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"
36 #include "frontends/Application.h"
37 #include "frontends/alert.h"
39 #include "mathed/InsetMath.h"
40 #include "mathed/InsetMathHull.h"
41 #include "mathed/MathData.h"
42 #include "mathed/MathStream.h"
43 #include "mathed/MathSupport.h"
45 #include "support/debug.h"
46 #include "support/docstream.h"
47 #include "support/FileName.h"
48 #include "support/gettext.h"
49 #include "support/lassert.h"
50 #include "support/lstrings.h"
51 #include "support/textutils.h"
53 #include <unordered_map>
56 //#define ResultsDebug
57 #define USE_QT_FOR_SEARCH
58 #if defined(USE_QT_FOR_SEARCH)
59 #include <QtCore> // sets QT_VERSION
60 #if (QT_VERSION >= 0x050000)
61 #include <QRegularExpression>
71 using namespace lyx::support;
75 typedef unordered_map<string, string> AccentsMap;
76 typedef unordered_map<string,string>::const_iterator AccentsIterator;
77 static AccentsMap accents = unordered_map<string, string>();
79 // Helper class for deciding what should be ignored
83 IgnoreFormats() = default;
85 bool getFamily() const { return ignoreFamily_; }
87 bool getSeries() const { return ignoreSeries_; }
89 bool getShape() const { return ignoreShape_; }
91 bool getUnderline() const { return ignoreUnderline_; }
93 bool getMarkUp() const { return ignoreMarkUp_; }
95 bool getStrikeOut() const { return ignoreStrikeOut_; }
97 bool getSectioning() const { return ignoreSectioning_; }
99 bool getFrontMatter() const { return ignoreFrontMatter_; }
101 bool getColor() const { return ignoreColor_; }
103 bool getLanguage() const { return ignoreLanguage_; }
105 bool getDeleted() const { return ignoreDeleted_; }
107 void setIgnoreDeleted(bool value);
109 void setIgnoreFormat(string const & type, bool value, bool fromUser = true);
113 bool ignoreFamily_ = false;
115 bool ignoreSeries_ = false;
117 bool ignoreShape_ = false;
119 bool ignoreUnderline_ = false;
121 bool ignoreMarkUp_ = false;
123 bool ignoreStrikeOut_ = false;
125 bool ignoreSectioning_ = false;
127 bool ignoreFrontMatter_ = false;
129 bool ignoreColor_ = false;
131 bool ignoreLanguage_ = false;
132 bool userSelectedIgnoreLanguage_ = false;
134 bool ignoreDeleted_ = true;
137 void IgnoreFormats::setIgnoreFormat(string const & type, bool value, bool fromUser)
139 if (type == "color") {
140 ignoreColor_ = value;
142 else if (type == "language") {
144 userSelectedIgnoreLanguage_ = value;
145 ignoreLanguage_ = value;
148 ignoreLanguage_ = (value || userSelectedIgnoreLanguage_);
150 else if (type == "sectioning") {
151 ignoreSectioning_ = value;
152 ignoreFrontMatter_ = value;
154 else if (type == "font") {
155 ignoreSeries_ = value;
156 ignoreShape_ = value;
157 ignoreFamily_ = value;
159 else if (type == "series") {
160 ignoreSeries_ = value;
162 else if (type == "shape") {
163 ignoreShape_ = value;
165 else if (type == "family") {
166 ignoreFamily_ = value;
168 else if (type == "markup") {
169 ignoreMarkUp_ = value;
171 else if (type == "underline") {
172 ignoreUnderline_ = value;
174 else if (type == "strike") {
175 ignoreStrikeOut_ = value;
177 else if (type == "deleted") {
178 ignoreDeleted_ = value;
182 // The global variable that can be changed from outside
183 IgnoreFormats ignoreFormats;
186 void setIgnoreFormat(string const & type, bool value, bool fromUser)
188 ignoreFormats.setIgnoreFormat(type, value, fromUser);
194 bool parse_bool(docstring & howto, bool const defvalue = false)
199 howto = split(howto, var, ' ');
207 MatchString(docstring const & s, bool cs, bool mw)
208 : str(s), case_sens(cs), whole_words(mw)
211 // returns true if the specified string is at the specified position
212 // del specifies whether deleted strings in ct mode will be considered
213 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
215 return par.find(str, case_sens, whole_words, pos, del);
223 // match whole words only
228 int findForward(DocIterator & cur, DocIterator const endcur,
229 MatchString const & match,
230 bool find_del = true, bool onlysel = false)
232 for (; cur; cur.forwardChar()) {
233 if (onlysel && endcur.pit() == cur.pit()
234 && endcur.idx() == cur.idx() && endcur.pos() < cur.pos())
236 if (cur.inTexted()) {
237 int len = match(cur.paragraph(), cur.pos(), find_del);
246 int findBackwards(DocIterator & cur, DocIterator const endcur,
247 MatchString const & match,
248 bool find_del = true, bool onlysel = false)
252 if (onlysel && endcur.pit() == cur.pit()
253 && endcur.idx() == cur.idx() && endcur.pos() > cur.pos())
255 if (cur.inTexted()) {
256 int len = match(cur.paragraph(), cur.pos(), find_del);
265 bool searchAllowed(docstring const & str)
268 frontend::Alert::error(_("Search error"), _("Search string is empty"));
277 bool findOne(BufferView * bv, docstring const & searchstr,
278 bool case_sens, bool whole, bool forward,
279 bool find_del, bool check_wrap, bool auto_wrap,
280 bool instant, bool onlysel)
282 // Clean up previous selections with empty searchstr on instant
283 if (searchstr.empty() && instant) {
284 if (bv->cursor().selection()) {
285 bv->setCursor(bv->cursor().selectionBegin());
286 bv->clearSelection();
291 if (!searchAllowed(searchstr))
294 DocIterator const endcur = forward ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin();
296 if (onlysel && bv->cursor().selection()) {
297 docstring const matchstring = bv->cursor().selectionAsString(false);
298 docstring const lcmatchsting = support::lowercase(matchstring);
299 if (matchstring == searchstr || (!case_sens && lcmatchsting == lowercase(searchstr))) {
300 docstring q = _("The search string matches the selection, and search is limited to selection.\n"
301 "Continue search outside?");
302 int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
303 q, 0, 1, _("&Yes"), _("&No"));
304 if (search_answer == 0) {
305 bv->clearSelection();
306 if (findOne(bv, searchstr, case_sens, whole, forward,
307 find_del, check_wrap, auto_wrap, false, false))
314 DocIterator cur = forward
315 ? ((instant || onlysel) ? bv->cursor().selectionBegin() : bv->cursor().selectionEnd())
316 : ((instant || onlysel) ? bv->cursor().selectionEnd() : bv->cursor().selectionBegin());
318 MatchString const match(searchstr, case_sens, whole);
320 int match_len = forward
321 ? findForward(cur, endcur, match, find_del, onlysel)
322 : findBackwards(cur, endcur, match, find_del, onlysel);
325 bv->putSelectionAt(cur, match_len, !forward);
327 docstring q = _("The search string was not found within the selection.\n"
328 "Continue search outside?");
329 int search_answer = frontend::Alert::prompt(_("Search outside selection?"),
330 q, 0, 1, _("&Yes"), _("&No"));
331 if (search_answer == 0) {
332 bv->clearSelection();
333 if (findOne(bv, searchstr, case_sens, whole, forward,
334 find_del, check_wrap, auto_wrap, false, false))
339 else if (check_wrap) {
340 DocIterator cur_orig(bv->cursor());
344 q = _("End of file reached while searching forward.\n"
345 "Continue searching from the beginning?");
347 q = _("Beginning of file reached while searching backward.\n"
348 "Continue searching from the end?");
349 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
350 q, 0, 1, _("&Yes"), _("&No"));
351 auto_wrap = wrap_answer == 0;
355 bv->cursor().clear();
356 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
358 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
359 bv->cursor().backwardPos();
361 bv->clearSelection();
362 if (findOne(bv, searchstr, case_sens, whole, forward,
363 find_del, false, false, false, false))
366 bv->cursor().setCursor(cur_orig);
370 return match_len > 0;
376 int replaceAll(BufferView * bv,
377 docstring const & searchstr, docstring const & replacestr,
378 bool case_sens, bool whole, bool onlysel)
380 Buffer & buf = bv->buffer();
382 if (!searchAllowed(searchstr) || buf.isReadonly())
385 DocIterator startcur = bv->cursor().selectionBegin();
386 DocIterator endcur = bv->cursor().selectionEnd();
387 bool const had_selection = bv->cursor().selection();
389 MatchString const match(searchstr, case_sens, whole);
392 int const rsize = replacestr.size();
393 int const ssize = searchstr.size();
396 cur.setCursor(doc_iterator_begin(&buf));
397 int match_len = findForward(cur, endcur, match, false, onlysel);
398 while (match_len > 0) {
399 // Backup current cursor position and font.
400 pos_type const pos = cur.pos();
401 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
403 int ct_deleted_text = ssize -
404 cur.paragraph().eraseChars(pos, pos + match_len,
405 buf.params().track_changes);
406 cur.paragraph().insert(pos, replacestr, font,
407 Change(buf.params().track_changes
409 : Change::UNCHANGED));
410 for (int i = 0; i < rsize + ct_deleted_text
411 && cur.pos() < cur.lastpos(); ++i)
413 if (onlysel && cur.pit() == endcur.pit() && cur.idx() == endcur.idx()) {
414 // Adjust end of selection for replace-all in selection
416 int const offset = rsize - ssize;
417 for (int i = 0; i < offset + ct_deleted_text
418 && endcur.pos() < endcur.lastpos(); ++i)
421 int const offset = ssize - rsize;
422 for (int i = 0; i < offset && endcur.pos() > 0; ++i)
423 endcur.backwardPos();
424 for (int i = 0; i < ct_deleted_text
425 && endcur.pos() < endcur.lastpos(); ++i)
430 match_len = findForward(cur, endcur, match, false, onlysel);
433 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
435 startcur.fixIfBroken();
436 bv->setCursor(startcur);
438 // Reset selection, accounting for changes in selection
440 endcur.fixIfBroken();
441 bv->cursor().resetAnchor();
442 bv->setCursorSelectionTo(endcur);
449 // the idea here is that we are going to replace the string that
450 // is selected IF it is the search string.
451 // if there is a selection, but it is not the search string, then
452 // we basically ignore it. (FIXME We ought to replace only within
454 // if there is no selection, then:
455 // (i) if some search string has been provided, then we find it.
456 // (think of how the dialog works when you hit "replace" the
458 // (ii) if no search string has been provided, then we treat the
459 // word the cursor is in as the search string. (why? i have no
460 // idea.) but this only works in text?
462 // returns the number of replacements made (one, if any) and
463 // whether anything at all was done.
464 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
465 docstring const & replacestr, bool case_sens,
466 bool whole, bool forward, bool findnext, bool wrap,
469 Cursor & cur = bv->cursor();
470 if (!cur.selection() || onlysel) {
471 // no selection, non-empty search string: find it
472 if (!searchstr.empty()) {
473 bool const found = findOne(bv, searchstr, case_sens, whole,
474 forward, true, findnext, wrap, false, onlysel);
475 return make_pair(found, 0);
477 // empty search string
480 return make_pair(false, 0);
481 // select current word and treat it as the search string.
482 // This causes a minor bug as undo will restore this selection,
483 // which the user did not create (#8986).
484 cur.innerText()->selectWord(cur, WHOLE_WORD);
485 searchstr = cur.selectionAsString(false, true);
488 // if we still don't have a search string, report the error
490 if (!searchAllowed(searchstr))
491 return make_pair(false, 0);
493 bool have_selection = cur.selection();
494 docstring const selected = cur.selectionAsString(false, true);
497 ? searchstr == selected
498 : compare_no_case(searchstr, selected) == 0;
500 // no selection or current selection is not search word:
501 // just find the search word
502 if (!have_selection || !match) {
503 bool const found = findOne(bv, searchstr, case_sens, whole, forward,
504 true, findnext, wrap, false, onlysel);
505 return make_pair(found, 0);
508 // we're now actually ready to replace. if the buffer is
509 // read-only, we can't, though.
