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) ?|[\\[\\]\\{\\}]))");
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")
780 replace = sub.str(3);
782 if (lastpos < (size_t) sub.position(2))
783 s += t.substr(lastpos, sub.position(2) - lastpos);
785 lastpos = sub.position(2) + sub.length(2);
789 else if (lastpos < t.length())
790 s += t.substr(lastpos, t.length() - lastpos);
794 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
795 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
796 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
797 string escape_for_regex(string s)
801 while (lastpos < s.size()) {
802 size_t regex_pos = s.find("\\regexp{", lastpos);
803 if (regex_pos == string::npos) {
804 regex_pos = s.size();
806 if (regex_pos > lastpos) {
807 result += string2regex(s.substr(lastpos, regex_pos-lastpos));
809 if (lastpos == s.size())
812 size_t end_pos = s.find("\\endregexp{}}", regex_pos + 8);
813 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)));
814 lastpos = end_pos + 13;
820 /// Wrapper for lyx::regex_replace with simpler interface
821 bool regex_replace(string const & s, string & t, string const & searchstr,
822 string const & replacestr)
824 regex e(searchstr, regex_constants::ECMAScript);
826 ostream_iterator<char, char> it(oss);
827 regex_replace(it, s.begin(), s.end(), e, replacestr);
828 // tolerate t and s be references to the same variable
829 bool rv = (s != oss.str());
835 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
837 ** Verify that closed braces exactly match open braces. This avoids that, for example,
838 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
841 ** Number of open braces that must remain open at the end for the verification to succeed.
844 bool braces_match(QString const & beg,
847 bool braces_match(string const & beg,
853 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg.toStdString() << "'");
855 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg << "'");
857 int lastidx = beg.size();
858 for (int i=0; i < lastidx; ++i) {
859 // Skip escaped braces in the count
869 } else if (c == '{') {
871 } else if (c == '}') {
872 if (open_pars == 0) {
873 LYXERR(Debug::FIND, "Found unmatched closed brace");
879 if (open_pars != unmatched) {
880 LYXERR(Debug::FIND, "Found " << open_pars
881 << " instead of " << unmatched
882 << " unmatched open braces at the end of count");
885 LYXERR(Debug::FIND, "Braces match as expected");
903 vector <string> result = vector <string>();
904 MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1) {};
907 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
909 int range = oldres.match_len;
910 if (range < 2) range = 2;
911 if (newres.match2end < oldres.match2end - oldres.match_len)
912 return MatchResult::newIsTooFar;
913 if (newres.match_len < oldres.match_len)
914 return MatchResult::newIsTooFar;
915 if ((newres.match_len == oldres.match_len) &&
916 (newres.match2end < oldres.match2end + range) &&
917 (newres.match2end > oldres.match2end - range)) {
918 return MatchResult::newIsBetter;
920 return MatchResult::newIsInvalid;
923 /** The class performing a match between a position in the document and the FindAdvOptions.
926 class MatchStringAdv {
928 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
930 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
931 ** constructor as opt.search, under the opt.* options settings.
934 ** If set, then match is searched only against beginning of text starting at cur.
935 ** If unset, then match is searched anywhere in text starting at cur.
938 ** The length of the matching text, or zero if no match was found.
940 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
949 /// first buffer on which search was started
950 lyx::Buffer * const p_first_buf;
952 FindAndReplaceOptions const & opt;
955 /// Auxiliary find method (does not account for opt.matchword)
956 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
957 void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
959 /** Normalize a stringified or latexified LyX paragraph.
963 ** <li>if search is not casesensitive, then lowercase the string;
964 ** <li>remove any newline at begin or end of the string;
965 ** <li>replace any newline in the middle of the string with a simple space;
966 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
969 ** @todo Normalization should also expand macros, if the corresponding
970 ** search option was checked.
972 string normalize(docstring const & s) const;
973 // normalized string to search
974 string par_as_string;
975 // regular expression to use for searching
976 // regexp2 is same as regexp, but prefixed with a ".*?"
978 QRegularExpression regexp;
979 QRegularExpression regexp2;
984 // leading format material as string
985 string lead_as_string;
986 // par_as_string after removal of lead_as_string
987 string par_as_string_nolead;
988 // unmatched open braces in the search string/regexp
990 // number of (.*?) subexpressions added at end of search regexp for closing
991 // environments, math mode, styles, etc...
994 // Are we searching with regular expressions ?
996 static int valid_matches;
997 static vector <string> matches;
998 void FillResults(MatchResult &found_mr);
1001 int MatchStringAdv::valid_matches = 0;
1002 vector <string> MatchStringAdv::matches = vector <string>(10);
1004 void MatchStringAdv::FillResults(MatchResult &found_mr)
1006 if (found_mr.match_len > 0) {
1007 valid_matches = found_mr.result.size();
1008 for (size_t i = 0; i < found_mr.result.size(); i++)
1009 matches[i] = found_mr.result[i];
1015 static docstring buffer_to_latex(Buffer & buffer)
1017 //OutputParams runparams(&buffer.params().encoding());
1018 OutputParams runparams(encodings.fromLyXName("utf8"));
1019 odocstringstream ods;
1021 runparams.nice = true;
1022 runparams.flavor = Flavor::XeTeX;
1023 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1024 // No side effect of file copying and image conversion
1025 runparams.dryrun = true;
1026 runparams.for_search = true;
1027 pit_type const endpit = buffer.paragraphs().size();
1028 for (pit_type pit = 0; pit != endpit; ++pit) {
1029 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
1030 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
1036 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1039 if (!opt.ignoreformat) {
1040 str = buffer_to_latex(buffer);
1042 // OutputParams runparams(&buffer.params().encoding());
1043 OutputParams runparams(encodings.fromLyXName("utf8"));
1044 runparams.nice = true;
1045 runparams.flavor = Flavor::XeTeX;
1046 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1047 runparams.dryrun = true;
1048 runparams.for_search = true;
1049 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1050 Paragraph const & par = buffer.paragraphs().at(pit);
1051 LYXERR(Debug::FIND, "Adding to search string: '"
1052 << par.asString(pos_type(0), par.size(),
1053 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1056 str += par.asString(pos_type(0), par.size(),
1057 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1060 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1061 string t = to_utf8(str);
1062 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
1069 /// Return separation pos between the leading material and the rest
1070 static size_t identifyLeading(string const & s)
1073 // @TODO Support \item[text]
1074 // Kornel: Added textsl, textsf, textit, texttt and noun
1075 // + allow to search for colored text too
1076 while (regex_replace(t, t, "^\\\\(("
1077 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1078 "lyxaddress|lyxrightaddress|"
1079 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1080 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1081 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1082 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1083 || regex_replace(t, t, "^\\$", "")
1084 || regex_replace(t, t, "^\\\\\\[", "")
1085 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1086 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1088 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1093 * Given a latexified string, retrieve some handled features
1094 * The features of the regex will later be compared with the features
1095 * of the searched text. If the regex features are not a
1096 * subset of the analized, then, in not format ignoring search
1097 * we can early stop the search in the relevant inset.
1099 typedef map<string, bool> Features;
1101 static Features identifyFeatures(string const & s)
1103 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1104 static regex const valid("^("
1106 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1107 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1108 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1109 "(u|uu)line|(s|x)out|uwave|"
1110 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1111 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1116 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1119 if (sub.str(1).compare("regexp") == 0) {
1123 string token = sub.str(1);
1125 if (regex_match(token, sub2, valid)) {
1133 if (sub.str(1).compare("endregexp") == 0) {
1143 * defines values features of a key "\\[a-z]+{"
1148 /* Char type with content discarded
1149 * like \hspace{1cm} */
1151 /* Char, like \backslash */
1153 /* replace starting backslash with '#' */
1155 /* \part, \section*, ... */
1157 /* title, author etc */
1159 /* \foreignlanguage{ngerman}, ... */
1162 * to discard language in content */
1165 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1167 /* fonts, colors, markups, ... */
1169 /* footnotesize, ... large, ...
1170 * Ignore all of them */
1173 /* inputencoding, ...
