3 * This file is part of LyX, the document processor.
4 * License details can be found in the file COPYING.
6 * \author Lars Gullik Bjønnes
9 * \author Alfredo Braunstein
10 * \author Tommaso Cucinotta
11 * \author Kornel Benko
13 * Full author contact details are available in file CREDITS.
21 #include "BufferList.h"
22 #include "BufferParams.h"
23 #include "BufferView.h"
26 #include "CutAndPaste.h"
27 #include "FuncRequest.h"
29 #include "output_latex.h"
30 #include "OutputParams.h"
31 #include "Paragraph.h"
35 #include "frontends/Application.h"
36 #include "frontends/alert.h"
38 #include "mathed/InsetMath.h"
39 #include "mathed/InsetMathHull.h"
40 #include "mathed/MathData.h"
41 #include "mathed/MathStream.h"
42 #include "mathed/MathSupport.h"
44 #include "support/debug.h"
45 #include "support/docstream.h"
46 #include "support/FileName.h"
47 #include "support/gettext.h"
48 #include "support/lassert.h"
49 #include "support/lstrings.h"
50 #include "support/textutils.h"
55 //#define ResultsDebug
56 #define USE_QT_FOR_SEARCH
57 #if defined(USE_QT_FOR_SEARCH)
58 #include <QtCore> // sets QT_VERSION
59 #if (QT_VERSION >= 0x050000)
60 #include <QRegularExpression>
70 using namespace lyx::support;
75 // Helper class for deciding what should be ignored
79 IgnoreFormats() = default;
81 bool getFamily() const { return ignoreFamily_; }
83 bool getSeries() const { return ignoreSeries_; }
85 bool getShape() const { return ignoreShape_; }
87 bool getUnderline() const { return ignoreUnderline_; }
89 bool getMarkUp() const { return ignoreMarkUp_; }
91 bool getStrikeOut() const { return ignoreStrikeOut_; }
93 bool getSectioning() const { return ignoreSectioning_; }
95 bool getFrontMatter() const { return ignoreFrontMatter_; }
97 bool getColor() const { return ignoreColor_; }
99 bool getLanguage() const { return ignoreLanguage_; }
101 void setIgnoreFormat(string const & type, bool value);
105 bool ignoreFamily_ = false;
107 bool ignoreSeries_ = false;
109 bool ignoreShape_ = false;
111 bool ignoreUnderline_ = false;
113 bool ignoreMarkUp_ = false;
115 bool ignoreStrikeOut_ = false;
117 bool ignoreSectioning_ = false;
119 bool ignoreFrontMatter_ = false;
121 bool ignoreColor_ = false;
123 bool ignoreLanguage_ = false;
127 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
129 if (type == "color") {
130 ignoreColor_ = value;
132 else if (type == "language") {
133 ignoreLanguage_ = value;
135 else if (type == "sectioning") {
136 ignoreSectioning_ = value;
137 ignoreFrontMatter_ = value;
139 else if (type == "font") {
140 ignoreSeries_ = value;
141 ignoreShape_ = value;
142 ignoreFamily_ = value;
144 else if (type == "series") {
145 ignoreSeries_ = value;
147 else if (type == "shape") {
148 ignoreShape_ = value;
150 else if (type == "family") {
151 ignoreFamily_ = value;
153 else if (type == "markup") {
154 ignoreMarkUp_ = value;
156 else if (type == "underline") {
157 ignoreUnderline_ = value;
159 else if (type == "strike") {
160 ignoreStrikeOut_ = value;
164 // The global variable that can be changed from outside
165 IgnoreFormats ignoreFormats;
168 void setIgnoreFormat(string const & type, bool value)
170 ignoreFormats.setIgnoreFormat(type, value);
176 bool parse_bool(docstring & howto)
181 howto = split(howto, var, ' ');
189 MatchString(docstring const & s, bool cs, bool mw)
190 : str(s), case_sens(cs), whole_words(mw)
193 // returns true if the specified string is at the specified position
194 // del specifies whether deleted strings in ct mode will be considered
195 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
197 return par.find(str, case_sens, whole_words, pos, del);
205 // match whole words only
210 int findForward(DocIterator & cur, MatchString const & match,
211 bool find_del = true)
213 for (; cur; cur.forwardChar())
214 if (cur.inTexted()) {
215 int len = match(cur.paragraph(), cur.pos(), find_del);
223 int findBackwards(DocIterator & cur, MatchString const & match,
224 bool find_del = true)
228 if (cur.inTexted()) {
229 int len = match(cur.paragraph(), cur.pos(), find_del);
238 bool searchAllowed(docstring const & str)
241 frontend::Alert::error(_("Search error"), _("Search string is empty"));
248 bool findOne(BufferView * bv, docstring const & searchstr,
249 bool case_sens, bool whole, bool forward,
250 bool find_del = true, bool check_wrap = false)
252 if (!searchAllowed(searchstr))
255 DocIterator cur = forward
256 ? bv->cursor().selectionEnd()
257 : bv->cursor().selectionBegin();
259 MatchString const match(searchstr, case_sens, whole);
261 int match_len = forward
262 ? findForward(cur, match, find_del)
263 : findBackwards(cur, match, find_del);
266 bv->putSelectionAt(cur, match_len, !forward);
267 else if (check_wrap) {
268 DocIterator cur_orig(bv->cursor());
271 q = _("End of file reached while searching forward.\n"
272 "Continue searching from the beginning?");
274 q = _("Beginning of file reached while searching backward.\n"
275 "Continue searching from the end?");
276 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
277 q, 0, 1, _("&Yes"), _("&No"));
278 if (wrap_answer == 0) {
280 bv->cursor().clear();
281 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
283 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
284 bv->cursor().backwardPos();
286 bv->clearSelection();
287 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
290 bv->cursor().setCursor(cur_orig);
294 return match_len > 0;
298 int replaceAll(BufferView * bv,
299 docstring const & searchstr, docstring const & replacestr,
300 bool case_sens, bool whole)
302 Buffer & buf = bv->buffer();
304 if (!searchAllowed(searchstr) || buf.isReadonly())
307 DocIterator cur_orig(bv->cursor());
309 MatchString const match(searchstr, case_sens, whole);
312 int const rsize = replacestr.size();
313 int const ssize = searchstr.size();
316 cur.setCursor(doc_iterator_begin(&buf));
317 int match_len = findForward(cur, match, false);
318 while (match_len > 0) {
319 // Backup current cursor position and font.
320 pos_type const pos = cur.pos();
321 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
323 int striked = ssize -
324 cur.paragraph().eraseChars(pos, pos + match_len,
325 buf.params().track_changes);
326 cur.paragraph().insert(pos, replacestr, font,
327 Change(buf.params().track_changes
329 : Change::UNCHANGED));
330 for (int i = 0; i < rsize + striked; ++i)
333 match_len = findForward(cur, match, false);
336 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
338 cur_orig.fixIfBroken();
339 bv->setCursor(cur_orig);
345 // the idea here is that we are going to replace the string that
346 // is selected IF it is the search string.
347 // if there is a selection, but it is not the search string, then
348 // we basically ignore it. (FIXME We ought to replace only within
350 // if there is no selection, then:
351 // (i) if some search string has been provided, then we find it.
352 // (think of how the dialog works when you hit "replace" the
354 // (ii) if no search string has been provided, then we treat the
355 // word the cursor is in as the search string. (why? i have no
356 // idea.) but this only works in text?
358 // returns the number of replacements made (one, if any) and
359 // whether anything at all was done.
360 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
361 docstring const & replacestr, bool case_sens,
362 bool whole, bool forward, bool findnext)
364 Cursor & cur = bv->cursor();
365 if (!cur.selection()) {
366 // no selection, non-empty search string: find it
367 if (!searchstr.empty()) {
368 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
369 return make_pair(found, 0);
371 // empty search string
374 return make_pair(false, 0);
375 // select current word and treat it as the search string.
376 // This causes a minor bug as undo will restore this selection,
377 // which the user did not create (#8986).
378 cur.innerText()->selectWord(cur, WHOLE_WORD);
379 searchstr = cur.selectionAsString(false);
382 // if we still don't have a search string, report the error
384 if (!searchAllowed(searchstr))
385 return make_pair(false, 0);
387 bool have_selection = cur.selection();
388 docstring const selected = cur.selectionAsString(false);
391 ? searchstr == selected
392 : compare_no_case(searchstr, selected) == 0;
394 // no selection or current selection is not search word:
395 // just find the search word
396 if (!have_selection || !match) {
397 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
398 return make_pair(found, 0);
401 // we're now actually ready to replace. if the buffer is
402 // read-only, we can't, though.
403 if (bv->buffer().isReadonly())
404 return make_pair(false, 0);
406 cap::replaceSelectionWithString(cur, replacestr);
408 cur.pos() += replacestr.length();
409 LASSERT(cur.pos() <= cur.lastpos(),
410 cur.pos() = cur.lastpos());
413 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
415 return make_pair(true, 1);
421 docstring const find2string(docstring const & search,
422 bool casesensitive, bool matchword, bool forward)
426 << int(casesensitive) << ' '
427 << int(matchword) << ' '
433 docstring const replace2string(docstring const & replace,
434 docstring const & search,
435 bool casesensitive, bool matchword,
436 bool all, bool forward, bool findnext)
439 ss << replace << '\n'
441 << int(casesensitive) << ' '
442 << int(matchword) << ' '
444 << int(forward) << ' '
450 bool lyxfind(BufferView * bv, FuncRequest const & ev)
452 if (!bv || ev.action() != LFUN_WORD_FIND)
455 //lyxerr << "find called, cmd: " << ev << endl;
457 // data is of the form
459 // <casesensitive> <matchword> <forward>"
461 docstring howto = split(ev.argument(), search, '\n');
463 bool casesensitive = parse_bool(howto);
464 bool matchword = parse_bool(howto);
465 bool forward = parse_bool(howto);
467 return findOne(bv, search, casesensitive, matchword, forward, true, true);
471 bool lyxreplace(BufferView * bv,
472 FuncRequest const & ev, bool has_deleted)
474 if (!bv || ev.action() != LFUN_WORD_REPLACE)
477 // data is of the form
480 // <casesensitive> <matchword> <all> <forward> <findnext>"
483 docstring howto = split(ev.argument(), rplc, '\n');
484 howto = split(howto, search, '\n');
486 bool casesensitive = parse_bool(howto);
487 bool matchword = parse_bool(howto);
488 bool all = parse_bool(howto);
489 bool forward = parse_bool(howto);
490 bool findnext = howto.empty() ? true : parse_bool(howto);
495 int replace_count = 0;
497 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
498 update = replace_count > 0;
501 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
503 replace_count = rv.second;
506 Buffer const & buf = bv->buffer();
508 // emit message signal.