510 if (bv->buffer().isReadonly())
511 return make_pair(false, 0);
513 cap::replaceSelectionWithString(cur, replacestr);
515 cur.pos() += replacestr.length();
516 LASSERT(cur.pos() <= cur.lastpos(),
517 cur.pos() = cur.lastpos());
520 findOne(bv, searchstr, case_sens, whole,
521 forward, false, findnext, wrap, false, onlysel);
523 return make_pair(true, 1);
529 docstring const find2string(docstring const & search,
530 bool casesensitive, bool matchword,
531 bool forward, bool wrap, bool instant,
536 << int(casesensitive) << ' '
537 << int(matchword) << ' '
538 << int(forward) << ' '
540 << int(instant) << ' '
546 docstring const replace2string(docstring const & replace,
547 docstring const & search,
548 bool casesensitive, bool matchword,
549 bool all, bool forward, bool findnext,
550 bool wrap, bool onlysel)
553 ss << replace << '\n'
555 << int(casesensitive) << ' '
556 << int(matchword) << ' '
558 << int(forward) << ' '
559 << int(findnext) << ' '
566 docstring const string2find(docstring const & argument,
574 // data is of the form
576 // <casesensitive> <matchword> <forward> <wrap> <onlysel>"
578 docstring howto = split(argument, search, '\n');
580 casesensitive = parse_bool(howto);
581 matchword = parse_bool(howto);
582 forward = parse_bool(howto, true);
583 wrap = parse_bool(howto);
584 instant = parse_bool(howto);
585 onlysel = parse_bool(howto);
591 bool lyxfind(BufferView * bv, FuncRequest const & ev)
593 if (!bv || ev.action() != LFUN_WORD_FIND)
596 //lyxerr << "find called, cmd: " << ev << endl;
604 docstring search = string2find(ev.argument(), casesensitive,
605 matchword, forward, wrap, instant, onlysel);
607 return findOne(bv, search, casesensitive, matchword, forward,
608 false, true, wrap, instant, onlysel);
612 bool lyxreplace(BufferView * bv, FuncRequest const & ev)
614 if (!bv || ev.action() != LFUN_WORD_REPLACE)
617 // data is of the form
620 // <casesensitive> <matchword> <all> <forward> <findnext> <wrap> <onlysel>"
623 docstring howto = split(ev.argument(), rplc, '\n');
624 howto = split(howto, search, '\n');
626 bool casesensitive = parse_bool(howto);
627 bool matchword = parse_bool(howto);
628 bool all = parse_bool(howto);
629 bool forward = parse_bool(howto, true);
630 bool findnext = parse_bool(howto, true);
631 bool wrap = parse_bool(howto);
632 bool onlysel = parse_bool(howto);
634 if (!bv->cursor().selection())
635 // only selection only makes sense with selection
640 int replace_count = 0;
642 replace_count = replaceAll(bv, search, rplc, casesensitive,
644 update = replace_count > 0;
647 replaceOne(bv, search, rplc, casesensitive, matchword,
648 forward, findnext, wrap, onlysel);
650 replace_count = rv.second;
653 Buffer const & buf = bv->buffer();
655 // emit message signal.
657 buf.message(_("String not found in selection."));
659 buf.message(_("String not found."));
661 if (replace_count == 0) {
662 buf.message(_("String found."));
663 } else if (replace_count == 1) {
664 buf.message(_("String has been replaced."));
666 docstring const str = onlysel
667 ? bformat(_("%1$d strings have been replaced in the selection."), replace_count)
668 : bformat(_("%1$d strings have been replaced."), replace_count);
676 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
678 for (; cur; cur.forwardPos())
679 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
683 DocIterator cur_orig(bv->cursor());
684 docstring q = _("End of file reached while searching forward.\n"
685 "Continue searching from the beginning?");
686 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
687 q, 0, 1, _("&Yes"), _("&No"));
688 if (wrap_answer == 0) {
689 bv->cursor().clear();
690 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
691 bv->clearSelection();
692 cur.setCursor(bv->cursor().selectionBegin());
693 if (findNextChange(bv, cur, false))
696 bv->cursor().setCursor(cur_orig);
703 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
705 for (cur.backwardPos(); cur; cur.backwardPos()) {
706 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
711 DocIterator cur_orig(bv->cursor());
712 docstring q = _("Beginning of file reached while searching backward.\n"
713 "Continue searching from the end?");
714 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
715 q, 0, 1, _("&Yes"), _("&No"));
716 if (wrap_answer == 0) {
717 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
718 bv->cursor().backwardPos();
719 bv->clearSelection();
720 cur.setCursor(bv->cursor().selectionBegin());
721 if (findPreviousChange(bv, cur, false))
724 bv->cursor().setCursor(cur_orig);
731 bool selectChange(Cursor & cur, bool forward)
733 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
735 Change ch = cur.paragraph().lookupChange(cur.pos());
737 CursorSlice tip1 = cur.top();
738 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
739 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
740 if (!ch2.isSimilarTo(ch))
743 CursorSlice tip2 = cur.top();
744 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
746 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
747 if (!ch2.isSimilarTo(ch)) {
748 // take a step forward to correctly set the selection
756 cur.bv().mouseSetCursor(cur, false);
758 cur.bv().mouseSetCursor(cur, true);
766 bool findChange(BufferView * bv, bool forward)
769 cur.setCursor(forward ? bv->cursor().selectionEnd()
770 : bv->cursor().selectionBegin());
771 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
772 return selectChange(cur, forward);
777 bool findNextChange(BufferView * bv)
779 return findChange(bv, true);
783 bool findPreviousChange(BufferView * bv)
785 return findChange(bv, false);
792 typedef vector<pair<string, string> > Escapes;
794 string string2regex(string in)
796 static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\$\\])" };
797 string temp = std::regex_replace(in, specialChars, R"(\$&)" );
802 while (fl_pos < temp.size()) {
803 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
804 if (fl_pos == string::npos)
807 temp2 += temp.substr(lastpos, fl_pos - lastpos);
813 if (lastpos < temp.size()) {
814 temp2 += temp.substr(lastpos, temp.size() - lastpos);
819 string correctRegex(string t, bool withformat)
821 /* Convert \backslash => \
822 * and \{, \}, \[, \] => {, }, [, ]
825 regex wordre("(\\\\)*(\\\\((backslash|mathcircumflex) ?|[\\[\\]\\{\\}]))");
828 bool backslashed = false;
829 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
832 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
836 if (sub.str(4) == "backslash") {
839 // transforms '\backslash \{' into '\{'
841 string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
842 if ((next == "\\{") || (next == "\\}")) {
848 else if (sub.str(4) == "mathcircumflex")
850 else if (backslashed) {
852 if (withformat && (sub.str(3) == "{"))
853 replace = accents["braceleft"];
854 else if (withformat && (sub.str(3) == "}"))
855 replace = accents["braceright"];
857 // else part should not exist
862 replace = sub.str(3);
864 if (lastpos < (size_t) sub.position(2))
865 s += t.substr(lastpos, sub.position(2) - lastpos);
867 lastpos = sub.position(2) + sub.length(2);
871 else if (lastpos < t.length())
872 s += t.substr(lastpos, t.length() - lastpos);
876 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
877 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
878 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
879 string escape_for_regex(string s, bool withformat)
883 while (lastpos < s.size()) {
884 size_t regex_pos = s.find("\\regexp{", lastpos);
885 if (regex_pos == string::npos) {
886 regex_pos = s.size();
888 if (regex_pos > lastpos) {
889 result += string2regex(s.substr(lastpos, regex_pos-lastpos));
891 if (lastpos == s.size())
894 size_t end_pos = s.find("\\endregexp{}}", regex_pos + 8);
895 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
896 lastpos = end_pos + 13;
902 /// Wrapper for lyx::regex_replace with simpler interface
903 bool regex_replace(string const & s, string & t, string const & searchstr,
904 string const & replacestr)
906 regex e(searchstr, regex_constants::ECMAScript);
908 ostream_iterator<char, char> it(oss);
909 regex_replace(it, s.begin(), s.end(), e, replacestr);
910 // tolerate t and s be references to the same variable
911 bool rv = (s != oss.str());
930 vector <string> result = vector <string>();
931 MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {};
934 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
936 if (newres.match2end < oldres.match2end)
937 return MatchResult::newIsTooFar;
938 if (newres.match_len < oldres.match_len)
939 return MatchResult::newIsTooFar;
941 if (newres.match_len == oldres.match_len) {
942 if (newres.match2end == oldres.match2end)
943 return MatchResult::newIsBetter;
945 return MatchResult::newIsInvalid;
948 /** The class performing a match between a position in the document and the FindAdvOptions.
951 class MatchStringAdv {
953 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
955 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
956 ** constructor as opt.search, under the opt.* options settings.
959 ** If set, then match is searched only against beginning of text starting at cur.
960 ** If unset, then match is searched anywhere in text starting at cur.
963 ** The length of the matching text, or zero if no match was found.
965 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
974 /// first buffer on which search was started
975 lyx::Buffer * const p_first_buf;
977 FindAndReplaceOptions const & opt;
980 /// Auxiliary find method (does not account for opt.matchword)
981 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
982 void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
984 /** Normalize a stringified or latexified LyX paragraph.
988 ** <li>if search is not casesensitive, then lowercase the string;
989 ** <li>remove any newline at begin or end of the string;
990 ** <li>replace any newline in the middle of the string with a simple space;
991 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
994 ** @todo Normalization should also expand macros, if the corresponding
995 ** search option was checked.
997 string normalize(docstring const & s) const;
998 // normalized string to search
999 string par_as_string;
1000 // regular expression to use for searching
1001 // regexp2 is same as regexp, but prefixed with a ".*?"
1003 QRegularExpression regexp;
1004 QRegularExpression regexp2;
1009 // leading format material as string
1010 string lead_as_string;
1011 // par_as_string after removal of lead_as_string
1012 string par_as_string_nolead;
1013 // unmatched open braces in the search string/regexp
1015 // number of (.*?) subexpressions added at end of search regexp for closing
1016 // environments, math mode, styles, etc...
1017 int close_wildcards;
1019 // Are we searching with regular expressions ?
1021 static int valid_matches;
1022 static vector <string> matches;
1023 void FillResults(MatchResult &found_mr);
1026 int MatchStringAdv::valid_matches = 0;
1027 vector <string> MatchStringAdv::matches = vector <string>(10);
1029 void MatchStringAdv::FillResults(MatchResult &found_mr)
1031 if (found_mr.match_len > 0) {
1032 valid_matches = found_mr.result.size();
1033 for (size_t i = 0; i < found_mr.result.size(); i++)
1034 matches[i] = found_mr.result[i];
1040 static docstring buffer_to_latex(Buffer & buffer)
1042 //OutputParams runparams(&buffer.params().encoding());
1043 OutputParams runparams(encodings.fromLyXName("utf8"));
1044 odocstringstream ods;
1046 runparams.nice = true;
1047 runparams.flavor = Flavor::XeTeX;
1048 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1049 // No side effect of file copying and image conversion
1050 runparams.dryrun = true;
1051 if (ignoreFormats.getDeleted())
1052 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
1054 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
1055 pit_type const endpit = buffer.paragraphs().size();
1056 for (pit_type pit = 0; pit != endpit; ++pit) {
1057 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
1058 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
1064 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
1067 if (!opt.ignoreformat) {
1068 str = buffer_to_latex(buffer);
1070 // OutputParams runparams(&buffer.params().encoding());
1071 OutputParams runparams(encodings.fromLyXName("utf8"));
1072 runparams.nice = true;
1073 runparams.flavor = Flavor::XeTeX;
1074 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
1075 runparams.dryrun = true;
1076 int option = AS_STR_INSETS |AS_STR_PLAINTEXT;
1077 if (ignoreFormats.getDeleted()) {
1078 option |= AS_STR_SKIPDELETE;
1079 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
1082 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
1084 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1085 Paragraph const & par = buffer.paragraphs().at(pit);
1086 LYXERR(Debug::FIND, "Adding to search string: '"
1087 << par.asString(pos_type(0), par.size(),
1091 str += par.asString(pos_type(0), par.size(),
1095 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1096 string t = to_utf8(str);
1097 while (regex_replace(t, t, "\\\\(text|lyxmathsym|ensuremath)\\{([^\\}]*)\\}", "$2"));
1104 /// Return separation pos between the leading material and the rest
1105 static size_t identifyLeading(string const & s)
1108 // @TODO Support \item[text]
1109 // Kornel: Added textsl, textsf, textit, texttt and noun
1110 // + allow to search for colored text too
1111 while (regex_replace(t, t, "^\\\\(("
1112 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1113 "lyxaddress|lyxrightaddress|"
1114 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1115 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1116 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1117 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1118 || regex_replace(t, t, "^\\$", "")
1119 || regex_replace(t, t, "^\\\\\\[", "")
1120 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1121 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1123 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1128 * Given a latexified string, retrieve some handled features
1129 * The features of the regex will later be compared with the features
1130 * of the searched text. If the regex features are not a
1131 * subset of the analized, then, in not format ignoring search
1132 * we can early stop the search in the relevant inset.