1174 * Discard also content, because they do not help in search */
1177 * like remove, but also all arguments */
1179 /* item, listitem */
1181 /* tex, latex, ... like isChar */
1183 /* like \lettrine[lines=5]{}{} */
1185 // like isStandard, but always remove head
1187 /* End of arguments marker for lettrine,
1188 * so that they can be ignored */
1191 KeyInfo() = default;
1192 KeyInfo(KeyType type, int parcount, bool disable)
1194 parenthesiscount(parcount),
1195 disabled(disable) {}
1196 KeyType keytype = invalid;
1198 int _tokensize = -1;
1199 int _tokenstart = -1;
1200 int _dataStart = -1;
1202 int parenthesiscount = 1;
1203 bool disabled = false;
1204 bool used = false; /* by pattern */
1209 Border(int l=0, int u=0) : low(l), upper(u) {};
1214 #define MAXOPENED 30
1216 bool isPatternString_;
1218 explicit Intervall(bool isPattern, string const & p) :
1219 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1220 hasTitle(false), langcount(0)
1228 static vector<Border> borders;
1229 int depts[MAXOPENED];
1230 int closes[MAXOPENED];
1231 int actualdeptindex;
1232 int previousNotIgnored(int) const;
1233 int nextNotIgnored(int) const;
1234 void handleOpenP(int i);
1235 void handleCloseP(int i, bool closingAllowed);
1236 void resetOpenedP(int openPos);
1237 void addIntervall(int upper);
1238 void addIntervall(int low, int upper); /* if explicit */
1239 void removeAccents();
1240 void setForDefaultLang(KeyInfo const & defLang) const;
1241 int findclosing(int start, int end, char up, char down, int repeat);
1242 void handleParentheses(int lastpos, bool closingAllowed);
1244 int langcount; // Number of disabled language specs up to current position in actual interval
1245 int isOpeningPar(int pos) const;
1247 void output(ostringstream &os, int lastpos);
1248 // string show(int lastpos);
1251 vector<Border> Intervall::borders = vector<Border>(30);
1253 int Intervall::isOpeningPar(int pos) const
1255 if ((pos < 0) || (size_t(pos) >= par.size()))
1257 if (par[pos] != '{')
1259 if (size_t(pos) + 2 >= par.size())
1261 if (par[pos+2] != '}')
1263 if (par[pos+1] == '[' || par[pos+1] == ']')
1268 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1270 // Enable the use of first token again
1271 if (ignoreidx >= 0) {
1272 int value = defLang._tokenstart + defLang._tokensize;
1278 if (borders[borderidx].low < value)
1279 borders[borderidx].low = value;
1280 if (borders[borderidx].upper < value)
1281 borders[borderidx].upper = value;
1286 static void checkDepthIndex(int val)
1288 static int maxdepthidx = MAXOPENED-2;
1289 static int lastmaxdepth = 0;
1290 if (val > lastmaxdepth) {
1291 LYXERR(Debug::INFO, "Depth reached " << val);
1294 if (val > maxdepthidx) {
1296 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1301 // Not needed, because borders are now dynamically expanded
1302 static void checkIgnoreIdx(int val)
1304 static int lastmaxignore = -1;
1305 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1306 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1307 lastmaxignore = val;
1313 * Expand the region of ignored parts of the input latex string
1314 * The region is only relevant in output()
1316 void Intervall::addIntervall(int low, int upper)
1319 if (low == upper) return;
1320 for (idx = ignoreidx+1; idx > 0; --idx) {
1321 if (low > borders[idx-1].upper) {
1325 Border br(low, upper);
1326 if (idx > ignoreidx) {
1327 if (borders.size() <= size_t(idx)) {
1328 borders.push_back(br);
1334 // checkIgnoreIdx(ignoreidx);
1338 // Expand only if one of the new bound is inside the interwall
1339 // We know here that br.low > borders[idx-1].upper
1340 if (br.upper < borders[idx].low) {
1341 // We have to insert at this pos
1342 if (size_t(ignoreidx+1) >= borders.size()) {
1343 borders.push_back(borders[ignoreidx]);
1346 borders[ignoreidx+1] = borders[ignoreidx];
1348 for (int i = ignoreidx; i > idx; --i) {
1349 borders[i] = borders[i-1];
1353 // checkIgnoreIdx(ignoreidx);
1356 // Here we know, that we are overlapping
1357 if (br.low > borders[idx].low)
1358 br.low = borders[idx].low;
1359 // check what has to be concatenated
1361 for (int i = idx; i <= ignoreidx; i++) {
1362 if (br.upper >= borders[i].low) {
1364 if (br.upper < borders[i].upper)
1365 br.upper = borders[i].upper;
1371 // count should be >= 1 here
1374 for (int i = idx + count; i <= ignoreidx; i++) {
1375 borders[i-count+1] = borders[i];
1377 ignoreidx -= count - 1;
1383 typedef map<string, string> AccentsMap;
1384 static AccentsMap accents = map<string, string>();
1386 static void buildaccent(string n, string param, string values)
1390 const char delim = '|';
1391 while (getline(s, name, delim)) {
1393 for (char c : param) {
1394 string key = name + "{" + c + "}";
1395 // get the corresponding utf8-value
1396 if ((values[start] & 0xc0) != 0xc0) {
1397 // should not happen, utf8 encoding starts at least with 11xxxxxx
1398 // but value for '\dot{i}' is 'i', which is ascii
1399 if ((values[start] & 0x80) == 0) {
1401 accents[key] = values.substr(start, 1);
1402 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1407 for (int j = 1; ;j++) {
1408 if (start + j >= values.size()) {
1409 accents[key] = values.substr(start, j);
1410 start = values.size() - 1;
1413 else if ((values[start+j] & 0xc0) != 0x80) {
1414 // This is the first byte of following utf8 char
1415 accents[key] = values.substr(start, j);
1417 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1426 static string getutf8(unsigned uchar)
1429 string ret = string();
1431 if (uchar <= 0x7f) {
1432 c[maxc-1] = uchar & 0x7f;
1435 unsigned char rest = 0x40;
1436 unsigned char first = 0x80;
1438 for (int i = start; i >=0; --i) {
1440 c[i] = first + uchar;
1443 c[i] = 0x80 | (uchar & 0x3f);
1450 for (int i = 0; i < maxc; i++) {
1451 if (c[i] == 0) continue;
1457 static void buildAccentsMap()
1459 accents["imath"] = "ı";
1461 accents["jmath"] = "ȷ";
1462 accents["cdot"] = "·";
1463 accents["textasciicircum"] = "^";
1464 accents["mathcircumflex"] = "^";
1465 accents["sim"] = "~";
1466 accents["guillemotright"] = "»";
1467 accents["guillemotleft"] = "«";
1468 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1469 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1470 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1471 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1472 accents["negmedspace"] = getutf8(0xf0005);
1473 accents["thickspace"] = getutf8(0xf0006);
1474 accents["negthickspace"] = getutf8(0xf0007);
1475 accents["lyx"] = getutf8(0xf0010); // Used logos
1476 accents["LyX"] = getutf8(0xf0010);
1477 accents["tex"] = getutf8(0xf0011);
1478 accents["TeX"] = getutf8(0xf0011);
1479 accents["latex"] = getutf8(0xf0012);
1480 accents["LaTeX"] = getutf8(0xf0012);
1481 accents["latexe"] = getutf8(0xf0013);
1482 accents["LaTeXe"] = getutf8(0xf0013);
1483 accents["lyxarrow"] = getutf8(0xf0020);
1484 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1485 accents["backslash LyX"] = getutf8(0xf0010);
1486 accents["backslash tex"] = getutf8(0xf0011);
1487 accents["backslash TeX"] = getutf8(0xf0011);
1488 accents["backslash latex"] = getutf8(0xf0012);
1489 accents["backslash LaTeX"] = getutf8(0xf0012);
1490 accents["backslash latexe"] = getutf8(0xf0013);
1491 accents["backslash LaTeXe"] = getutf8(0xf0013);
1492 accents["backslash lyxarrow"] = getutf8(0xf0020);
1493 accents["ddot{\\imath}"] = "ï";
1494 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1495 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1496 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1497 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1498 accents["acute{\\imath}"] = "í";
1499 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1500 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1501 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1502 buildaccent("mathring|r", "aAuUwy",
1504 accents["check{\\imath}"] = "ǐ";
1505 accents["check{\\jmath}"] = "ǰ";
1506 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1507 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1508 accents["hat{\\imath}"] = "î";
1509 accents["hat{\\jmath}"] = "ĵ";
1510 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1511 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1512 accents["bar{\\imath}"] = "ī";
1513 buildaccent("bar|=", "aAeEiIoOuUyY",
1514 "āĀēĒīĪōŌūŪȳȲ"); // macron
1515 accents["tilde{\\imath}"] = "ĩ";
1516 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1517 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1518 accents["breve{\\imath}"] = "ĭ";
1519 buildaccent("breve|u", "aAeEgGiIoOuU",
1520 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1521 accents["grave{\\imath}"] = "ì";
1522 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1523 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1524 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1525 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1526 buildaccent("ogonek|k", "AaEeIiUuOo",
1527 "ĄąĘęĮįŲųǪǫ"); // ogonek
1528 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1529 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1530 buildaccent("subring|textsubring", "Aa",
1532 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1533 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1534 buildaccent("subtilde|textsubtilde", "EeIiUu",
1535 "ḚḛḬḭṴṵ"); // subtilde
1536 accents["dgrave{\\imath}"] = "ȉ";
1537 accents["textdoublegrave{\\i}"] = "ȉ";
1538 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1539 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1540 accents["rcap{\\imath}"] = "ȉ";
1541 accents["textroundcap{\\i}"] = "ȉ";
1542 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1543 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1544 buildaccent("slashed", "oO",
1549 * Created accents in math or regexp environment
1550 * are macros, but we need the utf8 equivalent
1552 void Intervall::removeAccents()
1554 if (accents.empty())
1556 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1557 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1558 "|((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]))");
1560 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1562 string key = sub.str(1);
1563 if (accents.find(key) != accents.end()) {
1564 string val = accents[key];
1565 size_t pos = sub.position(size_t(0));
1566 for (size_t i = 0; i < val.size(); i++) {
1567 par[pos+i] = val[i];
1569 // Remove possibly following space too
1570 if (par[pos+sub.str(0).size()] == ' ')
1571 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1573 addIntervall(pos+val.size(), pos + sub.str(0).size());
1574 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1575 // remove traces of any remaining chars
1580 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1585 void Intervall::handleOpenP(int i)
1588 depts[actualdeptindex] = i+1;
1589 closes[actualdeptindex] = -1;
1590 checkDepthIndex(actualdeptindex);
1593 void Intervall::handleCloseP(int i, bool closingAllowed)
1595 if (actualdeptindex <= 0) {
1596 if (! closingAllowed)
1597 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1598 // if we are at the very end
1599 addIntervall(i, i+1);
1602 closes[actualdeptindex] = i+1;
1607 void Intervall::resetOpenedP(int openPos)
1609 // Used as initializer for foreignlanguage entry
1610 actualdeptindex = 1;
1611 depts[1] = openPos+1;
1615 int Intervall::previousNotIgnored(int start) const
1617 int idx = 0; /* int intervalls */
1618 for (idx = ignoreidx; idx >= 0; --idx) {
1619 if (start > borders[idx].upper)
1621 if (start >= borders[idx].low)
1622 start = borders[idx].low-1;
1627 int Intervall::nextNotIgnored(int start) const
1629 int idx = 0; /* int intervalls */
1630 for (idx = 0; idx <= ignoreidx; idx++) {
1631 if (start < borders[idx].low)
1633 if (start < borders[idx].upper)
1634 start = borders[idx].upper;
1639 typedef map<string, KeyInfo> KeysMap;
1640 typedef vector< KeyInfo> Entries;
1641 static KeysMap keys = map<string, KeyInfo>();
1647 Intervall interval_;
1648 void buildKeys(bool);
1649 void buildEntries(bool);