509 buf.message(_("String not found."));
511 if (replace_count == 0) {
512 buf.message(_("String found."));
513 } else if (replace_count == 1) {
514 buf.message(_("String has been replaced."));
516 docstring const str =
517 bformat(_("%1$d strings have been replaced."), replace_count);
521 } else if (findnext) {
522 // if we have deleted characters, we do not replace at all, but
523 // rather search for the next occurence
524 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
527 bv->message(_("String not found."));
533 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
535 for (; cur; cur.forwardPos())
536 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
540 DocIterator cur_orig(bv->cursor());
541 docstring q = _("End of file reached while searching forward.\n"
542 "Continue searching from the beginning?");
543 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
544 q, 0, 1, _("&Yes"), _("&No"));
545 if (wrap_answer == 0) {
546 bv->cursor().clear();
547 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
548 bv->clearSelection();
549 cur.setCursor(bv->cursor().selectionBegin());
550 if (findNextChange(bv, cur, false))
553 bv->cursor().setCursor(cur_orig);
560 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
562 for (cur.backwardPos(); cur; cur.backwardPos()) {
563 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
568 DocIterator cur_orig(bv->cursor());
569 docstring q = _("Beginning of file reached while searching backward.\n"
570 "Continue searching from the end?");
571 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
572 q, 0, 1, _("&Yes"), _("&No"));
573 if (wrap_answer == 0) {
574 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
575 bv->cursor().backwardPos();
576 bv->clearSelection();
577 cur.setCursor(bv->cursor().selectionBegin());
578 if (findPreviousChange(bv, cur, false))
581 bv->cursor().setCursor(cur_orig);
588 bool selectChange(Cursor & cur, bool forward)
590 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
592 Change ch = cur.paragraph().lookupChange(cur.pos());
594 CursorSlice tip1 = cur.top();
595 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
596 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
597 if (!ch2.isSimilarTo(ch))
600 CursorSlice tip2 = cur.top();
601 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
603 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
604 if (!ch2.isSimilarTo(ch)) {
605 // take a step forward to correctly set the selection
613 cur.bv().mouseSetCursor(cur, false);
615 cur.bv().mouseSetCursor(cur, true);
623 bool findChange(BufferView * bv, bool forward)
626 cur.setCursor(forward ? bv->cursor().selectionEnd()
627 : bv->cursor().selectionBegin());
628 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
629 return selectChange(cur, forward);
634 bool findNextChange(BufferView * bv)
636 return findChange(bv, true);
640 bool findPreviousChange(BufferView * bv)
642 return findChange(bv, false);
649 typedef vector<pair<string, string> > Escapes;
651 /// A map of symbols and their escaped equivalent needed within a regex.
652 /// @note Beware of order
653 Escapes const & get_regexp_escapes()
655 typedef std::pair<std::string, std::string> P;
657 static Escapes escape_map;
658 if (escape_map.empty()) {
659 escape_map.push_back(P("$", "_x_$"));
660 escape_map.push_back(P("{", "_x_{"));
661 escape_map.push_back(P("}", "_x_}"));
662 escape_map.push_back(P("[", "_x_["));
663 escape_map.push_back(P("]", "_x_]"));
664 escape_map.push_back(P("(", "_x_("));
665 escape_map.push_back(P(")", "_x_)"));
666 escape_map.push_back(P("+", "_x_+"));
667 escape_map.push_back(P("*", "_x_*"));
668 escape_map.push_back(P(".", "_x_."));
669 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
670 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
671 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
672 escape_map.push_back(P("_x_", "\\"));
677 /// A map of lyx escaped strings and their unescaped equivalent.
678 Escapes const & get_lyx_unescapes()
680 typedef std::pair<std::string, std::string> P;
682 static Escapes escape_map;
683 if (escape_map.empty()) {
684 escape_map.push_back(P("\\%", "%"));
685 escape_map.push_back(P("\\{", "{"));
686 escape_map.push_back(P("\\}", "}"));
687 escape_map.push_back(P("\\mathcircumflex ", "^"));
688 escape_map.push_back(P("\\mathcircumflex", "^"));
689 escape_map.push_back(P("\\backslash ", "\\"));
690 escape_map.push_back(P("\\backslash", "\\"));
691 escape_map.push_back(P("\\sim ", "~"));
692 escape_map.push_back(P("\\sim", "~"));
697 /// A map of escapes turning a regexp matching text to one matching latex.
698 Escapes const & get_regexp_latex_escapes()
700 typedef std::pair<std::string, std::string> P;
702 static Escapes escape_map;
703 if (escape_map.empty()) {
704 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
705 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
706 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
707 escape_map.push_back(P("\\[", "\\{\\[\\}"));
708 escape_map.push_back(P("\\]", "\\{\\]\\}"));
709 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
710 escape_map.push_back(P("%", "\\\\\\%"));
711 escape_map.push_back(P("#", "\\\\#"));
716 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
717 ** the found occurrence were escaped.
719 string apply_escapes(string s, Escapes const & escape_map)
721 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
722 Escapes::const_iterator it;
723 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
724 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
725 unsigned int pos = 0;
726 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
727 s.replace(pos, it->first.length(), it->second);
728 LYXERR(Debug::FIND, "After escape: " << s);
729 pos += it->second.length();
730 // LYXERR(Debug::FIND, "pos: " << pos);
733 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
738 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
739 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
740 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
741 string escape_for_regex(string s, bool match_latex)
744 while (pos < s.size()) {
745 size_t new_pos = s.find("\\regexp{", pos);
746 if (new_pos == string::npos)
751 LYXERR(Debug::FIND, "new_pos: " << new_pos);
752 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
753 LYXERR(Debug::FIND, "t [lyx]: " << t);
754 t = apply_escapes(t, get_regexp_escapes());
755 LYXERR(Debug::FIND, "t [rxp]: " << t);
756 s.replace(pos, new_pos - pos, t);
757 new_pos = pos + t.size();
758 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
759 LYXERR(Debug::FIND, "new_pos: " << new_pos);
760 if (new_pos == s.size())
763 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
764 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
765 LYXERR(Debug::FIND, "end_pos: " << end_pos);
766 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
767 LYXERR(Debug::FIND, "t in regexp : " << t);
768 t = apply_escapes(t, get_lyx_unescapes());
769 LYXERR(Debug::FIND, "t in regexp after unescapes [lyx]: " << t);
771 t = apply_escapes(t, get_regexp_latex_escapes());
772 LYXERR(Debug::FIND, "t in regexp after latex_escapes [ltx]: " << t);
774 if (end_pos == s.size()) {
775 s.replace(new_pos, end_pos - new_pos, t);
776 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
779 s.replace(new_pos, end_pos + 13 - new_pos, t);
780 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
781 pos = new_pos + t.size();
782 LYXERR(Debug::FIND, "pos: " << pos);
788 /// Wrapper for lyx::regex_replace with simpler interface
789 bool regex_replace(string const & s, string & t, string const & searchstr,
790 string const & replacestr)
792 regex e(searchstr, regex_constants::ECMAScript);
794 ostream_iterator<char, char> it(oss);
795 regex_replace(it, s.begin(), s.end(), e, replacestr);
796 // tolerate t and s be references to the same variable
797 bool rv = (s != oss.str());
803 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
805 ** Verify that closed braces exactly match open braces. This avoids that, for example,
806 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
809 ** Number of open braces that must remain open at the end for the verification to succeed.
812 bool braces_match(QString const & beg,
815 bool braces_match(string const & beg,
821 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg.toStdString() << "'");
823 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg << "'");
825 int lastidx = beg.size();
826 for (int i=0; i < lastidx; ++i) {
827 // Skip escaped braces in the count
837 } else if (c == '{') {
839 } else if (c == '}') {
840 if (open_pars == 0) {
841 LYXERR(Debug::FIND, "Found unmatched closed brace");
847 if (open_pars != unmatched) {
848 LYXERR(Debug::FIND, "Found " << open_pars
849 << " instead of " << unmatched
850 << " unmatched open braces at the end of count");
853 LYXERR(Debug::FIND, "Braces match as expected");
870 #if defined(ResultsDebug)
871 vector <string> result = vector <string>();
873 MatchResult(): match_len(0),match_prefix(0),match2end(0), pos(0),leadsize(0) {};
876 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
878 int range = oldres.match_len;
879 if (range < 2) range = 2;
880 if (newres.match2end < oldres.match2end - oldres.match_len)
881 return MatchResult::newIsTooFar;
882 if (newres.match_len < oldres.match_len)
883 return MatchResult::newIsTooFar;
884 if ((newres.match_len == oldres.match_len) &&
885 (newres.match2end < oldres.match2end + range) &&
886 (newres.match2end > oldres.match2end - range)) {
887 return MatchResult::newIsBetter;
889 return MatchResult::newIsInvalid;
892 /** The class performing a match between a position in the document and the FindAdvOptions.
895 class MatchStringAdv {
897 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
899 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
900 ** constructor as opt.search, under the opt.* options settings.
903 ** If set, then match is searched only against beginning of text starting at cur.
904 ** If unset, then match is searched anywhere in text starting at cur.
907 ** The length of the matching text, or zero if no match was found.
909 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
918 /// first buffer on which search was started
919 lyx::Buffer * const p_first_buf;
921 FindAndReplaceOptions const & opt;
924 /// Auxiliary find method (does not account for opt.matchword)
925 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
927 /** Normalize a stringified or latexified LyX paragraph.
931 ** <li>if search is not casesensitive, then lowercase the string;
932 ** <li>remove any newline at begin or end of the string;
933 ** <li>replace any newline in the middle of the string with a simple space;
934 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
937 ** @todo Normalization should also expand macros, if the corresponding
938 ** search option was checked.
940 string normalize(docstring const & s, bool hack_braces) const;
941 // normalized string to search
942 string par_as_string;
943 // regular expression to use for searching
944 // regexp2 is same as regexp, but prefixed with a ".*?"
946 QRegularExpression regexp;
947 QRegularExpression regexp2;
952 // leading format material as string
953 string lead_as_string;
954 // par_as_string after removal of lead_as_string
955 string par_as_string_nolead;
956 // unmatched open braces in the search string/regexp
958 // number of (.*?) subexpressions added at end of search regexp for closing
959 // environments, math mode, styles, etc...
962 // Are we searching with regular expressions ?
967 static docstring buffer_to_latex(Buffer & buffer)
969 //OutputParams runparams(&buffer.params().encoding());
970 OutputParams runparams(encodings.fromLyXName("utf8"));
971 odocstringstream ods;
973 runparams.nice = true;
974 runparams.flavor = Flavor::XeTeX;
975 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
976 // No side effect of file copying and image conversion
977 runparams.dryrun = true;
978 runparams.for_search = true;
979 pit_type const endpit = buffer.paragraphs().size();
980 for (pit_type pit = 0; pit != endpit; ++pit) {
981 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
982 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
988 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
991 if (!opt.ignoreformat) {
992 str = buffer_to_latex(buffer);
994 // OutputParams runparams(&buffer.params().encoding());
995 OutputParams runparams(encodings.fromLyXName("utf8"));
996 runparams.nice = true;
997 runparams.flavor = Flavor::XeTeX;
998 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
999 runparams.dryrun = true;
1000 runparams.for_search = true;
1001 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
1002 Paragraph const & par = buffer.paragraphs().at(pit);
1003 LYXERR(Debug::FIND, "Adding to search string: '"
1004 << par.asString(pos_type(0), par.size(),
1005 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1008 str += par.asString(pos_type(0), par.size(),
1009 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1012 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1013 string t = to_utf8(str);
1014 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
1021 /// Return separation pos between the leading material and the rest
1022 static size_t identifyLeading(string const & s)
1025 // @TODO Support \item[text]
1026 // Kornel: Added textsl, textsf, textit, texttt and noun
1027 // + allow to search for colored text too
1028 while (regex_replace(t, t, "^\\\\(("
1029 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1030 "lyxaddress|lyxrightaddress|"
1031 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1032 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1033 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1034 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1035 || regex_replace(t, t, "^\\$", "")
1036 || regex_replace(t, t, "^\\\\\\[", "")
1037 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1038 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1040 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1045 * Given a latexified string, retrieve some handled features
1046 * The features of the regex will later be compared with the features
1047 * of the searched text. If the regex features are not a
1048 * subset of the analized, then, in not format ignoring search
1049 * we can early stop the search in the relevant inset.