1134 typedef map<string, bool> Features;
1136 static Features identifyFeatures(string const & s)
1138 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1139 static regex const valid("^("
1141 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1142 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1143 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1144 "(u|uu)line|(s|x)out|uwave|"
1145 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1146 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1151 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1154 if (sub.str(1).compare("regexp") == 0) {
1158 string token = sub.str(1);
1160 if (regex_match(token, sub2, valid)) {
1168 if (sub.str(1).compare("endregexp") == 0) {
1178 * defines values features of a key "\\[a-z]+{"
1183 /* Char type with content discarded
1184 * like \hspace{1cm} */
1186 /* Char, like \backslash */
1188 /* replace starting backslash with '#' */
1190 /* \part, \section*, ... */
1192 /* title, author etc */
1194 /* \foreignlanguage{ngerman}, ... */
1197 * to discard language in content */
1200 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1202 /* fonts, colors, markups, ... */
1204 /* footnotesize, ... large, ...
1205 * Ignore all of them */
1208 /* inputencoding, ...
1209 * Discard also content, because they do not help in search */
1212 * like remove, but also all arguments */
1214 /* item, listitem */
1216 /* tex, latex, ... like isChar */
1218 /* like \lettrine[lines=5]{}{} */
1220 // like isStandard, but always remove head
1222 /* End of arguments marker for lettrine,
1223 * so that they can be ignored */
1226 KeyInfo() = default;
1227 KeyInfo(KeyType type, int parcount, bool disable)
1229 parenthesiscount(parcount),
1230 disabled(disable) {}
1231 KeyType keytype = invalid;
1233 int _tokensize = -1;
1234 int _tokenstart = -1;
1235 int _dataStart = -1;
1237 int parenthesiscount = 1;
1238 bool disabled = false;
1239 bool used = false; /* by pattern */
1244 Border(int l=0, int u=0) : low(l), upper(u) {};
1249 #define MAXOPENED 30
1251 bool isPatternString_;
1253 explicit Intervall(bool isPattern, string const & p) :
1254 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1255 hasTitle(false), langcount(0)
1263 static vector<Border> borders;
1264 int depts[MAXOPENED];
1265 int closes[MAXOPENED];
1266 int actualdeptindex;
1267 int previousNotIgnored(int) const;
1268 int nextNotIgnored(int) const;
1269 void handleOpenP(int i);
1270 void handleCloseP(int i, bool closingAllowed);
1271 void resetOpenedP(int openPos);
1272 void addIntervall(int upper);
1273 void addIntervall(int low, int upper); /* if explicit */
1274 void removeAccents();
1275 void setForDefaultLang(KeyInfo const & defLang) const;
1276 int findclosing(int start, int end, char up, char down, int repeat);
1277 void handleParentheses(int lastpos, bool closingAllowed);
1279 int langcount; // Number of disabled language specs up to current position in actual interval
1280 int isOpeningPar(int pos) const;
1282 void output(ostringstream &os, int lastpos);
1283 // string show(int lastpos);
1286 vector<Border> Intervall::borders = vector<Border>(30);
1288 int Intervall::isOpeningPar(int pos) const
1290 if ((pos < 0) || (size_t(pos) >= par.size()))
1292 if (par[pos] != '{')
1294 if (size_t(pos) + 2 >= par.size())
1296 if (par[pos+2] != '}')
1298 if (par[pos+1] == '[' || par[pos+1] == ']')
1303 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1305 // Enable the use of first token again
1306 if (ignoreidx >= 0) {
1307 int value = defLang._tokenstart + defLang._tokensize;
1313 if (borders[borderidx].low < value)
1314 borders[borderidx].low = value;
1315 if (borders[borderidx].upper < value)
1316 borders[borderidx].upper = value;
1321 static void checkDepthIndex(int val)
1323 static int maxdepthidx = MAXOPENED-2;
1324 static int lastmaxdepth = 0;
1325 if (val > lastmaxdepth) {
1326 LYXERR(Debug::INFO, "Depth reached " << val);
1329 if (val > maxdepthidx) {
1331 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1336 // Not needed, because borders are now dynamically expanded
1337 static void checkIgnoreIdx(int val)
1339 static int lastmaxignore = -1;
1340 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1341 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1342 lastmaxignore = val;
1348 * Expand the region of ignored parts of the input latex string
1349 * The region is only relevant in output()
1351 void Intervall::addIntervall(int low, int upper)
1354 if (low == upper) return;
1355 for (idx = ignoreidx+1; idx > 0; --idx) {
1356 if (low > borders[idx-1].upper) {
1360 Border br(low, upper);
1361 if (idx > ignoreidx) {
1362 if (borders.size() <= size_t(idx)) {
1363 borders.push_back(br);
1369 // checkIgnoreIdx(ignoreidx);
1373 // Expand only if one of the new bound is inside the interwall
1374 // We know here that br.low > borders[idx-1].upper
1375 if (br.upper < borders[idx].low) {
1376 // We have to insert at this pos
1377 if (size_t(ignoreidx+1) >= borders.size()) {
1378 borders.push_back(borders[ignoreidx]);
1381 borders[ignoreidx+1] = borders[ignoreidx];
1383 for (int i = ignoreidx; i > idx; --i) {
1384 borders[i] = borders[i-1];
1388 // checkIgnoreIdx(ignoreidx);
1391 // Here we know, that we are overlapping
1392 if (br.low > borders[idx].low)
1393 br.low = borders[idx].low;
1394 // check what has to be concatenated
1396 for (int i = idx; i <= ignoreidx; i++) {
1397 if (br.upper >= borders[i].low) {
1399 if (br.upper < borders[i].upper)
1400 br.upper = borders[i].upper;
1406 // count should be >= 1 here
1409 for (int i = idx + count; i <= ignoreidx; i++) {
1410 borders[i-count+1] = borders[i];
1412 ignoreidx -= count - 1;
1418 static void buildaccent(string n, string param, string values)
1422 const char delim = '|';
1423 while (getline(s, name, delim)) {
1425 for (char c : param) {
1426 string key = name + "{" + c + "}";
1427 // get the corresponding utf8-value
1428 if ((values[start] & 0xc0) != 0xc0) {
1429 // should not happen, utf8 encoding starts at least with 11xxxxxx
1430 // but value for '\dot{i}' is 'i', which is ascii
1431 if ((values[start] & 0x80) == 0) {
1433 accents[key] = values.substr(start, 1);
1434 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1439 for (int j = 1; ;j++) {
1440 if (start + j >= values.size()) {
1441 accents[key] = values.substr(start, j);
1442 start = values.size() - 1;
1445 else if ((values[start+j] & 0xc0) != 0x80) {
1446 // This is the first byte of following utf8 char
1447 accents[key] = values.substr(start, j);
1449 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1458 static string getutf8(unsigned uchar)
1461 string ret = string();
1463 if (uchar <= 0x7f) {
1464 c[maxc-1] = uchar & 0x7f;
1467 unsigned char rest = 0x40;
1468 unsigned char first = 0x80;
1470 for (int i = start; i >=0; --i) {
1472 c[i] = first + uchar;
1475 c[i] = 0x80 | (uchar & 0x3f);
1482 for (int i = 0; i < maxc; i++) {
1483 if (c[i] == 0) continue;
1489 static void buildAccentsMap()
1491 accents["imath"] = "ı";
1493 accents["jmath"] = "ȷ";
1494 accents["cdot"] = "·";
1495 accents["textasciicircum"] = "^";
1496 accents["mathcircumflex"] = "^";
1497 accents["sim"] = "~";
1498 accents["guillemotright"] = "»";
1499 accents["guillemotleft"] = "«";
1500 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1501 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1502 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1503 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1504 accents["negmedspace"] = getutf8(0xf0005);
1505 accents["thickspace"] = getutf8(0xf0006);
1506 accents["negthickspace"] = getutf8(0xf0007);
1507 accents["lyx"] = getutf8(0xf0010); // Used logos
1508 accents["LyX"] = getutf8(0xf0010);
1509 accents["tex"] = getutf8(0xf0011);
1510 accents["TeX"] = getutf8(0xf0011);
1511 accents["latex"] = getutf8(0xf0012);
1512 accents["LaTeX"] = getutf8(0xf0012);
1513 accents["latexe"] = getutf8(0xf0013);
1514 accents["LaTeXe"] = getutf8(0xf0013);
1515 accents["lyxarrow"] = getutf8(0xf0020);
1516 accents["braceleft"] = getutf8(0xf0030);
1517 accents["braceright"] = getutf8(0xf0031);
1518 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1519 accents["backslash LyX"] = getutf8(0xf0010);
1520 accents["backslash tex"] = getutf8(0xf0011);
1521 accents["backslash TeX"] = getutf8(0xf0011);
1522 accents["backslash latex"] = getutf8(0xf0012);
1523 accents["backslash LaTeX"] = getutf8(0xf0012);
1524 accents["backslash latexe"] = getutf8(0xf0013);
1525 accents["backslash LaTeXe"] = getutf8(0xf0013);
1526 accents["backslash lyxarrow"] = getutf8(0xf0020);
1527 accents["ddot{\\imath}"] = "ï";
1528 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1529 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1530 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1531 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1532 accents["acute{\\imath}"] = "í";
1533 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1534 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1535 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1536 buildaccent("mathring|r", "aAuUwy",
1538 accents["check{\\imath}"] = "ǐ";
1539 accents["check{\\jmath}"] = "ǰ";
1540 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1541 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1542 accents["hat{\\imath}"] = "î";
1543 accents["hat{\\jmath}"] = "ĵ";
1544 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1545 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1546 accents["bar{\\imath}"] = "ī";
1547 buildaccent("bar|=", "aAeEiIoOuUyY",
1548 "āĀēĒīĪōŌūŪȳȲ"); // macron
1549 accents["tilde{\\imath}"] = "ĩ";
1550 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1551 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1552 accents["breve{\\imath}"] = "ĭ";
1553 buildaccent("breve|u", "aAeEgGiIoOuU",
1554 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1555 accents["grave{\\imath}"] = "ì";
1556 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1557 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1558 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1559 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1560 buildaccent("ogonek|k", "AaEeIiUuOo",
1561 "ĄąĘęĮįŲųǪǫ"); // ogonek
1562 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1563 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1564 buildaccent("subring|textsubring", "Aa",
1566 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1567 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1568 buildaccent("subtilde|textsubtilde", "EeIiUu",
1569 "ḚḛḬḭṴṵ"); // subtilde
1570 accents["dgrave{\\imath}"] = "ȉ";
1571 accents["textdoublegrave{\\i}"] = "ȉ";
1572 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1573 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1574 accents["rcap{\\imath}"] = "ȉ";
1575 accents["textroundcap{\\i}"] = "ȉ";
1576 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1577 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1578 buildaccent("slashed", "oO",
1583 * Created accents in math or regexp environment
1584 * are macros, but we need the utf8 equivalent
1586 void Intervall::removeAccents()
1588 if (accents.empty())
1590 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1591 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1592 "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))|(brace|guillemot)(left|right)|textasciicircum|mathcircumflex|sim)(?![a-zA-Z]))");
1594 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1596 string key = sub.str(1);
1597 AccentsIterator it_ac = accents.find(key);
1598 if (it_ac != accents.end()) {
1599 string val = it_ac->second;
1600 size_t pos = sub.position(size_t(0));
1601 for (size_t i = 0; i < val.size(); i++) {
1602 par[pos+i] = val[i];
1604 // Remove possibly following space too
1605 if (par[pos+sub.str(0).size()] == ' ')
1606 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1608 addIntervall(pos+val.size(), pos + sub.str(0).size());
1609 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1610 // remove traces of any remaining chars
1615 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1620 void Intervall::handleOpenP(int i)
1623 depts[actualdeptindex] = i+1;
1624 closes[actualdeptindex] = -1;
1625 checkDepthIndex(actualdeptindex);
1628 void Intervall::handleCloseP(int i, bool closingAllowed)
1630 if (actualdeptindex <= 0) {
1631 if (! closingAllowed)
1632 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1633 // if we are at the very end
1634 addIntervall(i, i+1);
1637 closes[actualdeptindex] = i+1;
1642 void Intervall::resetOpenedP(int openPos)
1644 // Used as initializer for foreignlanguage entry
1645 actualdeptindex = 1;
1646 depts[1] = openPos+1;
1650 int Intervall::previousNotIgnored(int start) const
1652 int idx = 0; /* int intervalls */
1653 for (idx = ignoreidx; idx >= 0; --idx) {
1654 if (start > borders[idx].upper)
1656 if (start >= borders[idx].low)
1657 start = borders[idx].low-1;
1662 int Intervall::nextNotIgnored(int start) const
1664 int idx = 0; /* int intervalls */
1665 for (idx = 0; idx <= ignoreidx; idx++) {
1666 if (start < borders[idx].low)
1668 if (start < borders[idx].upper)
1669 start = borders[idx].upper;
1674 typedef unordered_map<string, KeyInfo> KeysMap;
1675 typedef unordered_map<string, KeyInfo>::const_iterator KeysIterator;
1676 typedef vector< KeyInfo> Entries;
1677 static KeysMap keys = unordered_map<string, KeyInfo>();
1683 Intervall interval_;
1684 void buildKeys(bool);
1685 void buildEntries(bool);
