
3648 for (MathData::const_iterator it = md.begin() + cs.pos();
3652 ods << asString(md2);
3653 // Retrieve the math environment type, and add '$' or '$]'
3654 // or others (\end{equation}) accordingly
3655 for (int s = cur.depth() - 1; s >= 0; --s) {
3656 CursorSlice const & cs2 = cur[s];
3657 InsetMath * inset = cs2.asInsetMath();
3658 if (inset && inset->asHullInset()) {
3659 TeXMathStream ws(os);
3660 inset->asHullInset()->footer_write(ws);
3664 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3666 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3671 #if defined(ResultsDebug)
3673 static void displayMResult(MatchResult &mres, int increment, string from)
3675 LYXERR0( "from:\t\t\t" << from);
3676 LYXERR0( "pos: " << mres.pos << " increment " << increment);
3677 LYXERR0( "leadsize: " << mres.leadsize);
3678 LYXERR0( "match_len: " << mres.match_len);
3679 LYXERR0( "match_prefix: " << mres.match_prefix);
3680 LYXERR0( "match2end: " << mres.match2end);
3681 LYXERR0( "pos_len: " << mres.pos_len); // Set in finalize
3682 for (size_t i = 0; i < mres.result.size(); i++)
3683 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3685 #define displayMres(s,i, txt) displayMResult(s,i, xtx);
3687 #define displayMres(s,i, txt)
3691 * Not good, we miss possible matches containing also characters not found in
3692 * the innermost depth.
3693 static bool findAdvForwardInnermost(DocIterator & cur)
3696 DocIterator old_cur = cur;
3697 int forwardCount = 0;
3705 if (cur.depth() > d) {
3709 if (cur.depth() == d)
3713 if (forwardCount > 0) {
3714 LYXERR(Debug::FIND, "Forwarded " << forwardCount << " step(s) (searching for innermost match)");
3722 /** Finalize an advanced find operation, advancing the cursor to the innermost
3723 ** position that matches, plus computing the length of the matching text to
3725 ** Return the cur.pos() difference between start and end of found match
3727 MatchResult &findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
3729 // Search the foremost position that matches (avoids find of entire math
3730 // inset when match at start of it)
3731 DocIterator old_cur(cur.buffer());
3733 static MatchResult fail = MatchResult();
3734 static MatchResult max_match;
3735 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3736 // Happens with e.g. hyperlinks
3737 // either one sees "http://www.bla.bla" or nothing
3738 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3739 // and although we search for only 3 chars, we find the whole hyperlink inset
3740 bool at_begin = (expected.match_prefix == 0);
3741 //if (findAdvForwardInnermost(cur)) {
3742 if (expected.match_len > 0) {
3744 max_match = expected;
3746 size_t d = cur.depth();
3750 if (cur.depth() < d)
3752 if ((cur.depth() == d) && at_begin)
3754 mres = match(cur, -1, at_begin);
3755 displayMres(mres, -1, "Checking innermost");
3756 if (mres.match_len < expected.match_len)
3758 if (!at_begin && (mres.match_prefix > max_match.match_prefix))
3761 at_begin = (max_match.match_prefix == 0);
3767 // (expected.match_len <= 0)
3768 mres = match(cur); /* match valid only if not searching whole words */
3769 displayMres(mres, 0, "Start with negative match");
3772 if (max_match.match_len <= 0) return fail;
3773 LYXERR(Debug::FIND, "Ok");
3775 // Compute the match length
3777 if (cur.pos() + len > cur.lastpos())
3779 // regexp should use \w+, \S+, or \b(some string)\b
3780 // to search for whole words
3781 if (match.opt.matchword && !match.use_regexp) {
3782 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3783 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3785 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3787 // Length of matched text (different from len param)
3788 static MatchResult old_match = match(cur, len, at_begin);
3789 if (old_match.match_len < 0)
3791 MatchResult new_match;
3792 // Greedy behaviour while matching regexps
3793 while ((new_match = match(cur, len + 1, at_begin)).match_len > old_match.match_len) {
3795 old_match = new_match;
3796 LYXERR(Debug::FIND, "verifying match with len = " << len);
3802 int maxl = cur.lastpos() - cur.pos();
3803 // Greedy behaviour while matching regexps
3804 while (maxl > minl) {
3806 mres2 = match(cur, len, at_begin);
3807 displayMres(mres2, len, "Finalize loop");
3808 int actual_match = mres2.match_len;
3809 if (actual_match >= max_match.match_len) {
3810 // actual_match > max_match _can_ happen,
3811 // if the search area splits
3812 // some following word so that the regex
3813 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3815 // This means, the len value is too big
3817 if (maxl - minl < 4)
3818 len = (int)((maxl + minl)/2);
3820 len = (int)(minl + (maxl - minl + 3)/4);
3823 // (actual_match < max_match)