1051 typedef map<string, bool> Features;
1053 static Features identifyFeatures(string const & s)
1055 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1056 static regex const valid("^("
1058 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1059 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1060 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1061 "(u|uu)line|(s|x)out|uwave|"
1062 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1063 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1068 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1071 if (sub.str(1).compare("regexp") == 0) {
1075 string token = sub.str(1);
1077 if (regex_match(token, sub2, valid)) {
1085 if (sub.str(1).compare("endregexp") == 0) {
1095 * defines values features of a key "\\[a-z]+{"
1100 /* Char type with content discarded
1101 * like \hspace{1cm} */
1103 /* Char, like \backslash */
1105 /* replace starting backslash with '#' */
1107 /* \part, \section*, ... */
1109 /* title, author etc */
1111 /* \foreignlanguage{ngerman}, ... */
1114 * to discard language in content */
1117 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1119 /* fonts, colors, markups, ... */
1121 /* footnotesize, ... large, ...
1122 * Ignore all of them */
1125 /* inputencoding, ...
1126 * Discard also content, because they do not help in search */
1129 * like remove, but also all arguments */
1131 /* item, listitem */
1133 /* tex, latex, ... like isChar */
1135 /* like \lettrine[lines=5]{}{} */
1137 // like isStandard, but always remove head
1139 /* End of arguments marker for lettrine,
1140 * so that they can be ignored */
1143 KeyInfo() = default;
1144 KeyInfo(KeyType type, int parcount, bool disable)
1146 parenthesiscount(parcount),
1147 disabled(disable) {}
1148 KeyType keytype = invalid;
1150 int _tokensize = -1;
1151 int _tokenstart = -1;
1152 int _dataStart = -1;
1154 int parenthesiscount = 1;
1155 bool disabled = false;
1156 bool used = false; /* by pattern */
1161 Border(int l=0, int u=0) : low(l), upper(u) {};
1166 #define MAXOPENED 30
1168 bool isPatternString_;
1170 explicit Intervall(bool isPattern, string const & p) :
1171 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1172 hasTitle(false), langcount(0)
1180 static vector<Border> borders;
1181 int depts[MAXOPENED];
1182 int closes[MAXOPENED];
1183 int actualdeptindex;
1184 int previousNotIgnored(int) const;
1185 int nextNotIgnored(int) const;
1186 void handleOpenP(int i);
1187 void handleCloseP(int i, bool closingAllowed);
1188 void resetOpenedP(int openPos);
1189 void addIntervall(int upper);
1190 void addIntervall(int low, int upper); /* if explicit */
1191 void removeAccents();
1192 void setForDefaultLang(KeyInfo const & defLang) const;
1193 int findclosing(int start, int end, char up, char down, int repeat);
1194 void handleParentheses(int lastpos, bool closingAllowed);
1196 int langcount; // Number of disabled language specs up to current position in actual interval
1197 int isOpeningPar(int pos) const;
1199 void output(ostringstream &os, int lastpos);
1200 // string show(int lastpos);
1203 vector<Border> Intervall::borders = vector<Border>(30);
1205 int Intervall::isOpeningPar(int pos) const
1207 if ((pos < 0) || (size_t(pos) >= par.size()))
1209 if (par[pos] != '{')
1211 if (size_t(pos) + 2 >= par.size())
1213 if (par[pos+2] != '}')
1215 if (par[pos+1] == '[' || par[pos+1] == ']')
1220 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1222 // Enable the use of first token again
1223 if (ignoreidx >= 0) {
1224 int value = defLang._tokenstart + defLang._tokensize;
1230 if (borders[borderidx].low < value)
1231 borders[borderidx].low = value;
1232 if (borders[borderidx].upper < value)
1233 borders[borderidx].upper = value;
1238 static void checkDepthIndex(int val)
1240 static int maxdepthidx = MAXOPENED-2;
1241 static int lastmaxdepth = 0;
1242 if (val > lastmaxdepth) {
1243 LYXERR(Debug::INFO, "Depth reached " << val);
1246 if (val > maxdepthidx) {
1248 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1253 // Not needed, because borders are now dynamically expanded
1254 static void checkIgnoreIdx(int val)
1256 static int lastmaxignore = -1;
1257 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1258 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1259 lastmaxignore = val;
1265 * Expand the region of ignored parts of the input latex string
1266 * The region is only relevant in output()
1268 void Intervall::addIntervall(int low, int upper)
1271 if (low == upper) return;
1272 for (idx = ignoreidx+1; idx > 0; --idx) {
1273 if (low > borders[idx-1].upper) {
1277 Border br(low, upper);
1278 if (idx > ignoreidx) {
1279 if (borders.size() <= size_t(idx)) {
1280 borders.push_back(br);
1286 // checkIgnoreIdx(ignoreidx);
1290 // Expand only if one of the new bound is inside the interwall
1291 // We know here that br.low > borders[idx-1].upper
1292 if (br.upper < borders[idx].low) {
1293 // We have to insert at this pos
1294 if (size_t(ignoreidx+1) >= borders.size()) {
1295 borders.push_back(borders[ignoreidx]);
1298 borders[ignoreidx+1] = borders[ignoreidx];
1300 for (int i = ignoreidx; i > idx; --i) {
1301 borders[i] = borders[i-1];
1305 // checkIgnoreIdx(ignoreidx);
1308 // Here we know, that we are overlapping
1309 if (br.low > borders[idx].low)
1310 br.low = borders[idx].low;
1311 // check what has to be concatenated
1313 for (int i = idx; i <= ignoreidx; i++) {
1314 if (br.upper >= borders[i].low) {
1316 if (br.upper < borders[i].upper)
1317 br.upper = borders[i].upper;
1323 // count should be >= 1 here
1326 for (int i = idx + count; i <= ignoreidx; i++) {
1327 borders[i-count+1] = borders[i];
1329 ignoreidx -= count - 1;
1335 typedef map<string, string> AccentsMap;
1336 static AccentsMap accents = map<string, string>();
1338 static void buildaccent(string n, string param, string values)
1342 const char delim = '|';
1343 while (getline(s, name, delim)) {
1345 for (char c : param) {
1346 string key = name + "{" + c + "}";
1347 // get the corresponding utf8-value
1348 if ((values[start] & 0xc0) != 0xc0) {
1349 // should not happen, utf8 encoding starts at least with 11xxxxxx
1350 // but value for '\dot{i}' is 'i', which is ascii
1351 if ((values[start] & 0x80) == 0) {
1353 accents[key] = values.substr(start, 1);
1354 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1359 for (int j = 1; ;j++) {
1360 if (start + j >= values.size()) {
1361 accents[key] = values.substr(start, j);
1362 start = values.size() - 1;
1365 else if ((values[start+j] & 0xc0) != 0x80) {
1366 // This is the first byte of following utf8 char
1367 accents[key] = values.substr(start, j);
1369 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1378 static string getutf8(unsigned uchar)
1381 string ret = string();
1383 if (uchar <= 0x7f) {
1384 c[maxc-1] = uchar & 0x7f;
1387 unsigned char rest = 0x40;
1388 unsigned char first = 0x80;
1390 for (int i = start; i >=0; --i) {
1392 c[i] = first + uchar;
1395 c[i] = 0x80 | (uchar & 0x3f);
1402 for (int i = 0; i < maxc; i++) {
1403 if (c[i] == 0) continue;
1409 static void buildAccentsMap()
1411 accents["imath"] = "ı";
1413 accents["jmath"] = "ȷ";
1414 accents["cdot"] = "·";
1415 accents["guillemotright"] = "»";
1416 accents["guillemotleft"] = "«";
1417 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1418 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1419 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1420 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1421 accents["negmedspace"] = getutf8(0xf0005);
1422 accents["thickspace"] = getutf8(0xf0006);
1423 accents["negthickspace"] = getutf8(0xf0007);
1424 accents["lyx"] = getutf8(0xf0010); // Used logos
1425 accents["LyX"] = getutf8(0xf0010);
1426 accents["tex"] = getutf8(0xf0011);
1427 accents["TeX"] = getutf8(0xf0011);
1428 accents["latex"] = getutf8(0xf0012);
1429 accents["LaTeX"] = getutf8(0xf0012);
1430 accents["latexe"] = getutf8(0xf0013);
1431 accents["LaTeXe"] = getutf8(0xf0013);
1432 accents["lyxarrow"] = getutf8(0xf0020);
1433 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1434 accents["backslash LyX"] = getutf8(0xf0010);
1435 accents["backslash tex"] = getutf8(0xf0011);
1436 accents["backslash TeX"] = getutf8(0xf0011);
1437 accents["backslash latex"] = getutf8(0xf0012);
1438 accents["backslash LaTeX"] = getutf8(0xf0012);
1439 accents["backslash latexe"] = getutf8(0xf0013);
1440 accents["backslash LaTeXe"] = getutf8(0xf0013);
1441 accents["backslash lyxarrow"] = getutf8(0xf0020);
1442 accents["ddot{\\imath}"] = "ï";
1443 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1444 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1445 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1446 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1447 accents["acute{\\imath}"] = "í";
1448 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1449 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1450 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1451 buildaccent("mathring|r", "aAuUwy",
1453 accents["check{\\imath}"] = "ǐ";
1454 accents["check{\\jmath}"] = "ǰ";
1455 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1456 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1457 accents["hat{\\imath}"] = "î";
1458 accents["hat{\\jmath}"] = "ĵ";
1459 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1460 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1461 accents["bar{\\imath}"] = "ī";
1462 buildaccent("bar|=", "aAeEiIoOuUyY",
1463 "āĀēĒīĪōŌūŪȳȲ"); // macron
1464 accents["tilde{\\imath}"] = "ĩ";
1465 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1466 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1467 accents["breve{\\imath}"] = "ĭ";
1468 buildaccent("breve|u", "aAeEgGiIoOuU",
1469 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1470 accents["grave{\\imath}"] = "ì";
1471 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1472 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1473 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1474 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1475 buildaccent("ogonek|k", "AaEeIiUuOo",
1476 "ĄąĘęĮįŲųǪǫ"); // ogonek
1477 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1478 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1479 buildaccent("subring|textsubring", "Aa",
1481 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1482 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1483 buildaccent("subtilde|textsubtilde", "EeIiUu",
1484 "ḚḛḬḭṴṵ"); // subtilde
1485 accents["dgrave{\\imath}"] = "ȉ";
1486 accents["textdoublegrave{\\i}"] = "ȉ";
1487 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1488 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1489 accents["rcap{\\imath}"] = "ȉ";
1490 accents["textroundcap{\\i}"] = "ȉ";
1491 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1492 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1493 buildaccent("slashed", "oO",
1498 * Created accents in math or regexp environment
1499 * are macros, but we need the utf8 equivalent
1501 void Intervall::removeAccents()
1503 if (accents.empty())
1505 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1506 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1507 "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))|guillemot(left|right))(?![a-zA-Z]))");
1509 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1511 string key = sub.str(1);
1512 if (accents.find(key) != accents.end()) {
1513 string val = accents[key];
1514 size_t pos = sub.position(size_t(0));
1515 for (size_t i = 0; i < val.size(); i++) {
1516 par[pos+i] = val[i];
1518 // Remove possibly following space too
1519 if (par[pos+sub.str(0).size()] == ' ')
1520 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1522 addIntervall(pos+val.size(), pos + sub.str(0).size());
1523 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1524 // remove traces of any remaining chars
1529 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1534 void Intervall::handleOpenP(int i)
1537 depts[actualdeptindex] = i+1;
1538 closes[actualdeptindex] = -1;
1539 checkDepthIndex(actualdeptindex);
1542 void Intervall::handleCloseP(int i, bool closingAllowed)
1544 if (actualdeptindex <= 0) {
1545 if (! closingAllowed)
1546 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1547 // if we are at the very end
1548 addIntervall(i, i+1);
1551 closes[actualdeptindex] = i+1;
1556 void Intervall::resetOpenedP(int openPos)
1558 // Used as initializer for foreignlanguage entry
1559 actualdeptindex = 1;
1560 depts[1] = openPos+1;
1564 int Intervall::previousNotIgnored(int start) const
1566 int idx = 0; /* int intervalls */