1686 void makeKey(const string &, KeyInfo, bool isPatternString);
1687 void processRegion(int start, int region_end); /* remove {} parts */
1688 void removeHead(KeyInfo const &, int count=0);
1691 LatexInfo(string const & par, bool isPatternString)
1692 : entidx_(-1), interval_(isPatternString, par)
1694 buildKeys(isPatternString);
1695 entries_ = vector<KeyInfo>();
1696 buildEntries(isPatternString);
1700 if (entries_.empty()) {
1703 if (entries_[0].keytype == KeyInfo::isTitle) {
1704 interval_.hasTitle = true;
1705 if (! entries_[0].disabled) {
1706 interval_.titleValue = entries_[0].head;
1709 interval_.titleValue = "";
1711 removeHead(entries_[0]);
1712 if (entries_.size() > 1)
1721 if (int(entries_.size()) > entidx_) {
1728 bool setNextKey(int idx) {
1729 if ((idx == entidx_) && (entidx_ >= 0)) {
1736 int find(int start, KeyInfo::KeyType keytype) const {
1740 while (tmpIdx < int(entries_.size())) {
1741 if (entries_[tmpIdx].keytype == keytype)
1747 int process(ostringstream & os, KeyInfo const & actual);
1748 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1749 // string show(int lastpos) { return interval.show(lastpos);};
1750 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1751 KeyInfo &getKeyInfo(int keyinfo) {
1752 static KeyInfo invalidInfo = KeyInfo();
1753 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1756 return entries_[keyinfo];
1758 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1759 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1763 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1767 for (int i = start; i < end; i += 1 + skip) {
1771 if (c == '\\') skip = 1;
1775 else if (c == down) {
1778 if ((repeat <= 0) || (par[i+1] != up))
1792 size_t mathpostfixsize;
1794 size_t mathprefixsize;
1798 vector<MathEntry> entries_;
1803 void insert(string const & wait, size_t start, size_t prefixsize, size_t end, size_t postfixsize) {
1804 MathEntry m = MathEntry();
1806 m.mathStart = start;
1807 m.mathprefixsize = prefixsize;
1808 m.mathEnd = end + postfixsize;
1809 m.mathpostfixsize = postfixsize;
1810 m.mathSize = m.mathEnd - m.mathStart;
1811 entries_.push_back(m);
1813 bool empty() const { return entries_.empty(); };
1814 size_t getEndPos() const {
1815 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1818 return entries_[actualIdx_].mathEnd;
1820 size_t getStartPos() const {
1821 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1822 return 100000; /* definitely enough? */
1824 return entries_[actualIdx_].mathStart;
1826 size_t getPrefixSize() const {
1827 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1830 return entries_[actualIdx_].mathprefixsize;
1832 size_t getPostfixSize() const {
1833 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1836 return entries_[actualIdx_].mathpostfixsize;
1838 size_t getFirstPos() {
1840 return getStartPos();
1842 size_t getSize() const {
1843 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1846 return entries_[actualIdx_].mathSize;
1848 void incrEntry() { actualIdx_++; };
1851 void LatexInfo::buildEntries(bool isPatternString)
1853 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|x?x?alignat)\\*?\\})(\\{[0-9]+\\})?)");
1854 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1855 static bool disableLanguageOverride = false;
1856 smatch sub, submath;
1857 bool evaluatingRegexp = false;
1859 bool evaluatingMath = false;
1860 bool evaluatingCode = false;
1862 bool evaluatingOptional = false;
1863 size_t optionalEnd = 0;
1866 bool math_end_waiting = false;
1867 size_t math_pos = 10000;
1868 size_t math_prefix_size = 1;
1870 static vector<string> usedText = vector<string>();
1871 static bool removeMathHull = false;
1873 interval_.removeAccents();
1875 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1877 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1878 // prefixed by odd count of '\\'
1881 if (math_end_waiting) {
1882 size_t pos = submath.position(size_t(2));
1883 if ((math_end == "$") &&
1884 (submath.str(2) == "$")) {
1885 mi.insert("$", math_pos, 1, pos, 1);
1886 math_end_waiting = false;
1888 else if ((math_end == "\\]") &&
1889 (submath.str(2) == "\\]")) {
1890 mi.insert("\\]", math_pos, 2, pos, 2);
1891 math_end_waiting = false;
1893 else if ((submath.str(3).compare("end") == 0) &&
1894 (submath.str(5).compare(math_end) == 0)) {
1895 mi.insert(math_end, math_pos, math_prefix_size, pos, submath.str(2).length());
1896 math_end_waiting = false;
1902 if (submath.str(3).compare("begin") == 0) {
1903 math_end_waiting = true;
1904 math_end = submath.str(5);
1905 math_pos = submath.position(size_t(2));
1906 math_prefix_size = submath.str(2).length();
1908 else if (submath.str(2).compare("\\[") == 0) {
1909 math_end_waiting = true;
1911 math_pos = submath.position(size_t(2));
1913 else if (submath.str(2) == "$") {
1914 size_t pos = submath.position(size_t(2));
1915 math_end_waiting = true;
1921 // Ignore language if there is math somewhere in pattern-string
1922 if (isPatternString) {
1923 for (auto s: usedText) {
1924 // Remove entries created in previous search runs
1927 usedText = vector<string>();
1930 keys["foreignlanguage"].disabled = true;
1931 disableLanguageOverride = true;
1932 removeMathHull = false;
1935 removeMathHull = true; // used later if not isPatternString
1936 disableLanguageOverride = false;
1940 if (disableLanguageOverride) {
1941 keys["foreignlanguage"].disabled = true;
1944 math_pos = mi.getFirstPos();
1945 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1947 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1948 // prefixed by odd count of '\\'
1951 string key = sub.str(5);
1953 if (sub.str(2)[0] == '\\')
1954 key = sub.str(2)[1];
1959 KeysIterator it_key = keys.find(key);
1960 if (it_key != keys.end()) {
1961 if (it_key->second.keytype == KeyInfo::headRemove) {
1962 KeyInfo found1 = it_key->second;
1963 found1.disabled = true;
1964 found1.head = "\\" + key + "{";
1965 found1._tokenstart = sub.position(size_t(2));
1966 found1._tokensize = found1.head.length();
1967 found1._dataStart = found1._tokenstart + found1.head.length();
1968 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1969 found1._dataEnd = endpos;
1974 if (evaluatingRegexp) {
1975 if (sub.str(3).compare("endregexp") == 0) {
1976 evaluatingRegexp = false;
1977 // found._tokenstart already set
1978 found._dataEnd = sub.position(size_t(2)) + 13;
1979 found._dataStart = found._dataEnd;
1980 found._tokensize = found._dataEnd - found._tokenstart;
1981 found.parenthesiscount = 0;
1982 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1989 if (evaluatingMath) {
1990 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1992 evaluatingMath = false;
1994 math_pos = mi.getStartPos();
1996 if (it_key == keys.end()) {
1997 found = KeyInfo(KeyInfo::isStandard, 0, true);
1998 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1999 found = KeyInfo(KeyInfo::isText, 0, false);
2000 if (isPatternString) {
2001 found.keytype = KeyInfo::isChar;
2002 found.disabled = false;
2006 usedText.push_back(key);
2010 if (key.compare("regexp") == 0) {
2011 evaluatingRegexp = true;
2012 found._tokenstart = sub.position(size_t(2));
2013 found._tokensize = 0;
2017 // Handle the other params of key
2018 if (found.keytype == KeyInfo::isIgnored)
2020 else if (found.keytype == KeyInfo::isMath) {
2021 if (size_t(sub.position(size_t(2))) == math_pos) {
2023 found._tokenstart = sub.position(size_t(2));
2024 found._tokensize = mi.getSize();
2025 found._dataEnd = found._tokenstart + found._tokensize;
2026 found._dataStart = found._dataEnd;
2027 found.parenthesiscount = 0;
2028 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2029 if (removeMathHull) {
2030 interval_.addIntervall(found._tokenstart, found._tokenstart + mi.getPrefixSize());
2031 interval_.addIntervall(found._dataEnd - mi.getPostfixSize(), found._dataEnd);
2034 // Treate all math constructs as simple math
2035 interval_.par[found._tokenstart] = '$';
2036 interval_.par[found._dataEnd - mi.getPostfixSize()] = '$';
2037 interval_.addIntervall(found._tokenstart + 1, found._tokenstart + mi.getPrefixSize());
2038 interval_.addIntervall(found._dataEnd - mi.getPostfixSize() + 1, found._dataEnd);
2040 evaluatingMath = true;
2043 // begin|end of unknown env, discard
2044 // First handle tables
2045 // longtable|tabular
2046 bool discardComment;
2048 found.keytype = KeyInfo::doRemove;
2049 if ((sub.str(7).compare("longtable") == 0) ||
2050 (sub.str(7).compare("tabular") == 0)) {
2051 discardComment = true; /* '%' */
2054 discardComment = false;
2055 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
2057 string token = sub.str(7);
2058 if (regex_match(token, sub2, removeArgs)) {
2059 found.keytype = KeyInfo::removeWithArg;
2062 // discard spaces before pos(2)
2063 int pos = sub.position(size_t(2));
2065 for (count = 0; pos - count > 0; count++) {
2066 char c = interval_.par[pos-count-1];
2067 if (discardComment) {
2068 if ((c != ' ') && (c != '%'))
2074 found._tokenstart = pos - count;
2075 if (sub.str(3).compare(0, 5, "begin") == 0) {
2076 size_t pos1 = pos + sub.str(2).length();
2077 if (sub.str(7).compare("cjk") == 0) {
2078 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2079 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
2081 found.keytype = KeyInfo::isMain;
2082 found._dataStart = pos1;
2083 found._dataEnd = interval_.par.length();
2084 found.disabled = keys["foreignlanguage"].disabled;
2085 found.used = keys["foreignlanguage"].used;
2086 found._tokensize = pos1 - found._tokenstart;
2087 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2090 // Swallow possible optional params
2091 while (interval_.par[pos1] == '[') {
2092 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
2094 // Swallow also the eventual parameter
2095 if (interval_.par[pos1] == '{') {
2096 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
2099 found._dataEnd = pos1;
2101 found._dataStart = found._dataEnd;
2102 found._tokensize = count + found._dataEnd - pos;
2103 found.parenthesiscount = 0;
2104 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2105 found.disabled = true;
2109 // Handle "\end{...}"
2110 found._dataStart = pos + sub.str(2).length();
2111 found._dataEnd = found._dataStart;
2112 found._tokensize = count + found._dataEnd - pos;
2113 found.parenthesiscount = 0;
2114 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
2115 found.disabled = true;
2119 else if (found.keytype != KeyInfo::isRegex) {
2120 found._tokenstart = sub.position(size_t(2));
2121 if (found.parenthesiscount == 0) {
2122 // Probably to be discarded
2123 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2124 char following = interval_.par[following_pos];
2125 if (following == ' ')
2126 found.head = "\\" + sub.str(5) + " ";
2127 else if (following == '=') {
2128 // like \uldepth=1000pt
2129 found.head = sub.str(2);
2132 found.head = "\\" + key;
2133 found._tokensize = found.head.length();
2134 found._dataEnd = found._tokenstart + found._tokensize;
2135 found._dataStart = found._dataEnd;
2138 int params = found._tokenstart + key.length() + 1;
2139 if (evaluatingOptional) {
2140 if (size_t(found._tokenstart) > optionalEnd) {
2141 evaluatingOptional = false;
2144 found.disabled = true;
2147 int optend = params;
2148 while (interval_.par[optend] == '[') {
2149 // discard optional parameters
2150 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2152 if (optend > params) {
2153 key += interval_.par.substr(params, optend-params);
2154 evaluatingOptional = true;
2155 optionalEnd = optend;
2156 if (found.keytype == KeyInfo::isSectioning) {
2157 // Remove optional values (but still keep in header)
2158 interval_.addIntervall(params, optend);
2161 string token = sub.str(7);
2163 if (interval_.par[optend] != '{') {
2165 found.parenthesiscount = 0;
2166 found.head = "\\" + key;
2169 closings = found.parenthesiscount;
2170 if (found.parenthesiscount == 1) {
2171 found.head = "\\" + key + "{";
2173 else if (found.parenthesiscount > 1) {
2175 found.head = sub.str(2) + "{";
2176 closings = found.parenthesiscount - 1;
2179 found.head = "\\" + key + "{";
2182 found._tokensize = found.head.length();
2183 found._dataStart = found._tokenstart + found.head.length();
2184 if (found.keytype == KeyInfo::doRemove) {
2186 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2187 if (endpar >= interval_.par.length())
2188 found._dataStart = interval_.par.length();
2190 found._dataStart = endpar;
2191 found._tokensize = found._dataStart - found._tokenstart;
2194 found._dataStart = found._tokenstart + found._tokensize;
2198 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2199 found._dataStart += 15;
2203 endpos = found._dataStart - 1;
2205 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2206 if (found.keytype == KeyInfo::isList) {
2207 // Check if it really is list env
2208 static regex const listre("^([a-z]+)$");
2210 if (!regex_match(token, sub2, listre)) {
2211 // Change the key of this entry. It is not in a list/item environment
2212 found.keytype = KeyInfo::endArguments;
2215 if (found.keytype == KeyInfo::noMain) {
2216 evaluatingCode = true;
2218 codeStart = found._dataStart;
2220 else if (evaluatingCode) {
2221 if (size_t(found._dataStart) > codeEnd)
2222 evaluatingCode = false;
2223 else if (found.keytype == KeyInfo::isMain) {
2224 // Disable this key, treate it as standard
2225 found.keytype = KeyInfo::isStandard;
2226 found.disabled = true;
2227 if ((codeEnd +1 >= interval_.par.length()) &&
2228 (found._tokenstart == codeStart)) {
2229 // trickery, because the code inset starts
2230 // with \selectlanguage ...