1650 void makeKey(const string &, KeyInfo, bool isPatternString);
1651 void processRegion(int start, int region_end); /* remove {} parts */
1652 void removeHead(KeyInfo const &, int count=0);
1655 LatexInfo(string const & par, bool isPatternString)
1656 : entidx_(-1), interval_(isPatternString, par)
1658 buildKeys(isPatternString);
1659 entries_ = vector<KeyInfo>();
1660 buildEntries(isPatternString);
1664 if (entries_.empty()) {
1667 if (entries_[0].keytype == KeyInfo::isTitle) {
1668 interval_.hasTitle = true;
1669 if (! entries_[0].disabled) {
1670 interval_.titleValue = entries_[0].head;
1673 interval_.titleValue = "";
1675 removeHead(entries_[0]);
1676 if (entries_.size() > 1)
1685 if (int(entries_.size()) > entidx_) {
1692 bool setNextKey(int idx) {
1693 if ((idx == entidx_) && (entidx_ >= 0)) {
1700 int find(int start, KeyInfo::KeyType keytype) const {
1704 while (tmpIdx < int(entries_.size())) {
1705 if (entries_[tmpIdx].keytype == keytype)
1711 int process(ostringstream & os, KeyInfo const & actual);
1712 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1713 // string show(int lastpos) { return interval.show(lastpos);};
1714 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1715 KeyInfo &getKeyInfo(int keyinfo) {
1716 static KeyInfo invalidInfo = KeyInfo();
1717 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1720 return entries_[keyinfo];
1722 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1723 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1727 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1731 for (int i = start; i < end; i += 1 + skip) {
1735 if (c == '\\') skip = 1;
1739 else if (c == down) {
1742 if ((repeat <= 0) || (par[i+1] != up))
1760 vector<MathEntry> entries_;
1765 void insert(string const & wait, size_t start, size_t end) {
1766 MathEntry m = MathEntry();
1768 m.mathStart = start;
1770 m.mathSize = end - start;
1771 entries_.push_back(m);
1773 bool empty() const { return entries_.empty(); };
1774 size_t getEndPos() const {
1775 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1778 return entries_[actualIdx_].mathEnd;
1780 size_t getStartPos() const {
1781 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1782 return 100000; /* definitely enough? */
1784 return entries_[actualIdx_].mathStart;
1786 size_t getFirstPos() {
1788 return getStartPos();
1790 size_t getSize() const {
1791 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1794 return entries_[actualIdx_].mathSize;
1796 void incrEntry() { actualIdx_++; };
1799 void LatexInfo::buildEntries(bool isPatternString)
1801 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\})");
1802 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1803 static bool disableLanguageOverride = false;
1804 smatch sub, submath;
1805 bool evaluatingRegexp = false;
1807 bool evaluatingMath = false;
1808 bool evaluatingCode = false;
1810 bool evaluatingOptional = false;
1811 size_t optionalEnd = 0;
1814 bool math_end_waiting = false;
1815 size_t math_pos = 10000;
1817 static vector<string> usedText = vector<string>();
1819 interval_.removeAccents();
1821 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1823 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1824 // prefixed by odd count of '\\'
1827 if (math_end_waiting) {
1828 size_t pos = submath.position(size_t(2));
1829 if ((math_end == "$") &&
1830 (submath.str(2) == "$")) {
1831 mi.insert("$", math_pos, pos + 1);
1832 math_end_waiting = false;
1834 else if ((math_end == "\\]") &&
1835 (submath.str(2) == "\\]")) {
1836 mi.insert("\\]", math_pos, pos + 2);
1837 math_end_waiting = false;
1839 else if ((submath.str(3).compare("end") == 0) &&
1840 (submath.str(4).compare(math_end) == 0)) {
1841 mi.insert(math_end, math_pos, pos + submath.str(2).length());
1842 math_end_waiting = false;
1848 if (submath.str(3).compare("begin") == 0) {
1849 math_end_waiting = true;
1850 math_end = submath.str(4);
1851 math_pos = submath.position(size_t(2));
1853 else if (submath.str(2).compare("\\[") == 0) {
1854 math_end_waiting = true;
1856 math_pos = submath.position(size_t(2));
1858 else if (submath.str(2) == "$") {
1859 size_t pos = submath.position(size_t(2));
1860 math_end_waiting = true;
1866 // Ignore language if there is math somewhere in pattern-string
1867 if (isPatternString) {
1868 for (auto s: usedText) {
1869 // Remove entries created in previous search runs
1872 usedText = vector<string>();
1875 keys["foreignlanguage"].disabled = true;
1876 disableLanguageOverride = true;
1879 disableLanguageOverride = false;
1882 if (disableLanguageOverride) {
1883 keys["foreignlanguage"].disabled = true;
1886 math_pos = mi.getFirstPos();
1887 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1889 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1890 // prefixed by odd count of '\\'
1893 string key = sub.str(5);
1895 if (sub.str(2)[0] == '\\')
1896 key = sub.str(2)[1];
1901 if (keys.find(key) != keys.end()) {
1902 if (keys[key].keytype == KeyInfo::headRemove) {
1903 KeyInfo found1 = keys[key];
1904 found1.disabled = true;
1905 found1.head = "\\" + key + "{";
1906 found1._tokenstart = sub.position(size_t(2));
1907 found1._tokensize = found1.head.length();
1908 found1._dataStart = found1._tokenstart + found1.head.length();
1909 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1910 found1._dataEnd = endpos;
1915 if (evaluatingRegexp) {
1916 if (sub.str(3).compare("endregexp") == 0) {
1917 evaluatingRegexp = false;
1918 // found._tokenstart already set
1919 found._dataEnd = sub.position(size_t(2)) + 13;
1920 found._dataStart = found._dataEnd;
1921 found._tokensize = found._dataEnd - found._tokenstart;
1922 found.parenthesiscount = 0;
1923 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1930 if (evaluatingMath) {
1931 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1933 evaluatingMath = false;
1935 math_pos = mi.getStartPos();
1937 if (keys.find(key) == keys.end()) {
1938 found = KeyInfo(KeyInfo::isStandard, 0, true);
1939 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1940 found = KeyInfo(KeyInfo::isText, 0, false);
1941 if (isPatternString) {
1942 found.keytype = KeyInfo::isChar;
1943 found.disabled = false;
1947 usedText.push_back(key);
1951 if (key.compare("regexp") == 0) {
1952 evaluatingRegexp = true;
1953 found._tokenstart = sub.position(size_t(2));
1954 found._tokensize = 0;
1958 // Handle the other params of key
1959 if (found.keytype == KeyInfo::isIgnored)
1961 else if (found.keytype == KeyInfo::isMath) {
1962 if (size_t(sub.position(size_t(2))) == math_pos) {
1964 found._tokenstart = sub.position(size_t(2));
1965 found._tokensize = mi.getSize();
1966 found._dataEnd = found._tokenstart + found._tokensize;
1967 found._dataStart = found._dataEnd;
1968 found.parenthesiscount = 0;
1969 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1970 evaluatingMath = true;
1973 // begin|end of unknown env, discard
1974 // First handle tables
1975 // longtable|tabular
1976 bool discardComment;
1978 found.keytype = KeyInfo::doRemove;
1979 if ((sub.str(7).compare("longtable") == 0) ||
1980 (sub.str(7).compare("tabular") == 0)) {
1981 discardComment = true; /* '%' */
1984 discardComment = false;
1985 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1987 string token = sub.str(7);
1988 if (regex_match(token, sub2, removeArgs)) {
1989 found.keytype = KeyInfo::removeWithArg;
1992 // discard spaces before pos(2)
1993 int pos = sub.position(size_t(2));
1995 for (count = 0; pos - count > 0; count++) {
1996 char c = interval_.par[pos-count-1];
1997 if (discardComment) {
1998 if ((c != ' ') && (c != '%'))
2004 found._tokenstart = pos - count;
2005 if (sub.str(3).compare(0, 5, "begin") == 0) {
2006 size_t pos1 = pos + sub.str(2).length();
2007 if (sub.str(7).compare("cjk") == 0) {
2008 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2009 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2011 found.keytype = KeyInfo::isMain;
2012 found._dataStart = pos1;
2013 found._dataEnd = interval_.par.length();
2014 found.disabled = keys["foreignlanguage"].disabled;
2015 found.used = keys["foreignlanguage"].used;
2016 found._tokensize = pos1 - found._tokenstart;
2017 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2020 // Swallow possible optional params
2021 while (interval_.par[pos1] == '[') {
2022 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2024 // Swallow also the eventual parameter
2025 if (interval_.par[pos1] == '{') {
2026 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2029 found._dataEnd = pos1;
2031 found._dataStart = found._dataEnd;
2032 found._tokensize = count + found._dataEnd - pos;
2033 found.parenthesiscount = 0;
2034 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2035 found.disabled = true;
2039 // Handle "\end{...}"
2040 found._dataStart = pos + sub.str(2).length();
2041 found._dataEnd = found._dataStart;
2042 found._tokensize = count + found._dataEnd - pos;
2043 found.parenthesiscount = 0;
2044 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2045 found.disabled = true;
2049 else if (found.keytype != KeyInfo::isRegex) {
2050 found._tokenstart = sub.position(size_t(2));
2051 if (found.parenthesiscount == 0) {
2052 // Probably to be discarded
2053 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2054 char following = interval_.par[following_pos];
2055 if (following == ' ')
2056 found.head = "\\" + sub.str(5) + " ";
2057 else if (following == '=') {
2058 // like \uldepth=1000pt
2059 found.head = sub.str(2);
2062 found.head = "\\" + key;
2063 found._tokensize = found.head.length();
2064 found._dataEnd = found._tokenstart + found._tokensize;
2065 found._dataStart = found._dataEnd;
2068 int params = found._tokenstart + key.length() + 1;
2069 if (evaluatingOptional) {
2070 if (size_t(found._tokenstart) > optionalEnd) {
2071 evaluatingOptional = false;
2074 found.disabled = true;
2077 int optend = params;
2078 while (interval_.par[optend] == '[') {
2079 // discard optional parameters
2080 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2082 if (optend > params) {
2083 key += interval_.par.substr(params, optend-params);
2084 evaluatingOptional = true;
2085 optionalEnd = optend;
2087 string token = sub.str(7);
2089 if (interval_.par[optend] != '{') {
2091 found.parenthesiscount = 0;
2092 found.head = "\\" + key;
2095 closings = found.parenthesiscount;
2096 if (found.parenthesiscount == 1) {
2097 found.head = "\\" + key + "{";
2099 else if (found.parenthesiscount > 1) {
2101 found.head = sub.str(2) + "{";
2102 closings = found.parenthesiscount - 1;
2105 found.head = "\\" + key + "{";
2108 found._tokensize = found.head.length();
2109 found._dataStart = found._tokenstart + found.head.length();
2110 if (found.keytype == KeyInfo::doRemove) {
2112 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2113 if (endpar >= interval_.par.length())
2114 found._dataStart = interval_.par.length();
2116 found._dataStart = endpar;
2117 found._tokensize = found._dataStart - found._tokenstart;
2120 found._dataStart = found._tokenstart + found._tokensize;
2124 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
2125 found._dataStart += 15;
2129 endpos = found._dataStart - 1;
2131 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2132 if (found.keytype == KeyInfo::isList) {
2133 // Check if it really is list env
2134 static regex const listre("^([a-z]+)$");
2136 if (!regex_match(token, sub2, listre)) {
2137 // Change the key of this entry. It is not in a list/item environment
2138 found.keytype = KeyInfo::endArguments;
2141 if (found.keytype == KeyInfo::noMain) {
2142 evaluatingCode = true;
2144 codeStart = found._dataStart;
2146 else if (evaluatingCode) {
2147 if (size_t(found._dataStart) > codeEnd)
2148 evaluatingCode = false;
2149 else if (found.keytype == KeyInfo::isMain) {
2150 // Disable this key, treate it as standard
2151 found.keytype = KeyInfo::isStandard;
2152 found.disabled = true;
2153 if ((codeEnd +1 >= interval_.par.length()) &&
2154 (found._tokenstart == codeStart)) {
2155 // trickery, because the code inset starts
2156 // with \selectlanguage ...