1567 for (idx = ignoreidx; idx >= 0; --idx) {
1568 if (start > borders[idx].upper)
1570 if (start >= borders[idx].low)
1571 start = borders[idx].low-1;
1576 int Intervall::nextNotIgnored(int start) const
1578 int idx = 0; /* int intervalls */
1579 for (idx = 0; idx <= ignoreidx; idx++) {
1580 if (start < borders[idx].low)
1582 if (start < borders[idx].upper)
1583 start = borders[idx].upper;
1588 typedef map<string, KeyInfo> KeysMap;
1589 typedef vector< KeyInfo> Entries;
1590 static KeysMap keys = map<string, KeyInfo>();
1596 Intervall interval_;
1597 void buildKeys(bool);
1598 void buildEntries(bool);
1599 void makeKey(const string &, KeyInfo, bool isPatternString);
1600 void processRegion(int start, int region_end); /* remove {} parts */
1601 void removeHead(KeyInfo const &, int count=0);
1604 LatexInfo(string const & par, bool isPatternString)
1605 : entidx_(-1), interval_(isPatternString, par)
1607 buildKeys(isPatternString);
1608 entries_ = vector<KeyInfo>();
1609 buildEntries(isPatternString);
1613 if (entries_.empty()) {
1616 if (entries_[0].keytype == KeyInfo::isTitle) {
1617 interval_.hasTitle = true;
1618 if (! entries_[0].disabled) {
1619 interval_.titleValue = entries_[0].head;
1622 interval_.titleValue = "";
1624 removeHead(entries_[0]);
1625 if (entries_.size() > 1)
1634 if (int(entries_.size()) > entidx_) {
1641 bool setNextKey(int idx) {
1642 if ((idx == entidx_) && (entidx_ >= 0)) {
1649 int find(int start, KeyInfo::KeyType keytype) const {
1653 while (tmpIdx < int(entries_.size())) {
1654 if (entries_[tmpIdx].keytype == keytype)
1660 int process(ostringstream & os, KeyInfo const & actual);
1661 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1662 // string show(int lastpos) { return interval.show(lastpos);};
1663 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1664 KeyInfo &getKeyInfo(int keyinfo) {
1665 static KeyInfo invalidInfo = KeyInfo();
1666 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1669 return entries_[keyinfo];
1671 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1672 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1676 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1680 for (int i = start; i < end; i += 1 + skip) {
1684 if (c == '\\') skip = 1;
1688 else if (c == down) {
1691 if ((repeat <= 0) || (par[i+1] != up))
1709 vector<MathEntry> entries_;
1714 void insert(string const & wait, size_t start, size_t end) {
1715 MathEntry m = MathEntry();
1717 m.mathStart = start;
1719 m.mathSize = end - start;
1720 entries_.push_back(m);
1722 bool empty() const { return entries_.empty(); };
1723 size_t getEndPos() const {
1724 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1727 return entries_[actualIdx_].mathEnd;
1729 size_t getStartPos() const {
1730 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1731 return 100000; /* definitely enough? */
1733 return entries_[actualIdx_].mathStart;
1735 size_t getFirstPos() {
1737 return getStartPos();
1739 size_t getSize() const {
1740 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1743 return entries_[actualIdx_].mathSize;
1745 void incrEntry() { actualIdx_++; };
1748 void LatexInfo::buildEntries(bool isPatternString)
1750 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\})");
1751 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1752 static bool disableLanguageOverride = false;
1753 smatch sub, submath;
1754 bool evaluatingRegexp = false;
1756 bool evaluatingMath = false;
1757 bool evaluatingCode = false;
1759 bool evaluatingOptional = false;
1760 size_t optionalEnd = 0;
1763 bool math_end_waiting = false;
1764 size_t math_pos = 10000;
1766 static vector<string> usedText = vector<string>();
1768 interval_.removeAccents();
1770 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1772 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1773 // prefixed by odd count of '\\'
1776 if (math_end_waiting) {
1777 size_t pos = submath.position(size_t(2));
1778 if ((math_end == "$") &&
1779 (submath.str(2) == "$")) {
1780 mi.insert("$", math_pos, pos + 1);
1781 math_end_waiting = false;
1783 else if ((math_end == "\\]") &&
1784 (submath.str(2) == "\\]")) {
1785 mi.insert("\\]", math_pos, pos + 2);
1786 math_end_waiting = false;
1788 else if ((submath.str(3).compare("end") == 0) &&
1789 (submath.str(4).compare(math_end) == 0)) {
1790 mi.insert(math_end, math_pos, pos + submath.str(2).length());
1791 math_end_waiting = false;
1797 if (submath.str(3).compare("begin") == 0) {
1798 math_end_waiting = true;
1799 math_end = submath.str(4);
1800 math_pos = submath.position(size_t(2));
1802 else if (submath.str(2).compare("\\[") == 0) {
1803 math_end_waiting = true;
1805 math_pos = submath.position(size_t(2));
1807 else if (submath.str(2) == "$") {
1808 size_t pos = submath.position(size_t(2));
1809 math_end_waiting = true;
1815 // Ignore language if there is math somewhere in pattern-string
1816 if (isPatternString) {
1817 for (auto s: usedText) {
1818 // Remove entries created in previous search runs
1821 usedText = vector<string>();
1824 keys["foreignlanguage"].disabled = true;
1825 disableLanguageOverride = true;
1828 disableLanguageOverride = false;
1831 if (disableLanguageOverride) {
1832 keys["foreignlanguage"].disabled = true;
1835 math_pos = mi.getFirstPos();
1836 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1838 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1839 // prefixed by odd count of '\\'
1842 string key = sub.str(5);
1844 if (sub.str(2)[0] == '\\')
1845 key = sub.str(2)[1];
1850 if (keys.find(key) != keys.end()) {
1851 if (keys[key].keytype == KeyInfo::headRemove) {
1852 KeyInfo found1 = keys[key];
1853 found1.disabled = true;
1854 found1.head = "\\" + key + "{";
1855 found1._tokenstart = sub.position(size_t(2));
1856 found1._tokensize = found1.head.length();
1857 found1._dataStart = found1._tokenstart + found1.head.length();
1858 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1859 found1._dataEnd = endpos;
1864 if (evaluatingRegexp) {
1865 if (sub.str(3).compare("endregexp") == 0) {
1866 evaluatingRegexp = false;
1867 // found._tokenstart already set
1868 found._dataEnd = sub.position(size_t(2)) + 13;
1869 found._dataStart = found._dataEnd;
1870 found._tokensize = found._dataEnd - found._tokenstart;
1871 found.parenthesiscount = 0;
1872 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1879 if (evaluatingMath) {
1880 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1882 evaluatingMath = false;
1884 math_pos = mi.getStartPos();
1886 if (keys.find(key) == keys.end()) {
1887 found = KeyInfo(KeyInfo::isStandard, 0, true);
1888 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1889 found = KeyInfo(KeyInfo::isText, 0, false);
1890 if (isPatternString) {
1891 found.keytype = KeyInfo::isChar;
1892 found.disabled = false;
1896 usedText.push_back(key);
1900 if (key.compare("regexp") == 0) {
1901 evaluatingRegexp = true;
1902 found._tokenstart = sub.position(size_t(2));
1903 found._tokensize = 0;
1907 // Handle the other params of key
1908 if (found.keytype == KeyInfo::isIgnored)
1910 else if (found.keytype == KeyInfo::isMath) {
1911 if (size_t(sub.position(size_t(2))) == math_pos) {
1913 found._tokenstart = sub.position(size_t(2));
1914 found._tokensize = mi.getSize();
1915 found._dataEnd = found._tokenstart + found._tokensize;
1916 found._dataStart = found._dataEnd;
1917 found.parenthesiscount = 0;
1918 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1919 evaluatingMath = true;
1922 // begin|end of unknown env, discard
1923 // First handle tables
1924 // longtable|tabular
1925 bool discardComment;
1927 found.keytype = KeyInfo::doRemove;
1928 if ((sub.str(7).compare("longtable") == 0) ||
1929 (sub.str(7).compare("tabular") == 0)) {
1930 discardComment = true; /* '%' */
1933 discardComment = false;
1934 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1936 string token = sub.str(7);
1937 if (regex_match(token, sub2, removeArgs)) {
1938 found.keytype = KeyInfo::removeWithArg;
1941 // discard spaces before pos(2)
1942 int pos = sub.position(size_t(2));
1944 for (count = 0; pos - count > 0; count++) {
1945 char c = interval_.par[pos-count-1];
1946 if (discardComment) {
1947 if ((c != ' ') && (c != '%'))
1953 found._tokenstart = pos - count;
1954 if (sub.str(3).compare(0, 5, "begin") == 0) {
1955 size_t pos1 = pos + sub.str(2).length();
1956 if (sub.str(7).compare("cjk") == 0) {
1957 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1958 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1960 found.keytype = KeyInfo::isMain;
1961 found._dataStart = pos1;
1962 found._dataEnd = interval_.par.length();
1963 found.disabled = keys["foreignlanguage"].disabled;
1964 found.used = keys["foreignlanguage"].used;
1965 found._tokensize = pos1 - found._tokenstart;
1966 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1969 // Swallow possible optional params
1970 while (interval_.par[pos1] == '[') {
1971 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1973 // Swallow also the eventual parameter
1974 if (interval_.par[pos1] == '{') {
1975 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1978 found._dataEnd = pos1;
1980 found._dataStart = found._dataEnd;
1981 found._tokensize = count + found._dataEnd - pos;
1982 found.parenthesiscount = 0;
1983 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1984 found.disabled = true;
1988 // Handle "\end{...}"
1989 found._dataStart = pos + sub.str(2).length();
1990 found._dataEnd = found._dataStart;
1991 found._tokensize = count + found._dataEnd - pos;
1992 found.parenthesiscount = 0;
1993 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1994 found.disabled = true;
1998 else if (found.keytype != KeyInfo::isRegex) {
1999 found._tokenstart = sub.position(size_t(2));
2000 if (found.parenthesiscount == 0) {
2001 // Probably to be discarded
2002 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
2003 char following = interval_.par[following_pos];
2004 if (following == ' ')
2005 found.head = "\\" + sub.str(5) + " ";
2006 else if (following == '=') {
2007 // like \uldepth=1000pt
2008 found.head = sub.str(2);
2011 found.head = "\\" + key;
2012 found._tokensize = found.head.length();
2013 found._dataEnd = found._tokenstart + found._tokensize;
2014 found._dataStart = found._dataEnd;
2017 int params = found._tokenstart + key.length() + 1;
2018 if (evaluatingOptional) {
2019 if (size_t(found._tokenstart) > optionalEnd) {
2020 evaluatingOptional = false;
2023 found.disabled = true;
2026 int optend = params;
2027 while (interval_.par[optend] == '[') {
2028 // discard optional parameters
2029 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2031 if (optend > params) {
2032 key += interval_.par.substr(params, optend-params);
2033 evaluatingOptional = true;
2034 optionalEnd = optend;
2036 string token = sub.str(7);
2038 if (interval_.par[optend] != '{') {
2040 found.parenthesiscount = 0;
2041 found.head = "\\" + key;
2044 closings = found.parenthesiscount;
2045 if (found.parenthesiscount == 1) {
2046 found.head = "\\" + key + "{";
2048 else if (found.parenthesiscount > 1) {
2050 found.head = sub.str(2) + "{";
2051 closings = found.parenthesiscount - 1;
2054 found.head = "\\" + key + "{";
2057 found._tokensize = found.head.length();
2058 found._dataStart = found._tokenstart + found.head.length();
2059 if (found.keytype == KeyInfo::doRemove) {
2061 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2062 if (endpar >= interval_.par.length())
2063 found._dataStart = interval_.par.length();
2065 found._dataStart = endpar;
2066 found._tokensize = found._dataStart - found._tokenstart;
2069 found._dataStart = found._tokenstart + found._tokensize;
2073 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
2074 found._dataStart += 15;
2078 endpos = found._dataStart - 1;
2080 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2081 if (found.keytype == KeyInfo::isList) {
2082 // Check if it really is list env
2083 static regex const listre("^([a-z]+)$");
2085 if (!regex_match(token, sub2, listre)) {
2086 // Change the key of this entry. It is not in a list/item environment
2087 found.keytype = KeyInfo::endArguments;
2090 if (found.keytype == KeyInfo::noMain) {
2091 evaluatingCode = true;
2093 codeStart = found._dataStart;
2095 else if (evaluatingCode) {
2096 if (size_t(found._dataStart) > codeEnd)
2097 evaluatingCode = false;
2098 else if (found.keytype == KeyInfo::isMain) {
2099 // Disable this key, treate it as standard
2100 found.keytype = KeyInfo::isStandard;
2101 found.disabled = true;
2102 if ((codeEnd +1 >= interval_.par.length()) &&
2103 (found._tokenstart == codeStart)) {
2104 // trickery, because the code inset starts
2105 // with \selectlanguage ...