2232 if (entries_.size() > 1) {
2233 entries_[entries_.size()-1]._dataEnd = codeEnd;
2238 if ((endpos == interval_.par.length()) &&
2239 (found.keytype == KeyInfo::doRemove)) {
2240 // Missing closing => error in latex-input?
2241 // therefore do not delete remaining data
2242 found._dataStart -= 1;
2243 found._dataEnd = found._dataStart;
2246 found._dataEnd = endpos;
2248 if (isPatternString) {
2249 keys[key].used = true;
2252 entries_.push_back(found);
2256 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2258 stringstream s(keysstring);
2260 const char delim = '|';
2261 while (getline(s, key, delim)) {
2262 KeyInfo keyII(keyI);
2263 if (isPatternString) {
2266 else if ( !keys[key].used)
2267 keyII.disabled = true;
2272 void LatexInfo::buildKeys(bool isPatternString)
2275 static bool keysBuilt = false;
2276 if (keysBuilt && !isPatternString) return;
2278 // Keys to ignore in any case
2279 makeKey("text|textcyrillic|lyxmathsym|ensuremath", KeyInfo(KeyInfo::headRemove, 1, true), true);
2280 // Known standard keys with 1 parameter.
2281 // Split is done, if not at start of region
2282 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2283 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2284 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2285 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2286 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2287 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2289 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2290 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2291 makeKey("section*|subsection*|subsubsection*|paragraph*",
2292 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2293 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2294 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2296 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2298 // Split is done, if not at start of region
2299 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2300 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2302 // Split is done always.
2303 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2307 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2308 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2309 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2310 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2312 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2313 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2314 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2315 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2317 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2318 // Custom space/skip, remove the content (== length value)
2319 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2320 // Found in fr/UserGuide.lyx
2321 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2323 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2324 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2325 // Known macros to remove (including their parameter)
2327 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2328 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2329 // handle like standard keys with 1 parameter.
2330 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2332 // Ignore deleted text
2333 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2334 // but preserve added text
2335 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2337 // Macros to remove, but let the parameter survive
2339 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2341 // Remove language spec from content of these insets
2342 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2344 // Same effect as previous, parameter will survive (because there is no one anyway)
2346 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2347 // Remove table decorations
2348 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2349 // Discard shape-header.
2350 // For footnote or shortcut too, because of lang settings
2351 // and wrong handling if used 'KeyInfo::noMain'
2352 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2353 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2354 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2355 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2356 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2357 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2358 // like ('tiny{}' or '\tiny ' ... )
2359 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2361 // Survives, like known character
2362 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2363 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2364 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2366 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2367 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2368 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2370 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2371 // Remove RTL/LTR marker
2372 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2373 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2374 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2375 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2376 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2377 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2378 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2379 if (isPatternString) {
2380 // Allow the first searched string to rebuild the keys too
2384 // no need to rebuild again
2390 * Keep the list of actual opened parentheses actual
2391 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2393 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2396 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2400 if (c == '\\') skip = 1;
2401 else if (c == '{') {
2404 else if (c == '}') {
2405 handleCloseP(i, closingAllowed);
2411 string Intervall::show(int lastpos)
2413 int idx = 0; /* int intervalls */
2416 for (idx = 0; idx <= ignoreidx; idx++) {
2417 while (i < lastpos) {
2419 if (i <= borders[idx].low) {
2420 if (borders[idx].low > lastpos)
2421 printsize = lastpos - i;
2423 printsize = borders[idx].low - i;
2424 s += par.substr(i, printsize);
2426 if (i >= borders[idx].low)
2427 i = borders[idx].upper;
2430 i = borders[idx].upper;
2436 s += par.substr(i, lastpos-i);
2442 void Intervall::output(ostringstream &os, int lastpos)
2444 // get number of chars to output
2445 int idx = 0; /* int intervalls */
2448 string startTitle = titleValue;
2449 for (idx = 0; idx <= ignoreidx; idx++) {
2451 if (i <= borders[idx].low) {
2453 if (borders[idx].low > lastpos)
2454 printsize = lastpos - i;
2456 printsize = borders[idx].low - i;
2457 if (printsize > 0) {
2458 os << startTitle << par.substr(i, printsize);
2460 printed += printsize;
2463 handleParentheses(i, false);
2464 if (i >= borders[idx].low)
2465 i = borders[idx].upper;
2468 i = borders[idx].upper;
2475 os << startTitle << par.substr(i, lastpos-i);
2476 printed += lastpos-i;
2478 handleParentheses(lastpos, false);
2480 if (keys["foreignlanguage"].disabled)
2481 startindex = actualdeptindex-langcount;
2483 startindex = actualdeptindex;
2484 for (int i = startindex; i > 0; --i) {
2487 if (hasTitle && (printed > 0))
2489 if (! isPatternString_)
2491 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2494 void LatexInfo::processRegion(int start, int region_end)
2496 while (start < region_end) { /* Let {[} and {]} survive */
2497 int cnt = interval_.isOpeningPar(start);
2499 // Closing is allowed past the region
2500 int closing = interval_.findclosing(start+1, interval_.par.length());
2501 interval_.addIntervall(start, start+1);
2502 interval_.addIntervall(closing, closing+1);
2506 start = interval_.nextNotIgnored(start+1);
2510 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2512 if (actual.parenthesiscount == 0) {
2513 // "{\tiny{} ...}" ==> "{{} ...}"
2514 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2517 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2518 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2519 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2523 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2526 switch (actual.keytype)
2528 case KeyInfo::isTitle: {
2530 nextKeyIdx = getNextKey();
2533 case KeyInfo::cleanToStart: {
2534 actual._dataEnd = actual._dataStart;
2535 nextKeyIdx = getNextKey();
2536 // Search for end of arguments
2537 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2539 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2540 entries_[i].disabled = true;
2542 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2544 while (interval_.par[actual._dataEnd] == ' ')
2546 interval_.addIntervall(0, actual._dataEnd+1);
2547 interval_.actualdeptindex = 0;
2548 interval_.depts[0] = actual._dataEnd+1;
2549 interval_.closes[0] = -1;
2552 case KeyInfo::isText:
2553 interval_.par[actual._tokenstart] = '#';
2554 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2555 nextKeyIdx = getNextKey();
2557 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2558 if (actual.disabled)
2559 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2561 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2564 case KeyInfo::isChar: {
2565 nextKeyIdx = getNextKey();
2568 case KeyInfo::isSize: {
2569 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2570 if (actual.parenthesiscount == 0)
2571 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2573 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2575 nextKeyIdx = getNextKey();
2577 // Here _dataStart points to '{', so correct it
2578 actual._dataStart += 1;
2579 actual._tokensize += 1;
2580 actual.parenthesiscount = 1;
2581 if (interval_.par[actual._dataStart] == '}') {
2582 // Determine the end if used like '{\tiny{}...}'
2583 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2584 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2587 // Determine the end if used like '\tiny{...}'
2588 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2590 // Split on this key if not at start
2591 int start = interval_.nextNotIgnored(previousStart);
2592 if (start < actual._tokenstart) {
2593 interval_.output(os, actual._tokenstart);
2594 interval_.addIntervall(start, actual._tokenstart);
2596 // discard entry if at end of actual
2597 nextKeyIdx = process(os, actual);
2601 case KeyInfo::endArguments: {
2602 // Remove trailing '{}' too
2603 actual._dataStart += 1;
2604 actual._dataEnd += 1;
2605 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2606 nextKeyIdx = getNextKey();
2609 case KeyInfo::noMain:
2611 case KeyInfo::isStandard: {
2612 if (actual.disabled) {
2614 processRegion(actual._dataStart, actual._dataStart+1);
2615 nextKeyIdx = getNextKey();
2617 // Split on this key if not at datastart of calling entry
2618 int start = interval_.nextNotIgnored(previousStart);
2619 if (start < actual._tokenstart) {
2620 interval_.output(os, actual._tokenstart);
2621 interval_.addIntervall(start, actual._tokenstart);
2623 // discard entry if at end of actual
2624 nextKeyIdx = process(os, actual);
2628 case KeyInfo::removeWithArg: {
2629 nextKeyIdx = getNextKey();
2630 // Search for end of arguments
2631 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2633 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2634 entries_[i].disabled = true;
2636 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2638 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2641 case KeyInfo::doRemove: {
2642 // Remove the key with all parameters and following spaces
2645 if (interval_.par[actual._dataEnd-1] == ' ')
2646 start = actual._dataEnd;
2648 start = actual._dataEnd+1;
2649 for (pos = start; pos < interval_.par.length(); pos++) {
2650 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2653 // Remove also enclosing parentheses [] and {}
2656 while (actual._tokenstart > numpars) {
2657 if (pos+numpars >= interval_.par.size())
2659 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2661 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2667 if (interval_.par[pos+numpars] == ' ')
2671 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2672 nextKeyIdx = getNextKey();
2675 case KeyInfo::isList: {
2676 // Discard space before _tokenstart
2678 for (count = 0; count < actual._tokenstart; count++) {
2679 if (interval_.par[actual._tokenstart-count-1] != ' ')
2682 nextKeyIdx = getNextKey();
2683 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2685 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2687 // How else can we catch this one?