2158 if (entries_.size() > 1) {
2159 entries_[entries_.size()-1]._dataEnd = codeEnd;
2164 if ((endpos == interval_.par.length()) &&
2165 (found.keytype == KeyInfo::doRemove)) {
2166 // Missing closing => error in latex-input?
2167 // therefore do not delete remaining data
2168 found._dataStart -= 1;
2169 found._dataEnd = found._dataStart;
2172 found._dataEnd = endpos;
2174 if (isPatternString) {
2175 keys[key].used = true;
2178 entries_.push_back(found);
2182 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2184 stringstream s(keysstring);
2186 const char delim = '|';
2187 while (getline(s, key, delim)) {
2188 KeyInfo keyII(keyI);
2189 if (isPatternString) {
2192 else if ( !keys[key].used)
2193 keyII.disabled = true;
2198 void LatexInfo::buildKeys(bool isPatternString)
2201 static bool keysBuilt = false;
2202 if (keysBuilt && !isPatternString) return;
2204 // Keys to ignore in any case
2205 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2206 // Known standard keys with 1 parameter.
2207 // Split is done, if not at start of region
2208 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2209 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2210 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2211 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2212 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2213 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2215 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2216 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2217 makeKey("section*|subsection*|subsubsection*|paragraph*",
2218 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2219 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2220 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2222 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2224 // Split is done, if not at start of region
2225 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2226 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2228 // Split is done always.
2229 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2233 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2234 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2235 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2236 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2238 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2239 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2240 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2241 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2243 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2244 // Custom space/skip, remove the content (== length value)
2245 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2246 // Found in fr/UserGuide.lyx
2247 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2249 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2250 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2251 // Known macros to remove (including their parameter)
2253 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2254 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2255 // handle like standard keys with 1 parameter.
2256 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2258 // Ignore deleted text
2259 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2260 // but preserve added text
2261 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2263 // Macros to remove, but let the parameter survive
2265 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2267 // Remove language spec from content of these insets
2268 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2270 // Same effect as previous, parameter will survive (because there is no one anyway)
2272 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2273 // Remove table decorations
2274 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2275 // Discard shape-header.
2276 // For footnote or shortcut too, because of lang settings
2277 // and wrong handling if used 'KeyInfo::noMain'
2278 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2279 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2280 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2281 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2282 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2283 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2284 // like ('tiny{}' or '\tiny ' ... )
2285 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2287 // Survives, like known character
2288 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2289 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2290 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2292 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2293 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2294 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2296 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2297 // Remove RTL/LTR marker
2298 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2299 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2300 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2301 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2302 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2303 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2304 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2305 if (isPatternString) {
2306 // Allow the first searched string to rebuild the keys too
2310 // no need to rebuild again
2316 * Keep the list of actual opened parentheses actual
2317 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2319 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2322 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2326 if (c == '\\') skip = 1;
2327 else if (c == '{') {
2330 else if (c == '}') {
2331 handleCloseP(i, closingAllowed);
2337 string Intervall::show(int lastpos)
2339 int idx = 0; /* int intervalls */
2342 for (idx = 0; idx <= ignoreidx; idx++) {
2343 while (i < lastpos) {
2345 if (i <= borders[idx].low) {
2346 if (borders[idx].low > lastpos)
2347 printsize = lastpos - i;
2349 printsize = borders[idx].low - i;
2350 s += par.substr(i, printsize);
2352 if (i >= borders[idx].low)
2353 i = borders[idx].upper;
2356 i = borders[idx].upper;
2362 s += par.substr(i, lastpos-i);
2368 void Intervall::output(ostringstream &os, int lastpos)
2370 // get number of chars to output
2371 int idx = 0; /* int intervalls */
2374 string startTitle = titleValue;
2375 for (idx = 0; idx <= ignoreidx; idx++) {
2377 if (i <= borders[idx].low) {
2379 if (borders[idx].low > lastpos)
2380 printsize = lastpos - i;
2382 printsize = borders[idx].low - i;
2383 if (printsize > 0) {
2384 os << startTitle << par.substr(i, printsize);
2386 printed += printsize;
2389 handleParentheses(i, false);
2390 if (i >= borders[idx].low)
2391 i = borders[idx].upper;
2394 i = borders[idx].upper;
2401 os << startTitle << par.substr(i, lastpos-i);
2402 printed += lastpos-i;
2404 handleParentheses(lastpos, false);
2406 if (keys["foreignlanguage"].disabled)
2407 startindex = actualdeptindex-langcount;
2409 startindex = actualdeptindex;
2410 for (int i = startindex; i > 0; --i) {
2413 if (hasTitle && (printed > 0))
2415 if (! isPatternString_)
2417 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2420 void LatexInfo::processRegion(int start, int region_end)
2422 while (start < region_end) { /* Let {[} and {]} survive */
2423 int cnt = interval_.isOpeningPar(start);
2425 // Closing is allowed past the region
2426 int closing = interval_.findclosing(start+1, interval_.par.length());
2427 interval_.addIntervall(start, start+1);
2428 interval_.addIntervall(closing, closing+1);
2432 start = interval_.nextNotIgnored(start+1);
2436 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2438 if (actual.parenthesiscount == 0) {
2439 // "{\tiny{} ...}" ==> "{{} ...}"
2440 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2443 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2444 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2445 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2449 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2452 switch (actual.keytype)
2454 case KeyInfo::isTitle: {
2456 nextKeyIdx = getNextKey();
2459 case KeyInfo::cleanToStart: {
2460 actual._dataEnd = actual._dataStart;
2461 nextKeyIdx = getNextKey();
2462 // Search for end of arguments
2463 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2465 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2466 entries_[i].disabled = true;
2468 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2470 while (interval_.par[actual._dataEnd] == ' ')
2472 interval_.addIntervall(0, actual._dataEnd+1);
2473 interval_.actualdeptindex = 0;
2474 interval_.depts[0] = actual._dataEnd+1;
2475 interval_.closes[0] = -1;
2478 case KeyInfo::isText:
2479 interval_.par[actual._tokenstart] = '#';
2480 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2481 nextKeyIdx = getNextKey();
2483 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2484 if (actual.disabled)
2485 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2487 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2490 case KeyInfo::isChar: {
2491 nextKeyIdx = getNextKey();
2494 case KeyInfo::isSize: {
2495 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2496 if (actual.parenthesiscount == 0)
2497 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2499 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2501 nextKeyIdx = getNextKey();
2503 // Here _dataStart points to '{', so correct it
2504 actual._dataStart += 1;
2505 actual._tokensize += 1;
2506 actual.parenthesiscount = 1;
2507 if (interval_.par[actual._dataStart] == '}') {
2508 // Determine the end if used like '{\tiny{}...}'
2509 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2510 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2513 // Determine the end if used like '\tiny{...}'
2514 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2516 // Split on this key if not at start
2517 int start = interval_.nextNotIgnored(previousStart);
2518 if (start < actual._tokenstart) {
2519 interval_.output(os, actual._tokenstart);
2520 interval_.addIntervall(start, actual._tokenstart);
2522 // discard entry if at end of actual
2523 nextKeyIdx = process(os, actual);
2527 case KeyInfo::endArguments: {
2528 // Remove trailing '{}' too
2529 actual._dataStart += 1;
2530 actual._dataEnd += 1;
2531 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2532 nextKeyIdx = getNextKey();
2535 case KeyInfo::noMain:
2537 case KeyInfo::isStandard: {
2538 if (actual.disabled) {
2540 processRegion(actual._dataStart, actual._dataStart+1);
2541 nextKeyIdx = getNextKey();
2543 // Split on this key if not at datastart of calling entry
2544 int start = interval_.nextNotIgnored(previousStart);
2545 if (start < actual._tokenstart) {
2546 interval_.output(os, actual._tokenstart);
2547 interval_.addIntervall(start, actual._tokenstart);
2549 // discard entry if at end of actual
2550 nextKeyIdx = process(os, actual);
2554 case KeyInfo::removeWithArg: {
2555 nextKeyIdx = getNextKey();
2556 // Search for end of arguments
2557 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2559 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2560 entries_[i].disabled = true;
2562 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2564 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2567 case KeyInfo::doRemove: {
2568 // Remove the key with all parameters and following spaces
2570 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2571 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2574 // Remove also enclosing parentheses [] and {}
2577 while (actual._tokenstart > numpars) {
2578 if (pos+numpars >= interval_.par.size())
2580 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2582 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2588 if (interval_.par[pos+numpars] == ' ')
2592 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2593 nextKeyIdx = getNextKey();
2596 case KeyInfo::isList: {
2597 // Discard space before _tokenstart
2599 for (count = 0; count < actual._tokenstart; count++) {
2600 if (interval_.par[actual._tokenstart-count-1] != ' ')
2603 nextKeyIdx = getNextKey();
2604 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2606 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2608 // How else can we catch this one?