2107 if (entries_.size() > 1) {
2108 entries_[entries_.size()-1]._dataEnd = codeEnd;
2113 if ((endpos == interval_.par.length()) &&
2114 (found.keytype == KeyInfo::doRemove)) {
2115 // Missing closing => error in latex-input?
2116 // therefore do not delete remaining data
2117 found._dataStart -= 1;
2118 found._dataEnd = found._dataStart;
2121 found._dataEnd = endpos;
2123 if (isPatternString) {
2124 keys[key].used = true;
2127 entries_.push_back(found);
2131 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2133 stringstream s(keysstring);
2135 const char delim = '|';
2136 while (getline(s, key, delim)) {
2137 KeyInfo keyII(keyI);
2138 if (isPatternString) {
2141 else if ( !keys[key].used)
2142 keyII.disabled = true;
2147 void LatexInfo::buildKeys(bool isPatternString)
2150 static bool keysBuilt = false;
2151 if (keysBuilt && !isPatternString) return;
2153 // Keys to ignore in any case
2154 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2155 // Known standard keys with 1 parameter.
2156 // Split is done, if not at start of region
2157 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2158 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2159 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2160 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2161 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2162 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2164 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2165 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2166 makeKey("section*|subsection*|subsubsection*|paragraph*",
2167 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2168 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2169 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2171 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2173 // Split is done, if not at start of region
2174 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2175 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2177 // Split is done always.
2178 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2182 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2183 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2184 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2185 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2187 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2188 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2189 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2190 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2192 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2193 // Custom space/skip, remove the content (== length value)
2194 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2195 // Found in fr/UserGuide.lyx
2196 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2198 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2199 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2200 // Known macros to remove (including their parameter)
2202 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2203 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2204 // handle like standard keys with 1 parameter.
2205 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2207 // Ignore deleted text
2208 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2209 // but preserve added text
2210 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2212 // Macros to remove, but let the parameter survive
2214 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2216 // Remove language spec from content of these insets
2217 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2219 // Same effect as previous, parameter will survive (because there is no one anyway)
2221 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2222 // Remove table decorations
2223 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2224 // Discard shape-header.
2225 // For footnote or shortcut too, because of lang settings
2226 // and wrong handling if used 'KeyInfo::noMain'
2227 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2228 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2229 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2230 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2231 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2232 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2233 // like ('tiny{}' or '\tiny ' ... )
2234 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2236 // Survives, like known character
2237 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2238 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2239 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2241 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2242 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2243 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2245 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2246 // Remove RTL/LTR marker
2247 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2248 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2249 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2250 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2251 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2252 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2253 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2254 if (isPatternString) {
2255 // Allow the first searched string to rebuild the keys too
2259 // no need to rebuild again
2265 * Keep the list of actual opened parentheses actual
2266 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2268 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2271 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2275 if (c == '\\') skip = 1;
2276 else if (c == '{') {
2279 else if (c == '}') {
2280 handleCloseP(i, closingAllowed);
2286 string Intervall::show(int lastpos)
2288 int idx = 0; /* int intervalls */
2291 for (idx = 0; idx <= ignoreidx; idx++) {
2292 while (i < lastpos) {
2294 if (i <= borders[idx].low) {
2295 if (borders[idx].low > lastpos)
2296 printsize = lastpos - i;
2298 printsize = borders[idx].low - i;
2299 s += par.substr(i, printsize);
2301 if (i >= borders[idx].low)
2302 i = borders[idx].upper;
2305 i = borders[idx].upper;
2311 s += par.substr(i, lastpos-i);
2317 void Intervall::output(ostringstream &os, int lastpos)
2319 // get number of chars to output
2320 int idx = 0; /* int intervalls */
2323 string startTitle = titleValue;
2324 for (idx = 0; idx <= ignoreidx; idx++) {
2326 if (i <= borders[idx].low) {
2328 if (borders[idx].low > lastpos)
2329 printsize = lastpos - i;
2331 printsize = borders[idx].low - i;
2332 if (printsize > 0) {
2333 os << startTitle << par.substr(i, printsize);
2335 printed += printsize;
2338 handleParentheses(i, false);
2339 if (i >= borders[idx].low)
2340 i = borders[idx].upper;
2343 i = borders[idx].upper;
2350 os << startTitle << par.substr(i, lastpos-i);
2351 printed += lastpos-i;
2353 handleParentheses(lastpos, false);
2355 if (keys["foreignlanguage"].disabled)
2356 startindex = actualdeptindex-langcount;
2358 startindex = actualdeptindex;
2359 for (int i = startindex; i > 0; --i) {
2362 if (hasTitle && (printed > 0))
2364 if (! isPatternString_)
2366 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2369 void LatexInfo::processRegion(int start, int region_end)
2371 while (start < region_end) { /* Let {[} and {]} survive */
2372 int cnt = interval_.isOpeningPar(start);
2374 // Closing is allowed past the region
2375 int closing = interval_.findclosing(start+1, interval_.par.length());
2376 interval_.addIntervall(start, start+1);
2377 interval_.addIntervall(closing, closing+1);
2381 start = interval_.nextNotIgnored(start+1);
2385 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2387 if (actual.parenthesiscount == 0) {
2388 // "{\tiny{} ...}" ==> "{{} ...}"
2389 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2392 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2393 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2394 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2398 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2401 switch (actual.keytype)
2403 case KeyInfo::isTitle: {
2405 nextKeyIdx = getNextKey();
2408 case KeyInfo::cleanToStart: {
2409 actual._dataEnd = actual._dataStart;
2410 nextKeyIdx = getNextKey();
2411 // Search for end of arguments
2412 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2414 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2415 entries_[i].disabled = true;
2417 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2419 while (interval_.par[actual._dataEnd] == ' ')
2421 interval_.addIntervall(0, actual._dataEnd+1);
2422 interval_.actualdeptindex = 0;
2423 interval_.depts[0] = actual._dataEnd+1;
2424 interval_.closes[0] = -1;
2427 case KeyInfo::isText:
2428 interval_.par[actual._tokenstart] = '#';
2429 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2430 nextKeyIdx = getNextKey();
2432 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2433 if (actual.disabled)
2434 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2436 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2439 case KeyInfo::isChar: {
2440 nextKeyIdx = getNextKey();
2443 case KeyInfo::isSize: {
2444 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2445 if (actual.parenthesiscount == 0)
2446 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2448 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2450 nextKeyIdx = getNextKey();
2452 // Here _dataStart points to '{', so correct it
2453 actual._dataStart += 1;
2454 actual._tokensize += 1;
2455 actual.parenthesiscount = 1;
2456 if (interval_.par[actual._dataStart] == '}') {
2457 // Determine the end if used like '{\tiny{}...}'
2458 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2459 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2462 // Determine the end if used like '\tiny{...}'
2463 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2465 // Split on this key if not at start
2466 int start = interval_.nextNotIgnored(previousStart);
2467 if (start < actual._tokenstart) {
2468 interval_.output(os, actual._tokenstart);
2469 interval_.addIntervall(start, actual._tokenstart);
2471 // discard entry if at end of actual
2472 nextKeyIdx = process(os, actual);
2476 case KeyInfo::endArguments: {
2477 // Remove trailing '{}' too
2478 actual._dataStart += 1;
2479 actual._dataEnd += 1;
2480 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2481 nextKeyIdx = getNextKey();
2484 case KeyInfo::noMain:
2486 case KeyInfo::isStandard: {
2487 if (actual.disabled) {
2489 processRegion(actual._dataStart, actual._dataStart+1);
2490 nextKeyIdx = getNextKey();
2492 // Split on this key if not at datastart of calling entry
2493 int start = interval_.nextNotIgnored(previousStart);
2494 if (start < actual._tokenstart) {
2495 interval_.output(os, actual._tokenstart);
2496 interval_.addIntervall(start, actual._tokenstart);
2498 // discard entry if at end of actual
2499 nextKeyIdx = process(os, actual);
2503 case KeyInfo::removeWithArg: {
2504 nextKeyIdx = getNextKey();
2505 // Search for end of arguments
2506 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2508 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2509 entries_[i].disabled = true;
2511 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2513 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2516 case KeyInfo::doRemove: {
2517 // Remove the key with all parameters and following spaces
2519 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2520 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2523 // Remove also enclosing parentheses [] and {}
2526 while (actual._tokenstart > numpars) {
2527 if (pos+numpars >= interval_.par.size())
2529 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2531 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2537 if (interval_.par[pos+numpars] == ' ')
2541 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2542 nextKeyIdx = getNextKey();
2545 case KeyInfo::isList: {
2546 // Discard space before _tokenstart
2548 for (count = 0; count < actual._tokenstart; count++) {
2549 if (interval_.par[actual._tokenstart-count-1] != ' ')
2552 nextKeyIdx = getNextKey();
2553 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2555 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2557 // How else can we catch this one?