3825 len = (int)((maxl + minl)/2);
3829 // Search for real start of matched characters
3831 MatchResult actual_match;
3834 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3835 if (cur.depth() < old_cur.depth()) {
3837 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3840 if (cur.pos() != old_cur.pos()) {
3841 // OK, forwarded 1 pos in actual inset
3842 actual_match = match(cur, len-1, at_begin);
3843 if (actual_match.match_len == max_match.match_len) {
3844 // Ha, got it! The shorter selection has the same match length
3849 // OK, the shorter selection matches less chars, revert to previous value
3855 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3856 actual_match = match(cur, len, at_begin);
3857 if (actual_match.match_len == max_match.match_len) {
3859 max_match = actual_match;
3866 max_match.pos_len = len;
3873 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3877 while (!theApp()->longOperationCancelled() && cur) {
3878 //(void) findAdvForwardInnermost(cur);
3879 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3880 MatchResult mres = match(cur, -1, false);
3881 displayMres(mres,-1, "Starting findForwardAdv")
3882 int match_len = mres.match_len;
3883 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3884 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3887 if (match_len > 0) {
3888 // Try to find the begin of searched string
3890 int firstInvalid = 100000;
3891 if (mres.match_prefix + mres.pos - mres.leadsize > 1)
3892 increment = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3895 LYXERR(Debug::FIND, "Set increment to " << increment);
3896 while (increment > 0) {
3897 DocIterator old_cur = cur;
3898 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3900 if (! cur || (cur.pit() > old_cur.pit())) {
3901 // Are we outside of the paragraph?
3902 // This can happen if moving past some UTF8-encoded chars
3907 MatchResult mres2 = match(cur, -1, false);
3908 displayMres(mres2,increment, "findForwardAdv loop")
3909 switch (interpretMatch(mres, mres2)) {
3910 case MatchResult::newIsTooFar:
3911 // behind the expected match
3912 firstInvalid = increment;
3916 case MatchResult::newIsBetter:
3917 // not reached ye, but cur.pos()+increment is bettert
3919 firstInvalid -= increment;
3920 if (increment > firstInvalid*3/4)
3921 increment = firstInvalid*3/4;
3922 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3923 if (increment >= mres2.match_prefix)
3924 increment = (mres2.match_prefix+1)*3/4;
3929 // Handle not like MatchResult::newIsTooFar
3930 // LYXERR0( "Something is wrong: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3932 increment = increment*3/4;
3938 // LYXERR0("Leaving first loop");
3940 LYXERR(Debug::FIND, "Finalizing 1");
3941 MatchResult found_match = findAdvFinalize(cur, match, mres);
3942 if (found_match.match_len > 0) {
3943 LASSERT(found_match.pos_len > 0, /**/);
3944 match.FillResults(found_match);
3945 return found_match.pos_len;
3948 // try next possible match
3953 // The following code is newer reached
3954 // but parts of it may be needed in future
3955 int match_len_zero_count = 0;
3957 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3959 mres3 = match(cur, -1, false);
3960 displayMres(mres3, 1, 1 "Prepare finalize in findForwardAdv")
3961 int remaining_len = mres3.match_len;
3962 if (remaining_len <= 0) {
3963 // Apparently the searched string is not in the remaining part
3970 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3971 mres3 = match(cur, 1);
3972 displayMres(mres3, 1, "Prepare 2 finalize in findForwardAdv")
3973 int match_len3 = mres3.match_len;
3977 displayMres(mres3, 1, "Prepare 3 finalize in findForwardAdv")
3978 int match_len2 = mres3.match_len;
3979 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3980 if (match_len2 > 0) {
3981 // Sometimes in finalize we understand it wasn't a match
3982 // and we need to continue the outest loop
3983 LYXERR(Debug::FIND, "Finalizing 2");
3984 MatchResult mres4 = findAdvFinalize(cur, match, mres.match_len);
3985 if (mres4.match_len > 0) {
3986 match.FillResults(mres4);
3987 LASSERT(mres4.pos_len > 0, /**/);
3988 return mres4.pos_len;
3992 match_len_zero_count = 0;
3993 else if (match_len2 == 0)
3994 match_len_zero_count++;
3995 if (match_len2 < 0) {
3996 if (++match_len_zero_count > 3) {
3997 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
4005 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
4006 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
4009 // This should exit nested insets, if any, or otherwise undefine the currsor.
4010 cur.pos() = cur.lastpos();
4011 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
4019 /// Find the most backward consecutive match within same paragraph while searching backwards.