2609 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2610 entries_[i].disabled = true;
2612 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2614 else if (nextKeyIdx > 0) {
2615 // Ignore any lang entries inside data region
2616 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2617 if (entries_[i].keytype == KeyInfo::isMain)
2618 entries_[i].disabled = true;
2621 if (actual.disabled) {
2622 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2625 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2627 if (interval_.par[actual._dataEnd+1] == '[') {
2628 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2629 if ((interval_.par[actual._dataEnd+2] == '{') &&
2630 (interval_.par[posdown-1] == '}')) {
2631 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2632 interval_.addIntervall(posdown-1, posdown+1);
2635 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2636 interval_.addIntervall(posdown, posdown+1);
2638 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2639 if (blk > posdown) {
2640 // Discard at most 1 space after empty item
2642 for (count = 0; count < 1; count++) {
2643 if (interval_.par[blk+count] != ' ')
2647 interval_.addIntervall(blk, blk+count);
2652 case KeyInfo::isSectioning: {
2653 // Discard spaces before _tokenstart
2655 int val = actual._tokenstart;
2656 for (count = 0; count < actual._tokenstart;) {
2657 val = interval_.previousNotIgnored(val-1);
2658 if (val < 0 || interval_.par[val] != ' ')
2661 count = actual._tokenstart - val;
2664 if (actual.disabled) {
2665 removeHead(actual, count);
2666 nextKeyIdx = getNextKey();
2668 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2669 nextKeyIdx = process(os, actual);
2673 case KeyInfo::isMath: {
2674 // Same as regex, use the content unchanged
2675 nextKeyIdx = getNextKey();
2678 case KeyInfo::isRegex: {
2679 // DO NOT SPLIT ON REGEX
2681 nextKeyIdx = getNextKey();
2684 case KeyInfo::isIgnored: {
2685 // Treat like a character for now
2686 nextKeyIdx = getNextKey();
2689 case KeyInfo::isMain: {
2690 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2691 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2692 if (actual._tokenstart > 0) {
2693 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2694 if ((prev >= 0) && interval_.par[prev] == '%')
2695 interval_.addIntervall(prev, prev+1);
2697 if (actual.disabled) {
2699 interval_.langcount++;
2700 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2701 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2702 // Discard also the space before math-equation
2703 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2705 nextKeyIdx = getNextKey();
2706 // interval.resetOpenedP(actual._dataStart-1);
2709 if (actual._tokenstart < 26) {
2710 // for the first (and maybe dummy) language
2711 interval_.setForDefaultLang(actual);
2713 interval_.resetOpenedP(actual._dataStart-1);
2717 case KeyInfo::invalid:
2718 case KeyInfo::headRemove:
2719 // These two cases cannot happen, already handled
2722 // LYXERR(Debug::INFO, "Unhandled keytype");
2723 nextKeyIdx = getNextKey();
2730 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2732 int end = interval_.nextNotIgnored(actual._dataEnd);
2733 int oldStart = actual._dataStart;
2734 int nextKeyIdx = getNextKey();
2736 if ((nextKeyIdx < 0) ||
2737 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2738 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2739 if (oldStart <= end) {
2740 processRegion(oldStart, end);
2745 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2747 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2748 (void) dispatch(os, actual._dataStart, nextKey);
2749 end = nextKey._tokenstart;
2752 processRegion(oldStart, nextKey._tokenstart);
2753 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2755 oldStart = nextKey._dataEnd+1;
2757 // now nextKey is either invalid or is outside of actual._dataEnd
2758 // output the remaining and discard myself
2759 if (oldStart <= end) {
2760 processRegion(oldStart, end);
2762 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2764 // This is the normal case.
2765 // But if using the firstlanguage, the closing may be missing
2767 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2769 if (actual._dataEnd < end)
2770 output_end = interval_.nextNotIgnored(actual._dataEnd);
2771 else if (interval_.par.size() > (size_t) end)
2772 output_end = interval_.nextNotIgnored(end);
2774 output_end = interval_.par.size();
2775 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2776 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2778 // Remove possible empty data
2779 int dstart = interval_.nextNotIgnored(actual._dataStart);
2780 while (interval_.isOpeningPar(dstart) == 1) {
2781 interval_.addIntervall(dstart, dstart+1);
2782 int dend = interval_.findclosing(dstart+1, output_end);
2783 interval_.addIntervall(dend, dend+1);
2784 dstart = interval_.nextNotIgnored(dstart+1);
2786 if (dstart < output_end)
2787 interval_.output(os, output_end);
2788 interval_.addIntervall(actual._tokenstart, end);
2792 string splitOnKnownMacros(string par, bool isPatternString)
2795 LatexInfo li(par, isPatternString);
2796 // LYXERR(Debug::INFO, "Berfore split: " << par);
2797 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2799 DummyKey._tokensize = 0;
2800 DummyKey._dataStart = 0;
2801 DummyKey._dataEnd = par.length();
2802 DummyKey.disabled = true;
2803 int firstkeyIdx = li.getFirstKey();
2805 if (firstkeyIdx >= 0) {
2806 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2807 DummyKey._tokenstart = firstKey._tokenstart;
2809 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2810 // Use dummy firstKey
2811 firstKey = DummyKey;
2812 (void) li.setNextKey(firstkeyIdx);
2815 if (par.substr(firstKey._dataStart, 2) == "% ")
2816 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2818 nextkeyIdx = li.process(os, firstKey);
2819 while (nextkeyIdx >= 0) {
2820 // Check for a possible gap between the last
2821 // entry and this one
2822 int datastart = li.nextNotIgnored(firstKey._dataStart);
2823 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2824 if ((nextKey._tokenstart > datastart)) {
2826 firstKey._dataStart = datastart;
2827 firstKey._dataEnd = par.length();
2828 (void) li.setNextKey(nextkeyIdx);
2829 // Fake the last opened parenthesis
2830 li.setForDefaultLang(firstKey);
2831 nextkeyIdx = li.process(os, firstKey);
2834 if (nextKey.keytype != KeyInfo::isMain) {
2835 firstKey._dataStart = datastart;
2836 firstKey._dataEnd = nextKey._dataEnd+1;
2837 (void) li.setNextKey(nextkeyIdx);
2838 li.setForDefaultLang(firstKey);
2839 nextkeyIdx = li.process(os, firstKey);
2842 nextkeyIdx = li.process(os, nextKey);
2846 // Handle the remaining
2847 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2848 firstKey._dataEnd = par.length();
2850 if ((firstKey._dataStart < firstKey._dataEnd) &&
2851 (par[firstKey._dataStart] != '}')) {
2852 li.setForDefaultLang(firstKey);
2853 (void) li.process(os, firstKey);
2857 // return string definitelly impossible to match
2858 s = "\\foreignlanguage{ignore}{ }";
2862 s = par; /* no known macros found */
2863 // LYXERR(Debug::INFO, "After split: " << s);
2868 * Try to unify the language specs in the latexified text.
2869 * Resulting modified string is set to "", if
2870 * the searched tex does not contain all the features in the search pattern
2872 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2874 static Features regex_f;
2875 static int missed = 0;
2876 static bool regex_with_format = false;
2878 int parlen = par.length();
2880 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2883 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2884 // Happens to be there in case of description or labeling environment
2889 // Split the latex input into pieces which
2890 // can be digested by our search engine
2891 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2892 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2893 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2896 result = par.substr(0, parlen);
2897 if (isPatternString) {
2900 regex_f = identifyFeatures(result);
2901 string features = "";
2902 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2903 string a = it->first;
2904 regex_with_format = true;
2905 features += " " + a;
2906 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2908 LYXERR(Debug::FIND, "Identified Features" << features);
2911 } else if (regex_with_format) {
2912 Features info = identifyFeatures(result);
2913 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2914 string a = it->first;
2915 bool b = it->second;
2916 if (b && ! info[a]) {
2918 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2924 // LYXERR(Debug::INFO, "No regex formats");
2930 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2931 static int identifyClosing(string & t)
2933 int open_braces = 0;
2935 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2936 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2938 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2940 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2942 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2951 static int num_replaced = 0;
2952 static bool previous_single_replace = true;
2954 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
2957 // Handle \w properly
2958 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
2959 if (! opt.casesensitive) {
2960 popts |= QRegularExpression::CaseInsensitiveOption;
2962 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
2963 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
2965 if (regexp.isValid() && regexp2.isValid()) {
2966 regexIsValid = true;
2967 // Check '{', '}' pairs inside the regex
2970 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
2971 char c = par_as_string[i];
2978 else if (c == '}') {
2985 if (balanced != 0) {
2986 regexIsValid = false;
2987 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
2991 regexIsValid = false;
2992 if (!regexp.isValid())
2993 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
2995 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
2998 if (opt.casesensitive) {
2999 regexp = regex(regexp_str);
3000 regexp2 = regex(regexp2_str);
3003 regexp = regex(regexp_str, std::regex_constants::icase);
3004 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3009 static void modifyRegexForMatchWord(string &t)
3012 regex wordre("(\\\\)*((\\.|\\\\b))");
3015 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3017 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3020 else if (sub.str(2) == "\\\\b")
3022 if (lastpos < (size_t) sub.position(2))
3023 s += t.substr(lastpos, sub.position(2) - lastpos);
3025 lastpos = sub.position(2) + sub.length(2);
3029 else if (lastpos < t.length())
3030 s += t.substr(lastpos, t.length() - lastpos);
3031 t = "\\b" + s + "\\b";
3034 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3035 : p_buf(&buf), p_first_buf(&buf), opt(opt)
3037 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3038 docstring const & ds = stringifySearchBuffer(find_buf, opt);
3039 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3040 if (opt.replace_all && previous_single_replace) {
3041 previous_single_replace = false;
3044 else if (!opt.replace_all) {
3045 num_replaced = 0; // count number of replaced strings
3046 previous_single_replace = true;
3048 // When using regexp, braces are hacked already by escape_for_regex()
3049 par_as_string = normalize(ds);
3051 close_wildcards = 0;
3053 size_t lead_size = 0;
3054 // correct the language settings
3055 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
3056 opt.matchstart = false;
3058 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
3059 if (opt.ignoreformat) {
3063 lead_size = identifyLeading(par_as_string);
3065 lead_as_string = par_as_string.substr(0, lead_size);
3066 string lead_as_regex_string = string2regex(lead_as_string);
3067 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3068 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3069 /* Handle whole words too in this case
3071 if (opt.matchword) {
3072 par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3073 opt.matchword = false;
3075 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3076 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3077 CreateRegexp(opt, regexp_str, regexp2_str);
3079 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3080 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3084 if (!opt.ignoreformat) {
3085 lead_size = identifyLeading(par_as_string);
3086 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
3087 lead_as_string = par_as_string.substr(0, lead_size);
3088 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3091 // Here we are using regexp
3092 LASSERT(use_regexp, /**/);
3094 string lead_as_regexp;
3095 if (lead_size > 0) {
3096 lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3097 regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3098 par_as_string = par_as_string_nolead;
3099 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
3100 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3102 LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3103 par_as_string = escape_for_regex(par_as_string);
3104 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3105 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3106 LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
3110 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
3111 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
3112 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
3113 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
3114 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
3115 || regex_replace(par_as_string, par_as_string,
3116 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
3117 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
3118 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
3123 if (!opt.ignoreformat) {
3124 // Remove extra '\}' at end if not part of \{\.\}
3125 size_t lng = par_as_string.size();
3127 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3129 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3138 if (lng < par_as_string.size())
3139 par_as_string = par_as_string.substr(0,lng);
3140 if ((lng > 0) && (par_as_string[0] == '^')) {
3141 par_as_string = par_as_string.substr(1);
3143 opt.matchstart = true;
3146 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3147 LYXERR(Debug::FIND, "Open braces: " << open_braces);
3148 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3150 // If entered regexp must match at begin of searched string buffer
3151 // Kornel: Added parentheses to use $1 for size of the leading string
3155 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3156 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3157 // so the convert has no effect in that case
3158 for (int i = 7; i > 0; --i) {
3159 string orig = "\\\\" + std::to_string(i);
3160 string dest = "\\" + std::to_string(i+2);
3161 while (regex_replace(par_as_string, par_as_string, orig, dest));
3163 /* opt.matchword is ignored if using regex
3164 so expanding par_as_string with "\\b" seems appropriate here
3165 if regex contains for instance '.*' or '.+'
3166 1.) Nothing to do, if 'par_as_string' contains "\\b" already.