2688 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2689 entries_[i].disabled = true;
2691 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2693 else if (nextKeyIdx > 0) {
2694 // Ignore any lang entries inside data region
2695 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2696 if (entries_[i].keytype == KeyInfo::isMain)
2697 entries_[i].disabled = true;
2700 if (actual.disabled) {
2701 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2704 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2706 if (interval_.par[actual._dataEnd+1] == '[') {
2707 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2708 if ((interval_.par[actual._dataEnd+2] == '{') &&
2709 (interval_.par[posdown-1] == '}')) {
2710 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2711 interval_.addIntervall(posdown-1, posdown+1);
2714 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2715 interval_.addIntervall(posdown, posdown+1);
2717 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2718 if (blk > posdown) {
2719 // Discard at most 1 space after empty item
2721 for (count = 0; count < 1; count++) {
2722 if (interval_.par[blk+count] != ' ')
2726 interval_.addIntervall(blk, blk+count);
2731 case KeyInfo::isSectioning: {
2732 // Discard spaces before _tokenstart
2734 int val = actual._tokenstart;
2735 for (count = 0; count < actual._tokenstart;) {
2736 val = interval_.previousNotIgnored(val-1);
2737 if (val < 0 || interval_.par[val] != ' ')
2740 count = actual._tokenstart - val;
2743 if (actual.disabled) {
2744 removeHead(actual, count);
2745 nextKeyIdx = getNextKey();
2747 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2748 nextKeyIdx = process(os, actual);
2752 case KeyInfo::isMath: {
2753 // Same as regex, use the content unchanged
2754 nextKeyIdx = getNextKey();
2757 case KeyInfo::isRegex: {
2758 // DO NOT SPLIT ON REGEX
2760 nextKeyIdx = getNextKey();
2763 case KeyInfo::isIgnored: {
2764 // Treat like a character for now
2765 nextKeyIdx = getNextKey();
2768 case KeyInfo::isMain: {
2769 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2770 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2771 if (actual._tokenstart > 0) {
2772 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2773 if ((prev >= 0) && interval_.par[prev] == '%')
2774 interval_.addIntervall(prev, prev+1);
2776 if (actual.disabled) {
2778 interval_.langcount++;
2779 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2780 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2781 // Discard also the space before math-equation
2782 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2784 nextKeyIdx = getNextKey();
2785 // interval.resetOpenedP(actual._dataStart-1);
2788 if (actual._tokenstart < 26) {
2789 // for the first (and maybe dummy) language
2790 interval_.setForDefaultLang(actual);
2792 interval_.resetOpenedP(actual._dataStart-1);
2796 case KeyInfo::invalid:
2797 case KeyInfo::headRemove:
2798 // These two cases cannot happen, already handled
2801 // LYXERR(Debug::INFO, "Unhandled keytype");
2802 nextKeyIdx = getNextKey();
2809 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2811 int end = interval_.nextNotIgnored(actual._dataEnd);
2812 int oldStart = actual._dataStart;
2813 int nextKeyIdx = getNextKey();
2815 if ((nextKeyIdx < 0) ||
2816 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2817 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2818 if (oldStart <= end) {
2819 processRegion(oldStart, end);
2824 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2826 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2827 (void) dispatch(os, actual._dataStart, nextKey);
2828 end = nextKey._tokenstart;
2831 processRegion(oldStart, nextKey._tokenstart);
2832 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2834 oldStart = nextKey._dataEnd+1;
2836 // now nextKey is either invalid or is outside of actual._dataEnd
2837 // output the remaining and discard myself
2838 if (oldStart <= end) {
2839 processRegion(oldStart, end);
2841 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2843 // This is the normal case.
2844 // But if using the firstlanguage, the closing may be missing
2846 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2848 if (actual._dataEnd < end)
2849 output_end = interval_.nextNotIgnored(actual._dataEnd);
2850 else if (interval_.par.size() > (size_t) end)
2851 output_end = interval_.nextNotIgnored(end);
2853 output_end = interval_.par.size();
2854 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2855 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2857 // Remove possible empty data
2858 int dstart = interval_.nextNotIgnored(actual._dataStart);
2859 while (interval_.isOpeningPar(dstart) == 1) {
2860 interval_.addIntervall(dstart, dstart+1);
2861 int dend = interval_.findclosing(dstart+1, output_end);
2862 interval_.addIntervall(dend, dend+1);
2863 dstart = interval_.nextNotIgnored(dstart+1);
2865 if (dstart < output_end)
2866 interval_.output(os, output_end);
2868 interval_.addIntervall(0, end);
2870 interval_.addIntervall(actual._tokenstart, end);
2874 string splitOnKnownMacros(string par, bool isPatternString)
2877 LatexInfo li(par, isPatternString);
2878 // LYXERR(Debug::INFO, "Berfore split: " << par);
2879 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2881 DummyKey._tokensize = 0;
2882 DummyKey._dataStart = 0;
2883 DummyKey._dataEnd = par.length();
2884 DummyKey.disabled = true;
2885 int firstkeyIdx = li.getFirstKey();
2887 if (firstkeyIdx >= 0) {
2888 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2889 DummyKey._tokenstart = firstKey._tokenstart;
2891 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2892 // Use dummy firstKey
2893 firstKey = DummyKey;
2894 (void) li.setNextKey(firstkeyIdx);
2897 if (par.substr(firstKey._dataStart, 2) == "% ")
2898 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2900 nextkeyIdx = li.process(os, firstKey);
2901 while (nextkeyIdx >= 0) {
2902 // Check for a possible gap between the last
2903 // entry and this one
2904 int datastart = li.nextNotIgnored(firstKey._dataStart);
2905 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2906 if ((nextKey._tokenstart > datastart)) {
2908 firstKey._dataStart = datastart;
2909 firstKey._dataEnd = par.length();
2910 (void) li.setNextKey(nextkeyIdx);
2911 // Fake the last opened parenthesis
2912 li.setForDefaultLang(firstKey);
2913 nextkeyIdx = li.process(os, firstKey);
2916 if (nextKey.keytype != KeyInfo::isMain) {
2917 firstKey._dataStart = datastart;
2918 firstKey._dataEnd = nextKey._dataEnd+1;
2919 (void) li.setNextKey(nextkeyIdx);
2920 li.setForDefaultLang(firstKey);
2921 nextkeyIdx = li.process(os, firstKey);
2924 nextkeyIdx = li.process(os, nextKey);
2928 // Handle the remaining
2929 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2930 firstKey._dataEnd = par.length();
2932 if ((firstKey._dataStart < firstKey._dataEnd) &&
2933 (par[firstKey._dataStart] != '}')) {
2934 li.setForDefaultLang(firstKey);
2935 (void) li.process(os, firstKey);
2939 // return string definitelly impossible to match
2940 s = "\\foreignlanguage{ignore}{ }";
2944 s = par; /* no known macros found */
2945 // LYXERR(Debug::INFO, "After split: " << s);
2950 * Try to unify the language specs in the latexified text.
2951 * Resulting modified string is set to "", if
2952 * the searched tex does not contain all the features in the search pattern
2954 static string correctlanguagesetting(string par, bool isPatternString, bool withformat, lyx::Buffer *pbuf = nullptr)
2956 static Features regex_f;
2957 static int missed = 0;
2958 static bool regex_with_format = false;
2960 int parlen = par.length();
2962 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2965 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2966 // Happens to be there in case of description or labeling environment
2971 // Split the latex input into pieces which
2972 // can be digested by our search engine
2973 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2974 if (isPatternString && (pbuf != nullptr)) { // Check if we should disable/enable test for language
2975 // We check for polyglossia, because in runparams.flavor we use Flavor::XeTeX
2976 string doclang = pbuf->params().language->polyglossia();
2977 static regex langre("\\\\(foreignlanguage)\\{([^\\}]+)\\}");
2979 bool toIgnoreLang = true;
2980 for (sregex_iterator it(par.begin(), par.end(), langre), end; it != end; ++it) {
2982 if (sub.str(2) != doclang) {
2983 toIgnoreLang = false;
2987 setIgnoreFormat("language", toIgnoreLang, false);
2990 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2991 LYXERR(Debug::FIND, "After splitOnKnownMacros:\n\"" << result << "\"");
2994 result = par.substr(0, parlen);
2995 if (isPatternString) {
2998 regex_f = identifyFeatures(result);
2999 string features = "";
3000 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3001 string a = it->first;
3002 regex_with_format = true;
3003 features += " " + a;
3004 // LYXERR(Debug::INFO, "Identified regex format:" << a);
3006 LYXERR(Debug::FIND, "Identified Features" << features);
3009 } else if (regex_with_format) {
3010 Features info = identifyFeatures(result);
3011 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
3012 string a = it->first;
3013 bool b = it->second;
3014 if (b && ! info[a]) {
3016 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
3023 // LYXERR(Debug::INFO, "No regex formats");
3029 // Remove trailing closure of math, macros and environments, so to catch parts of them.
3030 static int identifyClosing(string & t)
3032 int open_braces = 0;
3034 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
3035 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
3037 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
3039 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
3041 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
3050 static int num_replaced = 0;
3051 static bool previous_single_replace = true;
3053 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
3056 // Handle \w properly
3057 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3058 if (! opt.casesensitive) {
3059 popts |= QRegularExpression::CaseInsensitiveOption;
3061 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3062 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3064 if (regexp.isValid() && regexp2.isValid()) {
3065 regexIsValid = true;
3066 // Check '{', '}' pairs inside the regex
3069 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3070 char c = par_as_string[i];
3077 else if (c == '}') {
3084 if (balanced != 0) {
3085 regexIsValid = false;
3086 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3090 regexIsValid = false;
3091 if (!regexp.isValid())
3092 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3094 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3097 if (opt.casesensitive) {
3098 regexp = regex(regexp_str);
3099 regexp2 = regex(regexp2_str);
3102 regexp = regex(regexp_str, std::regex_constants::icase);
3103 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3108 static void modifyRegexForMatchWord(string &t)
3111 regex wordre("(\\\\)*((\\.|\\\\b))");
3114 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
3116 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
3119 else if (sub.str(2) == "\\\\b")
3121 if (lastpos < (size_t) sub.position(2))
3122 s += t.substr(lastpos, sub.position(2) - lastpos);
3124 lastpos = sub.position(2) + sub.length(2);
3127 s = "\\b" + t + "\\b";
3131 else if (lastpos < t.length())
3132 s += t.substr(lastpos, t.length() - lastpos);
3133 t = "\\b" + s + "\\b";
3136 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
3137 : p_buf(&buf), p_first_buf(&buf), opt(opt)
3139 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
3140 docstring const & ds = stringifySearchBuffer(find_buf, opt);
3141 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
3142 if (opt.replace_all && previous_single_replace) {
3143 previous_single_replace = false;
3146 else if (!opt.replace_all) {
3147 num_replaced = 0; // count number of replaced strings
3148 previous_single_replace = true;
3150 // When using regexp, braces are hacked already by escape_for_regex()
3151 par_as_string = normalize(ds);
3153 close_wildcards = 0;
3155 size_t lead_size = 0;
3156 // correct the language settings
3157 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat, &buf);
3158 opt.matchAtStart = false;
3160 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
3161 if (opt.ignoreformat) {
3165 lead_size = identifyLeading(par_as_string);
3167 lead_as_string = par_as_string.substr(0, lead_size);
3168 string lead_as_regex_string = string2regex(lead_as_string);
3169 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3170 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
3171 /* Handle whole words too in this case
3173 if (opt.matchword) {
3174 par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
3175 opt.matchword = false;
3177 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
3178 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
3179 CreateRegexp(opt, regexp_str, regexp2_str);
3181 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3182 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3186 if (!opt.ignoreformat) {
3187 lead_size = identifyLeading(par_as_string);
3188 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
3189 lead_as_string = par_as_string.substr(0, lead_size);
3190 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3193 // Here we are using regexp
3194 LASSERT(use_regexp, /**/);
3196 string lead_as_regexp;
3197 if (lead_size > 0) {
3198 lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3199 regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3200 par_as_string = par_as_string_nolead;
3201 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
3202 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3204 // LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3205 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3206 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3207 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3209 size_t lng = par_as_string.size();
3210 if (!opt.ignoreformat) {
3211 // Remove extra '\}' at end if not part of \{\.\}
3213 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3215 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3224 if (lng < par_as_string.size())
3225 par_as_string = par_as_string.substr(0,lng);
3227 LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
3228 if ((lng > 0) && (par_as_string[0] == '^')) {
3229 par_as_string = par_as_string.substr(1);
3231 opt.matchAtStart = true;
3233 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3234 // LYXERR(Debug::FIND, "Open braces: " << open_braces);
3235 // LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3237 // If entered regexp must match at begin of searched string buffer
3238 // Kornel: Added parentheses to use $1 for size of the leading string
3242 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3243 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3244 // so the convert has no effect in that case
3245 for (int i = 7; i > 0; --i) {
3246 string orig = "\\\\" + std::to_string(i);
3247 string dest = "\\" + std::to_string(i+2);
3248 while (regex_replace(par_as_string, par_as_string, orig, dest));
3250 if (opt.matchword) {
3251 modifyRegexForMatchWord(par_as_string);
3252 opt.matchword = false;
3254 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3255 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3257 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3258 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3259 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3263 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3267 mres.searched_size = len;
3269 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3272 docstring docstr = stringifyFromForSearch(opt, cur, len);
3274 str = normalize(docstr);
3275 if (!opt.ignoreformat) {
3276 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3277 // remove closing '}' and '\n' to allow for use of '$' in regex
3278 size_t lng = str.size();
3279 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3281 if (lng != str.size()) {
3282 str = str.substr(0, lng);
3286 mres.match_len = -1;
3289 LYXERR(Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
3291 LASSERT(use_regexp, /**/);
3293 // use_regexp always true
3294 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3296 QString qstr = QString::fromStdString(str);
3297 QRegularExpression const *p_regexp;
3298 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3302 p_regexp = ®exp2;
3304 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3305 if (!match.hasMatch())
3308 regex const *p_regexp;
3309 regex_constants::match_flag_type flags;
3311 flags = regex_constants::match_continuous;
3314 flags = regex_constants::match_default;
3315 p_regexp = ®exp2;
3317 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3318 if (re_it == sregex_iterator())
3320 match_results<string::const_iterator> const & m = *re_it;
3322 // Whole found string, including the leading
3323 // std: m[0].second - m[0].first
3324 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3326 // Size of the leading string
3327 // std: m[1].second - m[1].first
3328 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3329 int leadingsize = 0;
3331 if (match.lastCapturedIndex() > 0) {
3332 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3337 leadingsize = m[1].second - m[1].first;
3341 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3342 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3343 // because of different number of closing at end of string
3344 // we have to 'unify' the length of the post-match.