4020 MatchResult &findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
4022 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4023 DocIterator tmp_cur = cur;
4024 static MatchResult mr = findAdvFinalize(tmp_cur, match, MatchResult(-1));
4025 Inset & inset = cur.inset();
4026 for (; cur != cur_begin; cur.backwardPos()) {
4027 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
4028 DocIterator new_cur = cur;
4029 new_cur.backwardPos();
4030 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
4032 MatchResult new_mr = findAdvFinalize(new_cur, match, MatchResult(-1));
4033 if (new_mr.match_len == mr.match_len)
4037 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
4043 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
4047 // Backup of original position
4048 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4049 if (cur == cur_begin)
4052 DocIterator cur_orig(cur);
4053 bool pit_changed = false;
4056 bool found_match = (match(cur, -1, false).match_len > 0);
4060 cur.pos() = cur.lastpos();
4062 cur.pos() = cur_orig.pos();
4063 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
4064 DocIterator cur_prev_iter;
4066 found_match = (match(cur).match_len > 0);
4067 LYXERR(Debug::FIND, "findBackAdv3: found_match="
4068 << found_match << ", cur: " << cur);
4070 MatchResult found_mr = findMostBackwards(cur, match);
4071 match.FillResults(found_mr);
4072 LASSERT(found_mr.pos_len > 0, /**/);
4073 return found_mr.pos_len;
4076 // Stop if begin of document reached
4077 if (cur == cur_begin)
4079 cur_prev_iter = cur;
4083 if (cur == cur_begin)
4090 } while (!theApp()->longOperationCancelled());
4098 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4099 DocIterator const & cur, int len)
4101 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4103 if (!opt.ignoreformat)
4104 return latexifyFromCursor(cur, len);
4106 return stringifyFromCursor(cur, len);
4110 FindAndReplaceOptions::FindAndReplaceOptions(
4111 docstring const & _find_buf_name, bool _casesensitive,
4112 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4113 docstring const & _repl_buf_name, bool _keep_case,
4114 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4115 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4116 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4117 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4125 /** Check if 'len' letters following cursor are all non-lowercase */
4126 static bool allNonLowercase(Cursor const & cur, int len)
4128 pos_type beg_pos = cur.selectionBegin().pos();
4129 pos_type end_pos = cur.selectionBegin().pos() + len;
4130 if (len > cur.lastpos() + 1 - beg_pos) {
4131 LYXERR(Debug::FIND, "This should not happen, more debug needed");
4132 len = cur.lastpos() + 1 - beg_pos;
4133 end_pos = beg_pos + len;
4135 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4136 if (isLowerCase(cur.paragraph().getChar(pos)))
4142 /** Check if first letter is upper case and second one is lower case */
4143 static bool firstUppercase(Cursor const & cur)
4146 pos_type pos = cur.selectionBegin().pos();
4147 if (pos >= cur.lastpos() - 1) {
4148 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
4151 ch1 = cur.paragraph().getChar(pos);
4152 ch2 = cur.paragraph().getChar(pos + 1);
4153 bool result = isUpperCase(ch1) && isLowerCase(ch2);
4154 LYXERR(Debug::FIND, "firstUppercase(): "
4155 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4156 << ch2 << "(" << char(ch2) << ")"
4157 << ", result=" << result << ", cur=" << cur);
4162 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4164 ** \fixme What to do with possible further paragraphs in replace buffer ?