2558 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2559 entries_[i].disabled = true;
2561 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2563 else if (nextKeyIdx > 0) {
2564 // Ignore any lang entries inside data region
2565 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2566 if (entries_[i].keytype == KeyInfo::isMain)
2567 entries_[i].disabled = true;
2570 if (actual.disabled) {
2571 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2574 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2576 if (interval_.par[actual._dataEnd+1] == '[') {
2577 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2578 if ((interval_.par[actual._dataEnd+2] == '{') &&
2579 (interval_.par[posdown-1] == '}')) {
2580 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2581 interval_.addIntervall(posdown-1, posdown+1);
2584 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2585 interval_.addIntervall(posdown, posdown+1);
2587 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2588 if (blk > posdown) {
2589 // Discard at most 1 space after empty item
2591 for (count = 0; count < 1; count++) {
2592 if (interval_.par[blk+count] != ' ')
2596 interval_.addIntervall(blk, blk+count);
2601 case KeyInfo::isSectioning: {
2602 // Discard spaces before _tokenstart
2604 int val = actual._tokenstart;
2605 for (count = 0; count < actual._tokenstart;) {
2606 val = interval_.previousNotIgnored(val-1);
2607 if (val < 0 || interval_.par[val] != ' ')
2610 count = actual._tokenstart - val;
2613 if (actual.disabled) {
2614 removeHead(actual, count);
2615 nextKeyIdx = getNextKey();
2617 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2618 nextKeyIdx = process(os, actual);
2622 case KeyInfo::isMath: {
2623 // Same as regex, use the content unchanged
2624 nextKeyIdx = getNextKey();
2627 case KeyInfo::isRegex: {
2628 // DO NOT SPLIT ON REGEX
2630 nextKeyIdx = getNextKey();
2633 case KeyInfo::isIgnored: {
2634 // Treat like a character for now
2635 nextKeyIdx = getNextKey();
2638 case KeyInfo::isMain: {
2639 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2640 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2641 if (actual._tokenstart > 0) {
2642 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2643 if ((prev >= 0) && interval_.par[prev] == '%')
2644 interval_.addIntervall(prev, prev+1);
2646 if (actual.disabled) {
2648 interval_.langcount++;
2649 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2650 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2651 // Discard also the space before math-equation
2652 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2654 nextKeyIdx = getNextKey();
2655 // interval.resetOpenedP(actual._dataStart-1);
2658 if (actual._tokenstart < 26) {
2659 // for the first (and maybe dummy) language
2660 interval_.setForDefaultLang(actual);
2662 interval_.resetOpenedP(actual._dataStart-1);
2666 case KeyInfo::invalid:
2667 case KeyInfo::headRemove:
2668 // These two cases cannot happen, already handled
2671 // LYXERR(Debug::INFO, "Unhandled keytype");
2672 nextKeyIdx = getNextKey();
2679 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2681 int end = interval_.nextNotIgnored(actual._dataEnd);
2682 int oldStart = actual._dataStart;
2683 int nextKeyIdx = getNextKey();
2685 if ((nextKeyIdx < 0) ||
2686 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2687 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2688 if (oldStart <= end) {
2689 processRegion(oldStart, end);
2694 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2696 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2697 (void) dispatch(os, actual._dataStart, nextKey);
2698 end = nextKey._tokenstart;
2701 processRegion(oldStart, nextKey._tokenstart);
2702 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2704 oldStart = nextKey._dataEnd+1;
2706 // now nextKey is either invalid or is outside of actual._dataEnd
2707 // output the remaining and discard myself
2708 if (oldStart <= end) {
2709 processRegion(oldStart, end);
2711 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2713 // This is the normal case.
2714 // But if using the firstlanguage, the closing may be missing
2716 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2718 if (actual._dataEnd < end)
2719 output_end = interval_.nextNotIgnored(actual._dataEnd);
2720 else if (interval_.par.size() > (size_t) end)
2721 output_end = interval_.nextNotIgnored(end);
2723 output_end = interval_.par.size();
2724 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2725 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2727 // Remove possible empty data
2728 int dstart = interval_.nextNotIgnored(actual._dataStart);
2729 while (interval_.isOpeningPar(dstart) == 1) {
2730 interval_.addIntervall(dstart, dstart+1);
2731 int dend = interval_.findclosing(dstart+1, output_end);
2732 interval_.addIntervall(dend, dend+1);
2733 dstart = interval_.nextNotIgnored(dstart+1);
2735 if (dstart < output_end)
2736 interval_.output(os, output_end);
2737 interval_.addIntervall(actual._tokenstart, end);
2741 string splitOnKnownMacros(string par, bool isPatternString)
2744 LatexInfo li(par, isPatternString);
2745 // LYXERR(Debug::INFO, "Berfore split: " << par);
2746 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2748 DummyKey._tokensize = 0;
2749 DummyKey._dataStart = 0;
2750 DummyKey._dataEnd = par.length();
2751 DummyKey.disabled = true;
2752 int firstkeyIdx = li.getFirstKey();
2754 if (firstkeyIdx >= 0) {
2755 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2756 DummyKey._tokenstart = firstKey._tokenstart;
2758 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2759 // Use dummy firstKey
2760 firstKey = DummyKey;
2761 (void) li.setNextKey(firstkeyIdx);
2764 if (par.substr(firstKey._dataStart, 2) == "% ")
2765 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2767 nextkeyIdx = li.process(os, firstKey);
2768 while (nextkeyIdx >= 0) {
2769 // Check for a possible gap between the last
2770 // entry and this one
2771 int datastart = li.nextNotIgnored(firstKey._dataStart);
2772 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2773 if ((nextKey._tokenstart > datastart)) {
2775 firstKey._dataStart = datastart;
2776 firstKey._dataEnd = par.length();
2777 (void) li.setNextKey(nextkeyIdx);
2778 // Fake the last opened parenthesis
2779 li.setForDefaultLang(firstKey);
2780 nextkeyIdx = li.process(os, firstKey);
2783 if (nextKey.keytype != KeyInfo::isMain) {
2784 firstKey._dataStart = datastart;
2785 firstKey._dataEnd = nextKey._dataEnd+1;
2786 (void) li.setNextKey(nextkeyIdx);
2787 li.setForDefaultLang(firstKey);
2788 nextkeyIdx = li.process(os, firstKey);
2791 nextkeyIdx = li.process(os, nextKey);
2795 // Handle the remaining
2796 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2797 firstKey._dataEnd = par.length();
2799 if ((firstKey._dataStart < firstKey._dataEnd) &&
2800 (par[firstKey._dataStart] != '}')) {
2801 li.setForDefaultLang(firstKey);
2802 (void) li.process(os, firstKey);
2806 // return string definitelly impossible to match
2807 s = "\\foreignlanguage{ignore}{ }";
2811 s = par; /* no known macros found */
2812 // LYXERR(Debug::INFO, "After split: " << s);
2817 * Try to unify the language specs in the latexified text.
2818 * Resulting modified string is set to "", if
2819 * the searched tex does not contain all the features in the search pattern
2821 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2823 static Features regex_f;
2824 static int missed = 0;
2825 static bool regex_with_format = false;
2827 int parlen = par.length();
2829 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2832 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2833 // Happens to be there in case of description or labeling environment
2838 // Split the latex input into pieces which
2839 // can be digested by our search engine
2840 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2841 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2842 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2845 result = par.substr(0, parlen);
2846 if (isPatternString) {
2849 regex_f = identifyFeatures(result);
2850 string features = "";
2851 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2852 string a = it->first;
2853 regex_with_format = true;
2854 features += " " + a;
2855 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2857 LYXERR(Debug::FIND, "Identified Features" << features);
2860 } else if (regex_with_format) {
2861 Features info = identifyFeatures(result);
2862 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2863 string a = it->first;
2864 bool b = it->second;
2865 if (b && ! info[a]) {
2867 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2873 // LYXERR(Debug::INFO, "No regex formats");
2879 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2880 static int identifyClosing(string & t)
2882 int open_braces = 0;
2884 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2885 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2887 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2889 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2891 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2900 static int num_replaced = 0;
2901 static bool previous_single_replace = true;
2903 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2904 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2906 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2907 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2908 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2909 if (opt.replace_all && previous_single_replace) {
2910 previous_single_replace = false;
2913 else if (!opt.replace_all) {
2914 num_replaced = 0; // count number of replaced strings
2915 previous_single_replace = true;
2917 // When using regexp, braces are hacked already by escape_for_regex()
2918 par_as_string = normalize(ds, !use_regexp);
2920 close_wildcards = 0;
2922 size_t lead_size = 0;
2923 // correct the language settings
2924 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2925 if (opt.ignoreformat) {
2927 // if par_as_string_nolead were emty,
2928 // the following call to findAux will always *find* the string
2929 // in the checked data, and thus always using the slow
2930 // examining of the current text part.