3167 (Means, that the user knows how to handle whole words
3168 2.) else replace '.' with "\\S" and wrap the regex with "\\b"
3170 if (opt.matchword) {
3171 modifyRegexForMatchWord(par_as_string);
3172 opt.matchword = false;
3174 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3175 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3177 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3178 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3179 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3184 // Count number of characters in string
3188 // \\[a-zA-Z]+ ==> 1
3190 static int computeSize(QStringRef s, int len)
3191 #define isLyxAlpha(arg) arg.isLetter()
3193 static int computeSize(string s, int len)
3194 #define isLyxAlpha(arg) isalpha(arg)
3201 for (int i = 0; i < len; i += skip, count++) {
3202 if (s.at(i) == '\\') {
3204 if (i + 1 < len && isLyxAlpha(s.at(i+1))) {
3205 for (int j = 2; i+j < len; j++) {
3206 if (! isLyxAlpha(s.at(i+j))) {
3207 if (s.at(i+j) == ' ')
3209 else if (s.at(i+j) == '{') {
3210 if (i+j+1 < len && s.at(i+j+1) == '}')
3212 else if (i + j + 1 >= len)
3221 else if (s.at(i) == '{') {
3222 if (i + 1 < len && s.at(i+1) == '}')
3227 else if (s.at(i) == '-') {
3228 if (i+1 < len && s.at(i+1) == '-') {
3229 if (i + 2 < len && s.at(i+2) == '-')
3245 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3250 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3253 docstring docstr = stringifyFromForSearch(opt, cur, len);
3255 if (use_regexp || opt.casesensitive)
3256 str = normalize(docstr);
3258 str = normalize(lowercase(docstr));
3259 if (!opt.ignoreformat) {
3260 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3263 mres.match_len = -1;
3266 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
3267 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
3270 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3272 QString qstr = QString::fromStdString(str);
3273 QRegularExpression const *p_regexp;
3274 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3278 p_regexp = ®exp2;
3280 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3281 if (!match.hasMatch())
3284 regex const *p_regexp;
3285 regex_constants::match_flag_type flags;
3287 flags = regex_constants::match_continuous;
3290 flags = regex_constants::match_default;
3291 p_regexp = ®exp2;
3293 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3294 if (re_it == sregex_iterator())
3296 match_results<string::const_iterator> const & m = *re_it;
3298 // Whole found string, including the leading
3299 // std: m[0].second - m[0].first
3300 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3302 // Size of the leading string
3303 // std: m[1].second - m[1].first
3304 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3305 int leadingsize = 0;
3307 if (match.lastCapturedIndex() > 0) {
3308 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3313 leadingsize = m[1].second - m[1].first;
3317 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3318 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3319 // because of different number of closing at end of string
3320 // we have to 'unify' the length of the post-match.
3321 // Done by ignoring closing parenthesis and linefeeds at string end
3322 int matchend = match.capturedEnd(0);
3323 while (mres.match_len > 0) {
3324 QChar c = qstr.at(matchend - 1);
3325 if ((c == '\n') || (c == '}') || (c == '{')) {
3332 size_t strsize = qstr.size();
3333 while (strsize > (size_t) match.capturedEnd(0)) {
3334 QChar c = qstr.at(strsize-1);
3335 if ((c == '\n') || (c == '}')) {
3341 // LYXERR0(qstr.toStdString());
3342 mres.match2end = strsize - matchend;
3343 mres.pos = match.capturedStart(2);
3345 mres.match_prefix = m[2].second - m[2].first;
3346 mres.match_len = m[0].second - m[2].second;
3347 // ignore closing parenthesis and linefeeds at string end
3348 size_t strend = m[0].second - m[0].first;
3349 int matchend = strend;
3350 while (mres.match_len > 0) {
3351 char c = str.at(matchend - 1);
3352 if ((c == '\n') || (c == '}') || (c == '{')) {
3359 size_t strsize = str.size();
3360 while (strsize > strend) {
3361 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3368 mres.match2end = strsize - matchend;
3369 mres.pos = m[2].first - m[0].first;;
3371 if (mres.match2end < 0)
3373 mres.leadsize = leadingsize;
3375 if (mres.match_len > 0) {
3376 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3377 mres.result.push_back(a0);
3378 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3379 mres.result.push_back(match.captured(i).toStdString());
3383 if (mres.match_len > 0) {
3384 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3385 mres.result.push_back(a0);
3386 for (size_t i = 3; i < m.size(); i++) {
3387 mres.result.push_back(m[i]);
3394 // else !use_regexp: but all code paths above return
3395 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
3396 << par_as_string << "', str='" << str << "'");
3397 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
3398 << lead_as_string << "', par_as_string_nolead='"
3399 << par_as_string_nolead << "'");
3402 LYXERR(Debug::FIND, "size=" << par_as_string.size()
3403 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
3404 if (str.substr(0, par_as_string.size()) == par_as_string) {
3405 mres.match_len = par_as_string.size();
3406 mres.match2end = str.size();
3411 // Start the search _after_ the leading part
3412 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3413 if (pos != string::npos) {
3414 mres.match_len = par_as_string.size();
3415 mres.match2end = str.size() - pos;
3424 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3426 MatchResult mres = findAux(cur, len, at_begin);
3427 int res = mres.match_len;
3429 "res=" << res << ", at_begin=" << at_begin
3430 << ", matchstart=" << opt.matchstart
3431 << ", inTexted=" << cur.inTexted());
3432 if (opt.matchstart) {
3435 else if (mres.match_prefix > 0)
3441 /* DEAD CODE follows
3442 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3444 if ((len > 0) && (res < len)) {
3448 Paragraph const & par = cur.paragraph();
3449 bool ws_left = (cur.pos() > 0)
3450 ? par.isWordSeparator(cur.pos() - 1)
3456 ws_right = (cur.pos() + len < par.size())
3457 ? par.isWordSeparator(cur.pos() + len)
3461 "cur.pos()=" << cur.pos() << ", res=" << res
3462 << ", separ: " << ws_left << ", " << ws_right
3465 if (ws_left && ws_right) {
3466 // Check for word separators inside the found 'word'
3467 for (int i = 0; i < len; i++) {
3468 if (par.isWordSeparator(cur.pos() + i)) {
3481 static bool simple_replace(string &t, string from, string to)
3483 regex repl("(\\\\)*(" + from + ")");
3487 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3489 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3491 if (lastpos < (size_t) sub.position(2))
3492 s += t.substr(lastpos, sub.position(2) - lastpos);
3494 lastpos = sub.position(2) + sub.length(2);
3498 else if (lastpos < t.length())
3499 s += t.substr(lastpos, t.length() - lastpos);
3505 string MatchStringAdv::normalize(docstring const & s) const
3508 t = lyx::to_utf8(s);
3509 // Remove \n at begin
3510 while (!t.empty() && t[0] == '\n')
3513 while (!t.empty() && t[t.size() - 1] == '\n')
3514 t = t.substr(0, t.size() - 1);
3516 // Handle all other '\n'
3517 while ((pos = t.find("\n")) != string::npos) {
3518 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3520 if (isAlnumASCII(t[pos+1])) {
3521 t.replace(pos-2, 3, " ");
3524 t.replace(pos-2, 3, "");
3527 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3528 // '\n' adjacent to non-alpha-numerics, discard
3529 t.replace(pos, 1, "");
3532 // Replace all other \n with spaces
3533 t.replace(pos, 1, " ");
3536 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3537 // Kornel: Added textsl, textsf, textit, texttt and noun
3538 // + allow to seach for colored text too
3539 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3540 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3541 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3542 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3543 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3544 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3550 docstring stringifyFromCursor(DocIterator const & cur, int len)
3552 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3553 if (cur.inTexted()) {
3554 Paragraph const & par = cur.paragraph();
3555 // TODO what about searching beyond/across paragraph breaks ?