3345 // Done by ignoring closing parenthesis and linefeeds at string end
3346 int matchend = match.capturedEnd(0);
3347 size_t strsize = qstr.size();
3348 if (!opt.ignoreformat) {
3349 while (mres.match_len > 0) {
3350 QChar c = qstr.at(matchend - 1);
3351 if ((c == '\n') || (c == '}') || (c == '{')) {
3358 while (strsize > (size_t) match.capturedEnd(0)) {
3359 QChar c = qstr.at(strsize-1);
3360 if ((c == '\n') || (c == '}')) {
3367 // LYXERR0(qstr.toStdString());
3368 mres.match2end = strsize - matchend;
3369 mres.pos = match.capturedStart(2);
3371 mres.match_prefix = m[2].second - m[2].first;
3372 mres.match_len = m[0].second - m[2].second;
3373 // ignore closing parenthesis and linefeeds at string end
3374 size_t strend = m[0].second - m[0].first;
3375 int matchend = strend;
3376 size_t strsize = str.size();
3377 if (!opt.ignoreformat) {
3378 while (mres.match_len > 0) {
3379 char c = str.at(matchend - 1);
3380 if ((c == '\n') || (c == '}') || (c == '{')) {
3387 while (strsize > strend) {
3388 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3396 mres.match2end = strsize - matchend;
3397 mres.pos = m[2].first - m[0].first;;
3399 if (mres.match2end < 0)
3401 mres.leadsize = leadingsize;
3403 if (mres.match_len > 0) {
3404 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3405 mres.result.push_back(a0);
3406 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3407 mres.result.push_back(match.captured(i).toStdString());
3411 if (mres.match_len > 0) {
3412 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3413 mres.result.push_back(a0);
3414 for (size_t i = 3; i < m.size(); i++) {
3415 mres.result.push_back(m[i]);
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 << ", matchAtStart=" << opt.matchAtStart
3431 << ", inTexted=" << cur.inTexted());
3432 if (opt.matchAtStart) {
3435 else if (mres.match_prefix > 0)
3444 static bool simple_replace(string &t, string from, string to)
3446 regex repl("(\\\\)*(" + from + ")");
3450 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3452 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3454 if (lastpos < (size_t) sub.position(2))
3455 s += t.substr(lastpos, sub.position(2) - lastpos);
3457 lastpos = sub.position(2) + sub.length(2);
3461 else if (lastpos < t.length())
3462 s += t.substr(lastpos, t.length() - lastpos);
3468 string MatchStringAdv::normalize(docstring const & s) const
3471 t = lyx::to_utf8(s);
3472 // Remove \n at begin
3473 while (!t.empty() && t[0] == '\n')
3476 while (!t.empty() && t[t.size() - 1] == '\n')
3477 t = t.substr(0, t.size() - 1);
3479 // Handle all other '\n'
3480 while ((pos = t.find("\n")) != string::npos) {
3481 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3483 if (isAlnumASCII(t[pos+1])) {
3484 t.replace(pos-2, 3, " ");
3487 t.replace(pos-2, 3, "");
3490 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3491 // '\n' adjacent to non-alpha-numerics, discard
3492 t.replace(pos, 1, "");
3495 // Replace all other \n with spaces
3496 t.replace(pos, 1, " ");
3499 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3500 // Kornel: Added textsl, textsf, textit, texttt and noun
3501 // + allow to seach for colored text too
3502 LYXERR(Debug::FIND, "Removing stale empty macros from: " << t);
3503 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3504 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3505 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3506 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3507 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3513 docstring stringifyFromCursor(DocIterator const & cur, int len)
3515 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3516 if (cur.inTexted()) {
3517 Paragraph const & par = cur.paragraph();
3518 // TODO what about searching beyond/across paragraph breaks ?
3519 // TODO Try adding a AS_STR_INSERTS as last arg
3520 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3521 int(par.size()) : cur.pos() + len;
3522 // OutputParams runparams(&cur.buffer()->params().encoding());
3523 OutputParams runparams(encodings.fromLyXName("utf8"));
3524 runparams.nice = true;
3525 runparams.flavor = Flavor::XeTeX;
3526 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3527 // No side effect of file copying and image conversion
3528 runparams.dryrun = true;
3529 int option = AS_STR_INSETS | AS_STR_PLAINTEXT;
3530 if (ignoreFormats.getDeleted()) {
3531 option |= AS_STR_SKIPDELETE;
3532 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
3535 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
3537 LYXERR(Debug::FIND, "Stringifying with cur: "
3538 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3539 return par.asString(cur.pos(), end,
3542 } else if (cur.inMathed()) {
3543 CursorSlice cs = cur.top();
3544 MathData md = cs.cell();
3545 MathData::const_iterator it_end =
3546 (( len == -1 || cs.pos() + len > int(md.size()))
3548 : md.begin() + cs.pos() + len );
3550 for (MathData::const_iterator it = md.begin() + cs.pos();
3553 docstring s = asString(md2);
3554 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3557 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3562 /** Computes the LaTeX export of buf starting from cur and ending len positions
3563 * after cur, if len is positive, or at the paragraph or innermost inset end
3566 docstring latexifyFromCursor(DocIterator const & cur, int len)
3569 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3570 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3571 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3573 Buffer const & buf = *cur.buffer();
3575 odocstringstream ods;
3577 //OutputParams runparams(&buf.params().encoding());
3578 OutputParams runparams(encodings.fromLyXName("utf8"));
3579 runparams.nice = false;
3580 runparams.flavor = Flavor::XeTeX;
3581 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3582 // No side effect of file copying and image conversion
3583 runparams.dryrun = true;
3584 if (ignoreFormats.getDeleted()) {
3585 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
3588 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
3591 if (cur.inTexted()) {
3592 // @TODO what about searching beyond/across paragraph breaks ?
3593 pos_type endpos = cur.paragraph().size();
3594 if (len != -1 && endpos > cur.pos() + len)
3595 endpos = cur.pos() + len;
3596 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3597 string(), cur.pos(), endpos);
3598 string s = lyx::to_utf8(ods.str());
3599 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3600 return(lyx::from_utf8(s));
3601 } else if (cur.inMathed()) {
3602 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3603 for (int s = cur.depth() - 1; s >= 0; --s) {
3604 CursorSlice const & cs = cur[s];
3605 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3606 TeXMathStream ws(os);
3607 cs.asInsetMath()->asHullInset()->header_write(ws);
3612 CursorSlice const & cs = cur.top();
3613 MathData md = cs.cell();
3614 MathData::const_iterator it_end =
3615 ((len == -1 || cs.pos() + len > int(md.size()))
3617 : md.begin() + cs.pos() + len);
3619 for (MathData::const_iterator it = md.begin() + cs.pos();
3623 ods << asString(md2);
3624 // Retrieve the math environment type, and add '$' or '$]'
3625 // or others (\end{equation}) accordingly
3626 for (int s = cur.depth() - 1; s >= 0; --s) {
3627 CursorSlice const & cs2 = cur[s];
3628 InsetMath * inset = cs2.asInsetMath();
3629 if (inset && inset->asHullInset()) {
3630 TeXMathStream ws(os);
3631 inset->asHullInset()->footer_write(ws);
3635 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3637 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3642 #if defined(ResultsDebug)
3644 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
3646 LYXERR0( "from:\t\t\t" << from);
3648 if (mres.pos_len > 0) {
3650 status = "FINALSEARCH";
3653 if (mres.match_len > 0) {
3654 if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
3655 status = "Good Match";
3657 status = "Matched in";
3660 status = "MissedSearch";
3663 LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
3664 if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
3665 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
3666 if ((mres.pos > 0) || (mres.match_prefix > 0))
3667 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
3668 for (size_t i = 0; i < mres.result.size(); i++)
3669 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3671 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
3673 #define displayMres(s, txt, cur)
3676 /** Finalize an advanced find operation, advancing the cursor to the innermost
3677 ** position that matches, plus computing the length of the matching text to
3679 ** Return the cur.pos() difference between start and end of found match
3681 MatchResult findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
3683 // Search the foremost position that matches (avoids find of entire math
3684 // inset when match at start of it)
3685 DocIterator old_cur(cur.buffer());
3687 static MatchResult fail = MatchResult();
3688 MatchResult max_match;
3689 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3690 // Happens with e.g. hyperlinks
3691 // either one sees "http://www.bla.bla" or nothing
3692 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3693 // and although we search for only 3 chars, we find the whole hyperlink inset
3694 bool at_begin = (expected.match_prefix == 0);
3695 if (!match.opt.forward && match.opt.ignoreformat) {
3696 if (expected.pos > 0)
3699 LASSERT(at_begin, /**/);
3700 if (expected.match_len > 0 && at_begin) {
3701 // Search for deepest match
3703 max_match = expected;
3705 size_t d = cur.depth();
3709 if (cur.depth() < d)
3711 if (cur.depth() == d)
3714 while (cur && cur.depth() > lastd) {
3715 lastd = cur.depth();
3716 mres = match(cur, -1, at_begin);
3717 displayMres(mres, "Checking innermost", cur);
3718 if (mres.match_len > 0)
3723 if (mres.match_len < expected.match_len)
3731 // (expected.match_len <= 0)
3732 mres = match(cur); /* match valid only if not searching whole words */
3733 displayMres(mres, "Start with negative match", cur);
3736 if (max_match.match_len <= 0) return fail;
3737 LYXERR(Debug::FIND, "Ok");
3739 // Compute the match length
3741 if (cur.pos() + len > cur.lastpos())
3744 LASSERT(match.use_regexp, /**/);
3747 int maxl = cur.lastpos() - cur.pos();
3748 // Greedy behaviour while matching regexps
3749 while (maxl > minl) {
3751 mres2 = match(cur, len, at_begin);
3752 displayMres(mres2, "Finalize loop", cur);
3753 int actual_match_len = mres2.match_len;
3754 if (actual_match_len >= max_match.match_len) {
3755 // actual_match_len > max_match _can_ happen,
3756 // if the search area splits
3757 // some following word so that the regex
3758 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3760 // This means, the len value is too big
3761 actual_match_len = max_match.match_len;
3763 max_match.match_len = actual_match_len;
3765 if (maxl - minl < 4)
3766 len = (int)((maxl + minl)/2);
3768 len = (int)(minl + (maxl - minl + 3)/4);
3771 // (actual_match_len < max_match.match_len)
3773 len = (int)((maxl + minl)/2);
3778 // Search for real start of matched characters
3780 MatchResult actual_match;
3783 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3784 if (cur.depth() < old_cur.depth()) {
3786 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3789 if (cur.pos() != old_cur.pos()) {
3790 // OK, forwarded 1 pos in actual inset
3791 actual_match = match(cur, len-1, at_begin);
3792 if (actual_match.match_len == max_match.match_len) {
3793 // Ha, got it! The shorter selection has the same match length
3796 max_match = actual_match;
3799 // OK, the shorter selection matches less chars, revert to previous value
3805 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3806 actual_match = match(cur, len, at_begin);
3807 if (actual_match.match_len == max_match.match_len) {
3809 max_match = actual_match;
3816 max_match.pos_len = len;
3817 displayMres(max_match, "SEARCH RESULT", cur)
3824 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3828 bool repeat = false;
3829 DocIterator orig_cur; // to be used if repeat not successful
3830 MatchResult orig_mres;
3831 while (!theApp()->longOperationCancelled() && cur) {
3832 //(void) findAdvForwardInnermost(cur);
3833 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3834 MatchResult mres = match(cur, -1, false);
3835 string msg = "Starting";
3838 displayMres(mres, msg + " findForwardAdv", cur)
3839 int match_len = mres.match_len;
3840 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3841 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3844 if (match_len <= 0) {
3845 // This should exit nested insets, if any, or otherwise undefine the currsor.