2931 par_as_string_nolead = par_as_string;
2934 lead_size = identifyLeading(par_as_string);
2935 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2936 lead_as_string = par_as_string.substr(0, lead_size);
2937 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2941 open_braces = identifyClosing(par_as_string);
2942 identifyClosing(par_as_string_nolead);
2943 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2944 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2946 string lead_as_regexp;
2947 if (lead_size > 0) {
2948 // @todo No need to search for \regexp{} insets in leading material
2949 static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\\])" };
2950 lead_as_regexp = std::regex_replace(par_as_string.substr(0, lead_size), specialChars, R"(\$&)" );
2951 // lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2952 par_as_string = par_as_string_nolead;
2953 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2954 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2956 LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
2957 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2958 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2959 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2961 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2962 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2963 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2964 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2965 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2966 || regex_replace(par_as_string, par_as_string,
2967 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2968 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2969 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2973 if (!opt.ignoreformat) {
2974 // Remove extra '\}' at end if not part of \{\.\}
2975 size_t lng = par_as_string.size();
2977 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2979 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2988 if (lng < par_as_string.size())
2989 par_as_string = par_as_string.substr(0,lng);
2992 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2993 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2994 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2995 // replace '[^...]' with '[^...\}\{\\]'
2996 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2997 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2998 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
3000 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
3003 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3004 LYXERR(Debug::FIND, "Open braces: " << open_braces);
3005 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
3006 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3008 // If entered regexp must match at begin of searched string buffer
3009 // Kornel: Added parentheses to use $1 for size of the leading string
3013 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3014 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3015 // so the convert has no effect in that case
3016 for (int i = 7; i > 0; --i) {
3017 string orig = "\\\\" + std::to_string(i);
3018 string dest = "\\" + std::to_string(i+2);
3019 while (regex_replace(par_as_string, par_as_string, orig, dest));
3021 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3022 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3024 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3025 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3027 // Handle \w properly
3028 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3029 if (! opt.casesensitive) {
3030 popts |= QRegularExpression::CaseInsensitiveOption;
3032 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3033 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3035 if (regexp.isValid() && regexp2.isValid()) {
3036 regexIsValid = true;
3037 // Check '{', '}' pairs inside the regex
3040 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3041 char c = par_as_string[i];
3048 else if (c == '}') {
3055 if (balanced != 0) {
3056 regexIsValid = false;
3057 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3061 regexIsValid = false;
3062 if (!regexp.isValid())
3063 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3064 if (!regexp2.isValid())
3065 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3068 if (opt.casesensitive) {
3069 regexp = regex(regexp_str);
3070 regexp2 = regex(regexp2_str);
3073 regexp = regex(regexp_str, std::regex_constants::icase);
3074 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3081 // Count number of characters in string
3085 // \\[a-zA-Z]+ ==> 1
3087 static int computeSize(QStringRef s, int len)
3088 #define isLyxAlpha(arg) arg.isLetter()
3090 static int computeSize(string s, int len)
3091 #define isLyxAlpha(arg) isalpha(arg)
3098 for (int i = 0; i < len; i += skip, count++) {
3099 if (s.at(i) == '\\') {
3101 if (i + 1 < len && isLyxAlpha(s.at(i+1))) {
3102 for (int j = 2; i+j < len; j++) {
3103 if (! isLyxAlpha(s.at(i+j))) {
3104 if (s.at(i+j) == ' ')
3106 else if (s.at(i+j) == '{') {
3107 if (i+j+1 < len && s.at(i+j+1) == '}')
3109 else if (i + j + 1 >= len)
3118 else if (s.at(i) == '{') {
3119 if (i + 1 < len && s.at(i+1) == '}')
3124 else if (s.at(i) == '-') {
3125 if (i+1 < len && s.at(i+1) == '-') {
3126 if (i + 2 < len && s.at(i+2) == '-')
3142 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3147 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3150 docstring docstr = stringifyFromForSearch(opt, cur, len);
3152 if (use_regexp || opt.casesensitive)
3153 str = normalize(docstr, true);
3155 str = normalize(lowercase(docstr), true);
3156 if (!opt.ignoreformat) {
3157 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3160 mres.match_len = -1;
3163 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
3164 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
3167 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3169 QString qstr = QString::fromStdString(str);
3170 QRegularExpression const *p_regexp;
3171 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3175 p_regexp = ®exp2;
3177 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3178 if (!match.hasMatch())
3181 regex const *p_regexp;
3182 regex_constants::match_flag_type flags;
3184 flags = regex_constants::match_continuous;
3187 flags = regex_constants::match_default;
3188 p_regexp = ®exp2;
3190 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3191 if (re_it == sregex_iterator())
3193 match_results<string::const_iterator> const & m = *re_it;
3195 // Whole found string, including the leading
3196 // std: m[0].second - m[0].first
3197 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3199 // Size of the leading string
3200 // std: m[1].second - m[1].first
3201 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3202 int leadingsize = 0;
3204 if (match.lastCapturedIndex() > 0) {
3205 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3210 leadingsize = m[1].second - m[1].first;
3214 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3215 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3216 // because of different number of closing at end of string
3217 // we have to 'unify' the length of the post-match.
3218 // Done by ignoring closing parenthesis and linefeeds at string end
3219 int matchend = match.capturedEnd(0);
3220 while (mres.match_len > 0) {
3221 QChar c = qstr.at(matchend - 1);
3222 if ((c == '\n') || (c == '}') || (c == '{')) {
3229 size_t strsize = qstr.size();
3230 while (strsize > (size_t) match.capturedEnd(0)) {
3231 QChar c = qstr.at(strsize-1);
3232 if ((c == '\n') || (c == '}')) {
3238 // LYXERR0(qstr.toStdString());
3239 mres.match2end = strsize - matchend;
3240 mres.pos = match.capturedStart(2);
3242 mres.match_prefix = m[2].second - m[2].first;
3243 mres.match_len = m[0].second - m[2].second;
3244 // ignore closing parenthesis and linefeeds at string end
3245 size_t strend = m[0].second - m[0].first;
3246 int matchend = strend;
3247 while (mres.match_len > 0) {
3248 char c = str.at(matchend - 1);
3249 if ((c == '\n') || (c == '}') || (c == '{')) {
3256 size_t strsize = str.size();
3257 while (strsize > strend) {
3258 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3265 mres.match2end = strsize - matchend;
3266 mres.pos = m[2].first - m[0].first;;
3268 if (mres.match2end < 0)
3270 mres.leadsize = leadingsize;
3271 #if defined(ResultsDebug)
3273 if (mres.match_len > 0) {
3274 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3275 mres.result.push_back(a0);
3276 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3277 mres.result.push_back(match.captured(i).toStdString());
3281 if (mres.match_len > 0) {
3282 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3283 mres.result.push_back(a0);
3284 for (size_t i = 3; i < m.size(); i++) {
3285 mres.result.push_back(m[i]);
3293 // else !use_regexp: but all code paths above return
3294 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
3295 << par_as_string << "', str='" << str << "'");
3296 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
3297 << lead_as_string << "', par_as_string_nolead='"
3298 << par_as_string_nolead << "'");
3301 LYXERR(Debug::FIND, "size=" << par_as_string.size()
3302 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
3303 if (str.substr(0, par_as_string.size()) == par_as_string) {
3304 mres.match_len = par_as_string.size();
3305 mres.match2end = str.size();
3310 // Start the search _after_ the leading part
3311 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3312 if (pos != string::npos) {
3313 mres.match_len = par_as_string.size();
3314 mres.match2end = str.size() - pos;
3323 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3325 MatchResult mres = findAux(cur, len, at_begin);
3326 int res = mres.match_len;
3328 "res=" << res << ", at_begin=" << at_begin
3329 << ", matchword=" << opt.matchword
3330 << ", inTexted=" << cur.inTexted());
3331 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3333 if ((len > 0) && (res < len)) {
3337 Paragraph const & par = cur.paragraph();
3338 bool ws_left = (cur.pos() > 0)
3339 ? par.isWordSeparator(cur.pos() - 1)
3345 ws_right = (cur.pos() + len < par.size())
3346 ? par.isWordSeparator(cur.pos() + len)
3350 "cur.pos()=" << cur.pos() << ", res=" << res
3351 << ", separ: " << ws_left << ", " << ws_right
3354 if (ws_left && ws_right) {
3355 // Check for word separators inside the found 'word'
3356 for (int i = 0; i < len; i++) {
3357 if (par.isWordSeparator(cur.pos() + i)) {
3368 static bool simple_replace(string &t, string from, string to)
3370 regex repl("(\\\\)*(" + from + ")");
3374 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3376 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3378 if (lastpos < (size_t) sub.position(2))
3379 s += t.substr(lastpos, sub.position(2) - lastpos);
3381 lastpos = sub.position(2) + sub.length(2);
3385 else if (lastpos < t.length())
3386 s += t.substr(lastpos, t.length() - lastpos);
3391 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
3394 t = lyx::to_utf8(s);
3395 // Remove \n at begin
3396 while (!t.empty() && t[0] == '\n')
3399 while (!t.empty() && t[t.size() - 1] == '\n')
3400 t = t.substr(0, t.size() - 1);
3402 // Handle all other '\n'
3403 while ((pos = t.find("\n")) != string::npos) {
3404 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3406 if (isAlnumASCII(t[pos+1])) {
3407 t.replace(pos-2, 3, " ");
3410 t.replace(pos-2, 3, "");
3413 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3414 // '\n' adjacent to non-alpha-numerics, discard
3415 t.replace(pos, 1, "");
3418 // Replace all other \n with spaces
3419 t.replace(pos, 1, " ");
3422 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3423 // Kornel: Added textsl, textsf, textit, texttt and noun
3424 // + allow to seach for colored text too
3425 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3426 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3427 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3428 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3429 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3430 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3431 // FIXME - check what preceeds the brace
3433 if (opt.ignoreformat)
3434 while (regex_replace(t, t, "\\{", "_x_<")
3435 || regex_replace(t, t, "\\}", "_x_>"))
3436 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3438 simple_replace(t, "\\\\\\{", "_x_<");
3439 simple_replace(t, "\\\\\\}", "_x_>");
3440 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3448 docstring stringifyFromCursor(DocIterator const & cur, int len)
3450 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3451 if (cur.inTexted()) {
3452 Paragraph const & par = cur.paragraph();
3453 // TODO what about searching beyond/across paragraph breaks ?
3454 // TODO Try adding a AS_STR_INSERTS as last arg
3455 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3456 int(par.size()) : cur.pos() + len;
3457 // OutputParams runparams(&cur.buffer()->params().encoding());
3458 OutputParams runparams(encodings.fromLyXName("utf8"));
3459 runparams.nice = true;
3460 runparams.flavor = Flavor::XeTeX;
3461 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3462 // No side effect of file copying and image conversion
3463 runparams.dryrun = true;
3464 runparams.for_search = true;
3465 LYXERR(Debug::FIND, "Stringifying with cur: "
3466 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3467 return par.asString(cur.pos(), end,
3468 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3470 } else if (cur.inMathed()) {
3471 CursorSlice cs = cur.top();
3472 MathData md = cs.cell();
3473 MathData::const_iterator it_end =
3474 (( len == -1 || cs.pos() + len > int(md.size()))
3476 : md.begin() + cs.pos() + len );
3478 for (MathData::const_iterator it = md.begin() + cs.pos();
3481 docstring s = asString(md2);
3482 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3485 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3490 /** Computes the LaTeX export of buf starting from cur and ending len positions
3491 * after cur, if len is positive, or at the paragraph or innermost inset end
3494 docstring latexifyFromCursor(DocIterator const & cur, int len)
3496 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3497 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3498 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3499 Buffer const & buf = *cur.buffer();
3501 odocstringstream ods;
3503 //OutputParams runparams(&buf.params().encoding());
3504 OutputParams runparams(encodings.fromLyXName("utf8"));
3505 runparams.nice = false;
3506 runparams.flavor = Flavor::XeTeX;
3507 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3508 // No side effect of file copying and image conversion
3509 runparams.dryrun = true;
3510 runparams.for_search = true;
3512 if (cur.inTexted()) {
3513 // @TODO what about searching beyond/across paragraph breaks ?