4166 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4168 ParagraphList::iterator pit = buffer.paragraphs().begin();
4169 LASSERT(!pit->empty(), /**/);
4170 pos_type right = pos_type(1);
4171 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4172 right = pit->size();
4173 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4178 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4180 // Should replace the string "$" + std::to_string(matchnum) with replacement
4181 // if the char '$' is not prefixed with odd number of char '\\'
4182 static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4186 for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4188 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4190 int num = stoi(sub.str(3), nullptr, 10);
4191 if (num >= maxmatchnum)
4193 if (lastpos < (size_t) sub.position(2))
4194 s += t.substr(lastpos, sub.position(2) - lastpos);
4195 s += replacements[num];
4196 lastpos = sub.position(2) + sub.length(2);
4200 else if (lastpos < t.length())
4201 s += t.substr(lastpos, t.length() - lastpos);
4208 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4210 Cursor & cur = bv->cursor();
4211 if (opt.repl_buf_name.empty()
4212 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4213 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4216 DocIterator sel_beg = cur.selectionBegin();
4217 DocIterator sel_end = cur.selectionEnd();
4218 if (&sel_beg.inset() != &sel_end.inset()
4219 || sel_beg.pit() != sel_end.pit()
4220 || sel_beg.idx() != sel_end.idx())
4222 int sel_len = sel_end.pos() - sel_beg.pos();
4223 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4224 << ", sel_len: " << sel_len << endl);
4227 LASSERT(sel_len > 0, return 0);
4229 if (!matchAdv(sel_beg, sel_len).match_len)
4232 // Build a copy of the replace buffer, adapted to the KeepCase option
4233 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4235 repl_buffer_orig.write(oss);
4236 string lyx = oss.str();
4237 if (matchAdv.valid_matches > 0) {
4238 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4240 Buffer repl_buffer("", false);
4241 repl_buffer.setUnnamed(true);
4242 LASSERT(repl_buffer.readString(lyx), return 0);
4243 if (opt.keep_case && sel_len >= 2) {
4244 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4245 if (cur.inTexted()) {
4246 if (firstUppercase(cur))
4247 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4248 else if (allNonLowercase(cur, sel_len))
4249 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4252 cap::cutSelection(cur, false);
4253 if (cur.inTexted()) {
4254 repl_buffer.changeLanguage(
4255 repl_buffer.language(),
4256 cur.getFont().language());
4257 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4258 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4259 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4260 repl_buffer.params().documentClassPtr(),
4261 bv->buffer().errorList("Paste"));
4262 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4263 sel_len = repl_buffer.paragraphs().begin()->size();
4264 } else if (cur.inMathed()) {
4265 odocstringstream ods;
4267 // OutputParams runparams(&repl_buffer.params().encoding());
4268 OutputParams runparams(encodings.fromLyXName("utf8"));
4269 runparams.nice = false;
4270 runparams.flavor = Flavor::XeTeX;
4271 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4272 runparams.dryrun = true;
4273 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4274 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4275 docstring repl_latex = ods.str();
4276 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4278 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4279 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4280 repl_latex = from_utf8(s);
4281 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4282 MathData ar(cur.buffer());
4283 asArray(repl_latex, ar, Parse::NORMAL);
4285 sel_len = ar.size();
4286 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4288 if (cur.pos() >= sel_len)
4289 cur.pos() -= sel_len;
4292 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4293 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4294 bv->processUpdateFlags(Update::Force);
4299 /// Perform a FindAdv operation.
4300 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4305 // e.g., when invoking word-findadv from mini-buffer wither with
4306 // wrong options syntax or before ever opening advanced F&R pane
4307 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4311 MatchStringAdv matchAdv(bv->buffer(), opt);
4313 if (!matchAdv.regexIsValid) {
4314 bv->message(lyx::from_utf8(matchAdv.regexError));
4318 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4320 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4321 num_replaced += findAdvReplace(bv, opt, matchAdv);
4324 pos_len = findForwardAdv(cur, matchAdv);
4326 pos_len = findBackwardsAdv(cur, matchAdv);
4327 } catch (exception & ex) {
4328 bv->message(from_utf8(ex.what()));
4333 if (num_replaced > 0) {
4334 switch (num_replaced)
4337 bv->message(_("One match has been replaced."));
4340 bv->message(_("Two matches have been replaced."));
4343 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4349 bv->message(_("Match not found."));
4354 if (num_replaced > 0)
4355 bv->message(_("Match has been replaced."));
4357 bv->message(_("Match found."));
4359 if (cur.pos() + pos_len > cur.lastpos()) {
4360 // Prevent crash in bv->putSelectionAt()
4361 // Should never happen, maybe LASSERT() here?
4362 pos_len = cur.lastpos() - cur.pos();
4364 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4365 bv->putSelectionAt(cur, pos_len, !opt.forward);
4371 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4373 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4374 << opt.casesensitive << ' '
4375 << opt.matchword << ' '
4376 << opt.forward << ' '
4377 << opt.expandmacros << ' '
4378 << opt.ignoreformat << ' '
4379 << opt.replace_all << ' '
4380 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4381 << opt.keep_case << ' '
4382 << int(opt.scope) << ' '
4385 LYXERR(Debug::FIND, "built: " << os.str());
4391 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4393 LYXERR(Debug::FIND, "parsing");
4397 while (line != "EOSS") {
4401 if (is.eof()) // Tolerate malformed request
4405 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4406 opt.find_buf_name = from_utf8(s);
4407 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4408 is.get(); // Waste space before replace string
4411 while (line != "EOSS") {
4415 if (is.eof()) // Tolerate malformed request
4419 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4420 opt.repl_buf_name = from_utf8(s);
4421 is >> opt.keep_case;
4424 opt.scope = FindAndReplaceOptions::SearchScope(i);
4426 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4428 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4429 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4430 << opt.scope << ' ' << opt.restr);