3556 // TODO Try adding a AS_STR_INSERTS as last arg
3557 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3558 int(par.size()) : cur.pos() + len;
3559 // OutputParams runparams(&cur.buffer()->params().encoding());
3560 OutputParams runparams(encodings.fromLyXName("utf8"));
3561 runparams.nice = true;
3562 runparams.flavor = Flavor::XeTeX;
3563 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3564 // No side effect of file copying and image conversion
3565 runparams.dryrun = true;
3566 runparams.for_search = true;
3567 LYXERR(Debug::FIND, "Stringifying with cur: "
3568 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3569 return par.asString(cur.pos(), end,
3570 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3572 } else if (cur.inMathed()) {
3573 CursorSlice cs = cur.top();
3574 MathData md = cs.cell();
3575 MathData::const_iterator it_end =
3576 (( len == -1 || cs.pos() + len > int(md.size()))
3578 : md.begin() + cs.pos() + len );
3580 for (MathData::const_iterator it = md.begin() + cs.pos();
3583 docstring s = asString(md2);
3584 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3587 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3592 /** Computes the LaTeX export of buf starting from cur and ending len positions
3593 * after cur, if len is positive, or at the paragraph or innermost inset end
3596 docstring latexifyFromCursor(DocIterator const & cur, int len)
3598 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3599 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3600 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3601 Buffer const & buf = *cur.buffer();
3603 odocstringstream ods;
3605 //OutputParams runparams(&buf.params().encoding());
3606 OutputParams runparams(encodings.fromLyXName("utf8"));
3607 runparams.nice = false;
3608 runparams.flavor = Flavor::XeTeX;
3609 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3610 // No side effect of file copying and image conversion
3611 runparams.dryrun = true;
3612 runparams.for_search = true;
3614 if (cur.inTexted()) {
3615 // @TODO what about searching beyond/across paragraph breaks ?
3616 pos_type endpos = cur.paragraph().size();
3617 if (len != -1 && endpos > cur.pos() + len)
3618 endpos = cur.pos() + len;
3619 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3620 string(), cur.pos(), endpos);
3621 string s = lyx::to_utf8(ods.str());
3622 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3623 return(lyx::from_utf8(s));
3624 } else if (cur.inMathed()) {
3625 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3626 for (int s = cur.depth() - 1; s >= 0; --s) {
3627 CursorSlice const & cs = cur[s];
3628 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3629 TeXMathStream ws(os);
3630 cs.asInsetMath()->asHullInset()->header_write(ws);
3635 CursorSlice const & cs = cur.top();
3636 MathData md = cs.cell();
3637 MathData::const_iterator it_end =
3638 ((len == -1 || cs.pos() + len > int(md.size()))
3640 : md.begin() + cs.pos() + len);
3642 for (MathData::const_iterator it = md.begin() + cs.pos();
3646 ods << asString(md2);
3647 // Retrieve the math environment type, and add '$' or '$]'
3648 // or others (\end{equation}) accordingly
3649 for (int s = cur.depth() - 1; s >= 0; --s) {
3650 CursorSlice const & cs2 = cur[s];
3651 InsetMath * inset = cs2.asInsetMath();
3652 if (inset && inset->asHullInset()) {
3653 TeXMathStream ws(os);
3654 inset->asHullInset()->footer_write(ws);
3658 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3660 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3665 #if defined(ResultsDebug)
3667 static void displayMResult(MatchResult &mres, int increment)
3669 LYXERR0( "pos: " << mres.pos << " increment " << increment);
3670 LYXERR0( "leadsize: " << mres.leadsize);
3671 LYXERR0( "match_len: " << mres.match_len);
3672 LYXERR0( "match_prefix: " << mres.match_prefix);
3673 LYXERR0( "match2end: " << mres.match2end);
3674 LYXERR0( "pos_len: " << mres.pos_len); // Set in finalize
3675 for (size_t i = 0; i < mres.result.size(); i++)
3676 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3678 #define displayMres(s,i) displayMResult(s,i);
3680 #define displayMres(s,i)
3684 * Not good, we miss possible matches containing also characters not found in
3685 * the innermost depth.
3686 static bool findAdvForwardInnermost(DocIterator & cur)
3689 DocIterator old_cur = cur;
3690 int forwardCount = 0;
3698 if (cur.depth() > d) {
3702 if (cur.depth() == d)
3706 if (forwardCount > 0) {
3707 LYXERR(Debug::FIND, "Forwarded " << forwardCount << " step(s) (searching for innermost match)");
3715 /** Finalize an advanced find operation, advancing the cursor to the innermost
3716 ** position that matches, plus computing the length of the matching text to
3718 ** Return the cur.pos() difference between start and end of found match
3720 MatchResult &findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
3722 // Search the foremost position that matches (avoids find of entire math
3723 // inset when match at start of it)
3724 DocIterator old_cur(cur.buffer());
3726 static MatchResult fail = MatchResult();
3727 static MatchResult max_match;
3728 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3729 // Happens with e.g. hyperlinks
3730 // either one sees "http://www.bla.bla" or nothing
3731 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3732 // and although we search for only 3 chars, we find the whole hyperlink inset
3733 bool at_begin = (expected.match_prefix == 0);
3734 //if (findAdvForwardInnermost(cur)) {
3736 mres = match(cur, -1, at_begin);
3737 displayMres(mres, 0);
3738 if (expected.match_len > 0) {
3739 if (mres.match_len < expected.match_len)
3743 if (mres.match_len <= 0)
3748 else if (expected.match_len < 0) {
3749 mres = match(cur); /* match valid only if not searching whole words */
3750 displayMres(mres, 0);
3754 max_match = expected;
3756 if (max_match.match_len <= 0) return fail;
3757 LYXERR(Debug::FIND, "Ok");
3759 // Compute the match length
3761 if (cur.pos() + len > cur.lastpos())
3763 // regexp should use \w+, \S+, or \b(some string)\b
3764 // to search for whole words
3765 if (match.opt.matchword && !match.use_regexp) {
3766 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3767 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3769 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3771 // Length of matched text (different from len param)
3772 static MatchResult old_match = match(cur, len, at_begin);
3773 if (old_match.match_len < 0)
3775 MatchResult new_match;
3776 // Greedy behaviour while matching regexps
3777 while ((new_match = match(cur, len + 1, at_begin)).match_len > old_match.match_len) {
3779 old_match = new_match;
3780 LYXERR(Debug::FIND, "verifying match with len = " << len);
3786 int maxl = cur.lastpos() - cur.pos();
3787 // Greedy behaviour while matching regexps
3788 while (maxl > minl) {
3790 mres2 = match(cur, len, at_begin);
3791 displayMres(mres2, len);
3792 int actual_match = mres2.match_len;
3793 if (actual_match >= max_match.match_len) {
3794 // actual_match > max_match _can_ happen,
3795 // if the search area splits
3796 // some following word so that the regex
3797 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3799 // This means, the len value is too big
3801 if (maxl - minl < 4)
3802 len = (int)((maxl + minl)/2);
3804 len = (int)(minl + (maxl - minl + 3)/4);
3807 // (actual_match < max_match)
3809 len = (int)((maxl + minl)/2);
3813 // Search for real start of matched characters
3815 MatchResult actual_match;
3818 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3819 if (cur.depth() < old_cur.depth()) {
3821 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3824 if (cur.pos() != old_cur.pos()) {
3825 // OK, forwarded 1 pos in actual inset
3826 actual_match = match(cur, len-1, at_begin);
3827 if (actual_match.match_len == max_match.match_len) {
3828 // Ha, got it! The shorter selection has the same match length
3833 // OK, the shorter selection matches less chars, revert to previous value
3839 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3840 actual_match = match(cur, len, at_begin);
3841 if (actual_match.match_len == max_match.match_len) {
3843 max_match = actual_match;
3850 max_match.pos_len = len;
3857 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3861 while (!theApp()->longOperationCancelled() && cur) {
3862 //(void) findAdvForwardInnermost(cur);
3863 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3864 MatchResult mres = match(cur, -1, false);
3865 displayMres(mres,-1)
3866 int match_len = mres.match_len;
3867 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3868 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3871 if (match_len > 0) {
3872 // Try to find the begin of searched string
3874 int firstInvalid = 100000;
3875 if (mres.match_prefix + mres.pos - mres.leadsize > 1)
3876 increment = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3879 LYXERR(Debug::FIND, "Set increment to " << increment);
3880 while (increment > 0) {
3881 DocIterator old_cur = cur;
3882 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3884 if (! cur || (cur.pit() > old_cur.pit())) {
3885 // Are we outside of the paragraph?
3886 // This can happen if moving past some UTF8-encoded chars
3891 MatchResult mres2 = match(cur, -1, false);
3892 displayMres(mres2,increment)
3893 switch (interpretMatch(mres, mres2)) {
3894 case MatchResult::newIsTooFar:
3895 // behind the expected match
3896 firstInvalid = increment;
3900 case MatchResult::newIsBetter:
3901 // not reached ye, but cur.pos()+increment is bettert
3903 firstInvalid -= increment;
3904 if (increment > firstInvalid*3/4)
3905 increment = firstInvalid*3/4;
3906 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3907 if (increment >= mres2.match_prefix)
3908 increment = (mres2.match_prefix+1)*3/4;
3913 // Handle not like MatchResult::newIsTooFar
3914 // LYXERR0( "Something is wrong: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3916 increment = increment*3/4;
3922 // LYXERR0("Leaving first loop");
3924 LYXERR(Debug::FIND, "Finalizing 1");
3925 MatchResult found_match = findAdvFinalize(cur, match, mres);
3926 if (found_match.match_len > 0) {
3927 LASSERT(found_match.pos_len > 0, /**/);
3928 match.FillResults(found_match);
3929 return found_match.pos_len;
3932 // try next possible match
3937 // The following code is newer reached
3938 // but parts of it may be needed in future
3939 int match_len_zero_count = 0;
3941 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3943 mres3 = match(cur, -1, false);
3944 displayMres(mres3, 1)
3945 int remaining_len = mres3.match_len;
3946 if (remaining_len <= 0) {
3947 // Apparently the searched string is not in the remaining part
3954 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3955 mres3 = match(cur, 1);
3956 displayMres(mres3, 1)
3957 int match_len3 = mres3.match_len;
3961 displayMres(mres3, 1)
3962 int match_len2 = mres3.match_len;
3963 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3964 if (match_len2 > 0) {
3965 // Sometimes in finalize we understand it wasn't a match
3966 // and we need to continue the outest loop
3967 LYXERR(Debug::FIND, "Finalizing 2");
3968 MatchResult mres4 = findAdvFinalize(cur, match, mres.match_len);
3969 if (mres4.match_len > 0) {
3970 match.FillResults(mres4);
3971 LASSERT(mres4.pos_len > 0, /**/);
3972 return mres4.pos_len;
3976 match_len_zero_count = 0;
3977 else if (match_len2 == 0)
3978 match_len_zero_count++;
3979 if (match_len2 < 0) {
3980 if (++match_len_zero_count > 3) {
3981 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3989 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3990 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3993 // This should exit nested insets, if any, or otherwise undefine the currsor.