3846 cur.pos() = cur.lastpos();
3847 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3850 else { // match_len > 0
3851 // Try to find the begin of searched string
3853 int firstInvalid = cur.lastpos() - cur.pos();
3855 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3856 int incrcur = (firstInvalid + 1 )*3/4;
3857 if (incrcur < incrmatch)
3858 increment = incrcur;
3860 increment = incrmatch;
3864 LYXERR(Debug::FIND, "Set increment to " << increment);
3865 while (increment > 0) {
3866 DocIterator old_cur = cur;
3867 if (cur.pos() + increment >= cur.lastpos()) {
3871 cur.pos() = cur.pos() + increment;
3872 MatchResult mres2 = match(cur, -1, false);
3873 displayMres(mres2, "findForwardAdv loop", cur)
3874 switch (interpretMatch(mres, mres2)) {
3875 case MatchResult::newIsTooFar:
3876 // behind the expected match
3877 firstInvalid = increment;
3881 case MatchResult::newIsBetter:
3882 // not reached yet, but cur.pos()+increment is bettert
3884 firstInvalid -= increment;
3885 if (increment > firstInvalid*3/4)
3886 increment = firstInvalid*3/4;
3887 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3888 if (increment >= mres2.match_prefix)
3889 increment = (mres2.match_prefix+1)*3/4;
3894 // Handle not like MatchResult::newIsTooFar
3895 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3897 increment = increment*3/4;
3902 if (mres.match_len > 0) {
3903 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
3904 // The match seems to indicate some deeper level
3913 // should never be reached.
3917 // LYXERR0("Leaving first loop");
3918 LYXERR(Debug::FIND, "Finalizing 1");
3919 MatchResult found_match = findAdvFinalize(cur, match, mres);
3920 if (found_match.match_len > 0) {
3921 LASSERT(found_match.pos_len > 0, /**/);
3922 match.FillResults(found_match);
3923 return found_match.pos_len;
3926 // try next possible match
3937 /// Find the most backward consecutive match within same paragraph while searching backwards.
3938 MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, MatchResult &expected)
3940 DocIterator cur_begin = cur;
3941 cur_begin.pos() = 0;
3942 DocIterator tmp_cur = cur;
3943 MatchResult mr = findAdvFinalize(tmp_cur, match, expected);
3944 Inset & inset = cur.inset();
3945 for (; cur != cur_begin; cur.backwardPos()) {
3946 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3947 DocIterator new_cur = cur;
3948 new_cur.backwardPos();
3949 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3951 MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
3952 if (new_mr.match_len == mr.match_len)
3956 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3962 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3966 // Backup of original position
3967 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3968 if (cur == cur_begin)
3971 DocIterator cur_orig(cur);
3972 bool pit_changed = false;
3975 MatchResult found_match = match(cur, -1, false);
3977 if (found_match.match_len > 0) {
3979 cur.pos() = cur.lastpos();
3981 cur.pos() = cur_orig.pos();
3982 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3983 DocIterator cur_prev_iter;
3985 found_match = match(cur);
3986 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3987 << (found_match.match_len > 0) << ", cur: " << cur);
3988 if (found_match.match_len > 0) {
3989 MatchResult found_mr = findMostBackwards(cur, match, found_match);
3990 if (found_mr.pos_len > 0) {
3991 match.FillResults(found_mr);
3992 return found_mr.pos_len;
3996 // Stop if begin of document reached
3997 if (cur == cur_begin)
3999 cur_prev_iter = cur;
4003 if (cur == cur_begin)
4010 } while (!theApp()->longOperationCancelled());
4018 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
4019 DocIterator const & cur, int len)
4021 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
4023 if (!opt.ignoreformat)
4024 return latexifyFromCursor(cur, len);
4026 return stringifyFromCursor(cur, len);
4030 FindAndReplaceOptions::FindAndReplaceOptions(
4031 docstring const & _find_buf_name, bool _casesensitive,
4032 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
4033 docstring const & _repl_buf_name, bool _keep_case,
4034 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
4035 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
4036 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
4037 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
4045 /** Check if 'len' letters following cursor are all non-lowercase */
4046 static bool allNonLowercase(Cursor const & cur, int len)
4048 pos_type beg_pos = cur.selectionBegin().pos();
4049 pos_type end_pos = cur.selectionBegin().pos() + len;
4050 if (len > cur.lastpos() + 1 - beg_pos) {
4051 LYXERR(Debug::FIND, "This should not happen, more debug needed");
4052 len = cur.lastpos() + 1 - beg_pos;
4053 end_pos = beg_pos + len;
4055 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
4056 if (isLowerCase(cur.paragraph().getChar(pos)))
4062 /** Check if first letter is upper case and second one is lower case */
4063 static bool firstUppercase(Cursor const & cur)
4066 pos_type pos = cur.selectionBegin().pos();
4067 if (pos >= cur.lastpos() - 1) {
4068 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
4071 ch1 = cur.paragraph().getChar(pos);
4072 ch2 = cur.paragraph().getChar(pos + 1);
4073 bool result = isUpperCase(ch1) && isLowerCase(ch2);
4074 LYXERR(Debug::FIND, "firstUppercase(): "
4075 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4076 << ch2 << "(" << char(ch2) << ")"
4077 << ", result=" << result << ", cur=" << cur);
4082 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4084 ** \fixme What to do with possible further paragraphs in replace buffer ?
4086 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4088 ParagraphList::iterator pit = buffer.paragraphs().begin();
4089 LASSERT(!pit->empty(), /**/);
4090 pos_type right = pos_type(1);
4091 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4092 right = pit->size();
4093 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4097 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
4099 // Should replace the string "$" + std::to_string(matchnum) with replacement
4100 // if the char '$' is not prefixed with odd number of char '\\'
4101 static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
4105 for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
4107 if ((sub.position(2) - sub.position(0)) % 2 == 1)
4109 int num = stoi(sub.str(3), nullptr, 10);
4110 if (num >= maxmatchnum)
4112 if (lastpos < (size_t) sub.position(2))
4113 s += t.substr(lastpos, sub.position(2) - lastpos);
4114 s += replacements[num];
4115 lastpos = sub.position(2) + sub.length(2);
4119 else if (lastpos < t.length())
4120 s += t.substr(lastpos, t.length() - lastpos);
4126 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4128 Cursor & cur = bv->cursor();
4129 if (opt.repl_buf_name.empty()
4130 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4131 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4134 DocIterator sel_beg = cur.selectionBegin();
4135 DocIterator sel_end = cur.selectionEnd();
4136 if (&sel_beg.inset() != &sel_end.inset()
4137 || sel_beg.pit() != sel_end.pit()
4138 || sel_beg.idx() != sel_end.idx())
4140 int sel_len = sel_end.pos() - sel_beg.pos();
4141 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4142 << ", sel_len: " << sel_len << endl);
4145 LASSERT(sel_len > 0, return 0);
4147 if (!matchAdv(sel_beg, sel_len).match_len)
4150 // Build a copy of the replace buffer, adapted to the KeepCase option
4151 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4153 repl_buffer_orig.write(oss);
4154 string lyx = oss.str();
4155 if (matchAdv.valid_matches > 0) {
4156 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
4158 Buffer repl_buffer("", false);
4159 repl_buffer.setUnnamed(true);
4160 LASSERT(repl_buffer.readString(lyx), return 0);
4161 if (opt.keep_case && sel_len >= 2) {
4162 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4163 if (cur.inTexted()) {
4164 if (firstUppercase(cur))
4165 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4166 else if (allNonLowercase(cur, sel_len))
4167 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4170 cap::cutSelection(cur, false);
4171 if (cur.inTexted()) {
4172 repl_buffer.changeLanguage(
4173 repl_buffer.language(),
4174 cur.getFont().language());
4175 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4176 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4177 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4178 repl_buffer.params().documentClassPtr(),
4179 repl_buffer.params().authors(),
4180 bv->buffer().errorList("Paste"));
4181 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4182 sel_len = repl_buffer.paragraphs().begin()->size();
4183 } else if (cur.inMathed()) {
4184 odocstringstream ods;
4186 // OutputParams runparams(&repl_buffer.params().encoding());
4187 OutputParams runparams(encodings.fromLyXName("utf8"));
4188 runparams.nice = false;
4189 runparams.flavor = Flavor::XeTeX;
4190 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4191 runparams.dryrun = true;
4192 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4193 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4194 docstring repl_latex = ods.str();
4195 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4197 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4198 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4199 repl_latex = from_utf8(s);
4200 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4201 MathData ar(cur.buffer());
4202 asArray(repl_latex, ar, Parse::NORMAL);
4204 sel_len = ar.size();
4205 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4207 if (cur.pos() >= sel_len)
4208 cur.pos() -= sel_len;
4211 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4212 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4213 bv->processUpdateFlags(Update::Force);
4218 /// Perform a FindAdv operation.
4219 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4224 // e.g., when invoking word-findadv from mini-buffer wither with
4225 // wrong options syntax or before ever opening advanced F&R pane
4226 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4230 MatchStringAdv matchAdv(bv->buffer(), opt);
4232 if (!matchAdv.regexIsValid) {
4233 bv->message(lyx::from_utf8(matchAdv.regexError));
4237 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4239 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4240 num_replaced += findAdvReplace(bv, opt, matchAdv);
4243 pos_len = findForwardAdv(cur, matchAdv);
4245 pos_len = findBackwardsAdv(cur, matchAdv);
4246 } catch (exception & ex) {
4247 bv->message(from_utf8(ex.what()));
4252 if (num_replaced > 0) {
4253 switch (num_replaced)
4256 bv->message(_("One match has been replaced."));
4259 bv->message(_("Two matches have been replaced."));
4262 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4268 bv->message(_("Match not found."));
4273 if (num_replaced > 0)
4274 bv->message(_("Match has been replaced."));
4276 bv->message(_("Match found."));
4278 if (cur.pos() + pos_len > cur.lastpos()) {
4279 // Prevent crash in bv->putSelectionAt()
4280 // Should never happen, maybe LASSERT() here?
4281 pos_len = cur.lastpos() - cur.pos();
4283 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4284 bv->putSelectionAt(cur, pos_len, !opt.forward);
4290 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4292 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4293 << opt.casesensitive << ' '
4294 << opt.matchword << ' '
4295 << opt.forward << ' '
4296 << opt.expandmacros << ' '
4297 << opt.ignoreformat << ' '
4298 << opt.replace_all << ' '
4299 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4300 << opt.keep_case << ' '
4301 << int(opt.scope) << ' '
4304 LYXERR(Debug::FIND, "built: " << os.str());
4310 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4312 // LYXERR(Debug::FIND, "parsing");
4316 while (line != "EOSS") {
4320 if (is.eof()) // Tolerate malformed request
4324 // LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4325 opt.find_buf_name = from_utf8(s);
4326 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4327 is.get(); // Waste space before replace string
4330 while (line != "EOSS") {
4334 if (is.eof()) // Tolerate malformed request
4338 // LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4339 opt.repl_buf_name = from_utf8(s);
4340 is >> opt.keep_case;
4343 opt.scope = FindAndReplaceOptions::SearchScope(i);
4345 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4348 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4349 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4350 << opt.scope << ' ' << opt.restr);