3994 cur.pos() = cur.lastpos();
3995 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
4003 /// Find the most backward consecutive match within same paragraph while searching backwards.
4004 MatchResult &findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
4006 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4007 DocIterator tmp_cur = cur;
4008 static MatchResult mr = findAdvFinalize(tmp_cur, match, MatchResult(-1));
4009 Inset & inset = cur.inset();
4010 for (; cur != cur_begin; cur.backwardPos()) {
4011 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
4012 DocIterator new_cur = cur;
4013 new_cur.backwardPos();
4014 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
4016 MatchResult new_mr = findAdvFinalize(new_cur, match, MatchResult(-1));
4017 if (new_mr.match_len == mr.match_len)
4021 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
4027 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
4031 // Backup of original position
4032 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
4033 if (cur == cur_begin)
4036 DocIterator cur_orig(cur);
4037 bool pit_changed = false;
4040 bool found_match = (match(cur, -1, false).match_len > 0);
4044 cur.pos() = cur.lastpos();
4046 cur.pos() = cur_orig.pos();
4047 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
4048 DocIterator cur_prev_iter;
4050 found_match = (match(cur).match_len > 0);
4051 LYXERR(Debug::FIND, "findBackAdv3: found_match="
4052 << found_match << ", cur: " << cur);
4054 MatchResult found_mr = findMostBackwards(cur, match);
4055 match.FillResults(found_mr);
4056 LASSERT(found_mr.pos_len > 0, /**/);
4057 return found_mr.pos_len;
4060 // Stop if begin of document reached
4061 if (cur == cur_begin)
4063 cur_prev_iter = cur;
4067 if (cur == cur_begin)
4074 } while (!theApp()->longOperationCancelled());
4082 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4083 DocIterator const & cur, int len)
4085 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4087 if (!opt.ignoreformat)
4088 return latexifyFromCursor(cur, len);
4090 return stringifyFromCursor(cur, len);
4094 FindAndReplaceOptions::FindAndReplaceOptions(
4095 docstring const & _find_buf_name, bool _casesensitive,
4096 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4097 docstring const & _repl_buf_name, bool _keep_case,
4098 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4099 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4100 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4101 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4109 /** Check if 'len' letters following cursor are all non-lowercase */
4110 static bool allNonLowercase(Cursor const & cur, int len)
4112 pos_type beg_pos = cur.selectionBegin().pos();
4113 pos_type end_pos = cur.selectionBegin().pos() + len;
4114 if (len > cur.lastpos() + 1 - beg_pos) {
4115 LYXERR(Debug::FIND, "This should not happen, more debug needed");
4116 len = cur.lastpos() + 1 - beg_pos;
4117 end_pos = beg_pos + len;
4119 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4120 if (isLowerCase(cur.paragraph().getChar(pos)))
4126 /** Check if first letter is upper case and second one is lower case */
4127 static bool firstUppercase(Cursor const & cur)
4130 pos_type pos = cur.selectionBegin().pos();
4131 if (pos >= cur.lastpos() - 1) {
4132 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
4135 ch1 = cur.paragraph().getChar(pos);
4136 ch2 = cur.paragraph().getChar(pos + 1);
4137 bool result = isUpperCase(ch1) && isLowerCase(ch2);
4138 LYXERR(Debug::FIND, "firstUppercase(): "
4139 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4140 << ch2 << "(" << char(ch2) << ")"
4141 << ", result=" << result << ", cur=" << cur);
4146 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4148 ** \fixme What to do with possible further paragraphs in replace buffer ?
4150 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4152 ParagraphList::iterator pit = buffer.paragraphs().begin();
4153 LASSERT(!pit->empty(), /**/);
4154 pos_type right = pos_type(1);
4155 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4156 right = pit->size();
4157 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4162 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4164 // Should replace the string "$" + std::to_string(matchnum) with replacement
4165 // if the char '$' is not prefixed with odd number of char '\\'
4166 static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4170 for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4172 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4174 int num = stoi(sub.str(3), nullptr, 10);
4175 if (num >= maxmatchnum)
4177 if (lastpos < (size_t) sub.position(2))
4178 s += t.substr(lastpos, sub.position(2) - lastpos);
4179 s += replacements[num];
4180 lastpos = sub.position(2) + sub.length(2);
4184 else if (lastpos < t.length())
4185 s += t.substr(lastpos, t.length() - lastpos);
4192 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4194 Cursor & cur = bv->cursor();
4195 if (opt.repl_buf_name.empty()
4196 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4197 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4200 DocIterator sel_beg = cur.selectionBegin();
4201 DocIterator sel_end = cur.selectionEnd();
4202 if (&sel_beg.inset() != &sel_end.inset()
4203 || sel_beg.pit() != sel_end.pit()
4204 || sel_beg.idx() != sel_end.idx())
4206 int sel_len = sel_end.pos() - sel_beg.pos();
4207 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4208 << ", sel_len: " << sel_len << endl);
4211 LASSERT(sel_len > 0, return 0);
4213 if (!matchAdv(sel_beg, sel_len).match_len)
4216 // Build a copy of the replace buffer, adapted to the KeepCase option
4217 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4219 repl_buffer_orig.write(oss);
4220 string lyx = oss.str();
4221 if (matchAdv.valid_matches > 0) {
4222 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4224 Buffer repl_buffer("", false);
4225 repl_buffer.setUnnamed(true);
4226 LASSERT(repl_buffer.readString(lyx), return 0);
4227 if (opt.keep_case && sel_len >= 2) {
4228 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4229 if (cur.inTexted()) {
4230 if (firstUppercase(cur))
4231 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4232 else if (allNonLowercase(cur, sel_len))
4233 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4236 cap::cutSelection(cur, false);
4237 if (cur.inTexted()) {
4238 repl_buffer.changeLanguage(
4239 repl_buffer.language(),
4240 cur.getFont().language());
4241 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4242 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4243 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4244 repl_buffer.params().documentClassPtr(),
4245 bv->buffer().errorList("Paste"));
4246 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4247 sel_len = repl_buffer.paragraphs().begin()->size();
4248 } else if (cur.inMathed()) {
4249 odocstringstream ods;
4251 // OutputParams runparams(&repl_buffer.params().encoding());
4252 OutputParams runparams(encodings.fromLyXName("utf8"));
4253 runparams.nice = false;
4254 runparams.flavor = Flavor::XeTeX;
4255 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4256 runparams.dryrun = true;
4257 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4258 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4259 docstring repl_latex = ods.str();
4260 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4262 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4263 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4264 repl_latex = from_utf8(s);
4265 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4266 MathData ar(cur.buffer());
4267 asArray(repl_latex, ar, Parse::NORMAL);
4269 sel_len = ar.size();
4270 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4272 if (cur.pos() >= sel_len)
4273 cur.pos() -= sel_len;
4276 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4277 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4278 bv->processUpdateFlags(Update::Force);
4283 /// Perform a FindAdv operation.
4284 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4289 // e.g., when invoking word-findadv from mini-buffer wither with
4290 // wrong options syntax or before ever opening advanced F&R pane
4291 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4295 MatchStringAdv matchAdv(bv->buffer(), opt);
4297 if (!matchAdv.regexIsValid) {
4298 bv->message(lyx::from_utf8(matchAdv.regexError));
4302 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4304 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4305 num_replaced += findAdvReplace(bv, opt, matchAdv);
4308 pos_len = findForwardAdv(cur, matchAdv);
4310 pos_len = findBackwardsAdv(cur, matchAdv);
4311 } catch (exception & ex) {
4312 bv->message(from_utf8(ex.what()));
4317 if (num_replaced > 0) {
4318 switch (num_replaced)
4321 bv->message(_("One match has been replaced."));
4324 bv->message(_("Two matches have been replaced."));
4327 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4333 bv->message(_("Match not found."));
4338 if (num_replaced > 0)
4339 bv->message(_("Match has been replaced."));
4341 bv->message(_("Match found."));
4343 if (cur.pos() + pos_len > cur.lastpos()) {
4344 // Prevent crash in bv->putSelectionAt()
4345 // Should never happen, maybe LASSERT() here?
4346 pos_len = cur.lastpos() - cur.pos();
4348 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4349 bv->putSelectionAt(cur, pos_len, !opt.forward);
4355 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4357 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4358 << opt.casesensitive << ' '
4359 << opt.matchword << ' '
4360 << opt.forward << ' '
4361 << opt.expandmacros << ' '
4362 << opt.ignoreformat << ' '
4363 << opt.replace_all << ' '
4364 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4365 << opt.keep_case << ' '
4366 << int(opt.scope) << ' '
4369 LYXERR(Debug::FIND, "built: " << os.str());
4375 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4377 LYXERR(Debug::FIND, "parsing");
4381 while (line != "EOSS") {
4385 if (is.eof()) // Tolerate malformed request
4389 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4390 opt.find_buf_name = from_utf8(s);
4391 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4392 is.get(); // Waste space before replace string
4395 while (line != "EOSS") {
4399 if (is.eof()) // Tolerate malformed request
4403 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4404 opt.repl_buf_name = from_utf8(s);
4405 is >> opt.keep_case;
4408 opt.scope = FindAndReplaceOptions::SearchScope(i);
4410 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4412 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4413 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4414 << opt.scope << ' ' << opt.restr);