3514 pos_type endpos = cur.paragraph().size();
3515 if (len != -1 && endpos > cur.pos() + len)
3516 endpos = cur.pos() + len;
3517 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3518 string(), cur.pos(), endpos);
3519 string s = lyx::to_utf8(ods.str());
3520 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3521 return(lyx::from_utf8(s));
3522 } else if (cur.inMathed()) {
3523 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3524 for (int s = cur.depth() - 1; s >= 0; --s) {
3525 CursorSlice const & cs = cur[s];
3526 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3527 TeXMathStream ws(os);
3528 cs.asInsetMath()->asHullInset()->header_write(ws);
3533 CursorSlice const & cs = cur.top();
3534 MathData md = cs.cell();
3535 MathData::const_iterator it_end =
3536 ((len == -1 || cs.pos() + len > int(md.size()))
3538 : md.begin() + cs.pos() + len);
3540 for (MathData::const_iterator it = md.begin() + cs.pos();
3544 ods << asString(md2);
3545 // Retrieve the math environment type, and add '$' or '$]'
3546 // or others (\end{equation}) accordingly
3547 for (int s = cur.depth() - 1; s >= 0; --s) {
3548 CursorSlice const & cs2 = cur[s];
3549 InsetMath * inset = cs2.asInsetMath();
3550 if (inset && inset->asHullInset()) {
3551 TeXMathStream ws(os);
3552 inset->asHullInset()->footer_write(ws);
3556 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3558 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3563 #if defined(ResultsDebug)
3565 static void displayMResult(MatchResult &mres, int increment)
3567 LYXERR0( "pos: " << mres.pos << " increment " << increment);
3568 LYXERR0( "leadsize: " << mres.leadsize);
3569 LYXERR0( "match_len: " << mres.match_len);
3570 LYXERR0( "match_prefix: " << mres.match_prefix);
3571 LYXERR0( "match2end: " << mres.match2end);
3572 for (size_t i = 0; i < mres.result.size(); i++)
3573 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3575 #define displayMres(s,i) displayMResult(s,i);
3577 #define displayMres(s,i)
3580 static bool findAdvForwardInnermost(DocIterator & cur)
3583 DocIterator old_cur(cur.buffer());
3584 int forwardCount = 0;
3592 if (cur.depth() > d) {
3596 if (cur.depth() == d)
3600 if (forwardCount > 0) {
3601 LYXERR(Debug::FIND, "Forwarded " << forwardCount << " step(s) (searching for innermost match)");
3608 /** Finalize an advanced find operation, advancing the cursor to the innermost
3609 ** position that matches, plus computing the length of the matching text to
3612 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, int expected_len, int prefix_len = 0)
3614 // Search the foremost position that matches (avoids find of entire math
3615 // inset when match at start of it)
3616 DocIterator old_cur(cur.buffer());
3619 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3620 // Happens with e.g. hyperlinks
3621 // either one sees "http://www.bla.bla" or nothing
3622 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3623 // and although we search for only 3 chars, we find the whole hyperlink inset
3624 bool at_begin = (prefix_len == 0);
3625 if (findAdvForwardInnermost(cur)) {
3626 mres = match(cur, -1, at_begin);
3627 displayMres(mres, 0);
3628 if (expected_len > 0) {
3629 if (mres.match_len < expected_len)
3633 if (mres.match_len <= 0)
3636 max_match = mres.match_len;
3638 else if (expected_len < 0) {
3639 mres = match(cur); /* match valid only if not searching whole words */
3640 displayMres(mres, 0);
3641 max_match = mres.match_len;
3644 max_match = expected_len;
3646 if (max_match <= 0) return 0;
3647 LYXERR(Debug::FIND, "Ok");
3649 // Compute the match length
3651 if (cur.pos() + len > cur.lastpos())
3653 // regexp should use \w+, \S+, or \b(some string)\b
3654 // to search for whole words
3655 if (match.opt.matchword && !match.use_regexp) {
3656 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3657 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3659 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3661 // Length of matched text (different from len param)
3662 int old_match = match(cur, len, at_begin).match_len;
3666 // Greedy behaviour while matching regexps
3667 while ((new_match = match(cur, len + 1, at_begin).match_len) > old_match) {
3669 old_match = new_match;
3670 LYXERR(Debug::FIND, "verifying match with len = " << len);
3677 int maxl = cur.lastpos() - cur.pos();
3678 // Greedy behaviour while matching regexps
3679 while (maxl > minl) {
3681 mres2 = match(cur, len, at_begin);
3682 displayMres(mres2, len);
3683 int actual_match = mres2.match_len;
3684 if (actual_match >= max_match) {
3685 // actual_match > max_match _can_ happen,
3686 // if the search area splits
3687 // some following word so that the regex
3688 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3690 // This means, the len value is too big
3692 if (maxl - minl < 4)
3693 len = (int)((maxl + minl)/2);
3695 len = (int)(minl + (maxl - minl + 3)/4);
3698 // (actual_match < max_match)
3700 len = (int)((maxl + minl)/2);
3704 // Search for real start of matched characters
3709 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3710 if (cur.depth() < old_cur.depth()) {
3712 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3715 if (cur.pos() != old_cur.pos()) {
3716 // OK, forwarded 1 pos in actual inset
3717 actual_match = match(cur, len-1, at_begin).match_len;
3718 if (actual_match == max_match) {
3719 // Ha, got it! The shorter selection has the same match length
3724 // OK, the shorter selection matches less chars, revert to previous value
3730 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3731 actual_match = match(cur, len, at_begin).match_len;
3732 if (actual_match == max_match)
3741 int findForwardAdv(DocIterator & cur, MatchStringAdv const & match)
3745 while (!theApp()->longOperationCancelled() && cur) {
3746 (void) findAdvForwardInnermost(cur);
3747 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3748 MatchResult mres = match(cur, -1, false);
3749 displayMres(mres,-1)
3750 int match_len = mres.match_len;
3751 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3752 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3755 if (match_len > 0) {
3756 // Try to find the begin of searched string
3758 int firstInvalid = 100000;
3759 if (mres.match_prefix + mres.pos - mres.leadsize > 1)
3760 increment = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3763 LYXERR(Debug::FIND, "Set increment to " << increment);
3764 while (increment > 0) {
3765 DocIterator old_cur = cur;
3766 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3768 if (! cur || (cur.pit() > old_cur.pit())) {
3769 // Are we outside of the paragraph?
3770 // This can happen if moving past some UTF8-encoded chars
3775 MatchResult mres2 = match(cur, -1, false);
3776 displayMres(mres2,increment)
3777 switch (interpretMatch(mres, mres2)) {
3778 case MatchResult::newIsTooFar:
3779 // behind the expected match
3780 firstInvalid = increment;
3784 case MatchResult::newIsBetter:
3785 // not reached ye, but cur.pos()+increment is bettert
3787 firstInvalid -= increment;
3788 if (increment > firstInvalid*3/4)
3789 increment = firstInvalid*3/4;
3790 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3791 if (increment >= mres2.match_prefix)
3792 increment = (mres2.match_prefix+1)*3/4;
3797 // Handle not like MatchResult::newIsTooFar
3798 // LYXERR0( "Something is wrong: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3800 increment = increment*3/4;
3806 // LYXERR0("Leaving first loop");
3808 LYXERR(Debug::FIND, "Finalizing 1");
3809 int len = findAdvFinalize(cur, match, mres.match_len, mres.match_prefix);
3813 // try next possible match
3818 // The following code is newer reached
3819 // but parts of it may be needed in future
3820 int match_len_zero_count = 0;
3822 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3824 mres3 = match(cur, -1, false);
3825 displayMres(mres3, 1)
3826 int remaining_len = mres3.match_len;
3827 if (remaining_len <= 0) {
3828 // Apparently the searched string is not in the remaining part
3835 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3836 mres3 = match(cur, 1);
3837 displayMres(mres3, 1)
3838 int match_len3 = mres3.match_len;
3842 displayMres(mres3, 1)
3843 int match_len2 = mres3.match_len;
3844 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3845 if (match_len2 > 0) {
3846 // Sometimes in finalize we understand it wasn't a match
3847 // and we need to continue the outest loop
3848 LYXERR(Debug::FIND, "Finalizing 2");
3849 int len = findAdvFinalize(cur, match, mres.match_len);
3855 match_len_zero_count = 0;
3856 else if (match_len2 == 0)
3857 match_len_zero_count++;
3858 if (match_len2 < 0) {
3859 if (++match_len_zero_count > 3) {
3860 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3868 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3869 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3872 // This should exit nested insets, if any, or otherwise undefine the currsor.
3873 cur.pos() = cur.lastpos();
3874 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3882 /// Find the most backward consecutive match within same paragraph while searching backwards.
3883 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3885 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3886 DocIterator tmp_cur = cur;
3887 int len = findAdvFinalize(tmp_cur, match, -1);
3888 Inset & inset = cur.inset();
3889 for (; cur != cur_begin; cur.backwardPos()) {
3890 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3891 DocIterator new_cur = cur;
3892 new_cur.backwardPos();
3893 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3895 int new_len = findAdvFinalize(new_cur, match, -1);
3900 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3906 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3910 // Backup of original position
3911 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3912 if (cur == cur_begin)
3915 DocIterator cur_orig(cur);
3916 bool pit_changed = false;
3919 bool found_match = (match(cur, -1, false).match_len > 0);
3923 cur.pos() = cur.lastpos();
3925 cur.pos() = cur_orig.pos();
3926 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3927 DocIterator cur_prev_iter;
3929 found_match = (match(cur).match_len > 0);
3930 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3931 << found_match << ", cur: " << cur);
3933 return findMostBackwards(cur, match);
3935 // Stop if begin of document reached
3936 if (cur == cur_begin)
3938 cur_prev_iter = cur;
3942 if (cur == cur_begin)
3949 } while (!theApp()->longOperationCancelled());
3957 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3958 DocIterator const & cur, int len)
3960 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3962 if (!opt.ignoreformat)
3963 return latexifyFromCursor(cur, len);
3965 return stringifyFromCursor(cur, len);
3969 FindAndReplaceOptions::FindAndReplaceOptions(
3970 docstring const & _find_buf_name, bool _casesensitive,
3971 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3972 docstring const & _repl_buf_name, bool _keep_case,
3973 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3974 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3975 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3976 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3984 /** Check if 'len' letters following cursor are all non-lowercase */
3985 static bool allNonLowercase(Cursor const & cur, int len)
3987 pos_type beg_pos = cur.selectionBegin().pos();
3988 pos_type end_pos = cur.selectionBegin().pos() + len;
3989 if (len > cur.lastpos() + 1 - beg_pos) {
3990 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3991 len = cur.lastpos() + 1 - beg_pos;
3992 end_pos = beg_pos + len;
3994 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3995 if (isLowerCase(cur.paragraph().getChar(pos)))
4001 /** Check if first letter is upper case and second one is lower case */
4002 static bool firstUppercase(Cursor const & cur)
4005 pos_type pos = cur.selectionBegin().pos();
4006 if (pos >= cur.lastpos() - 1) {
4007 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
4010 ch1 = cur.paragraph().getChar(pos);
4011 ch2 = cur.paragraph().getChar(pos + 1);
4012 bool result = isUpperCase(ch1) && isLowerCase(ch2);
4013 LYXERR(Debug::FIND, "firstUppercase(): "
4014 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
4015 << ch2 << "(" << char(ch2) << ")"
4016 << ", result=" << result << ", cur=" << cur);
4021 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
4023 ** \fixme What to do with possible further paragraphs in replace buffer ?
4025 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
4027 ParagraphList::iterator pit = buffer.paragraphs().begin();
4028 LASSERT(!pit->empty(), /**/);
4029 pos_type right = pos_type(1);
4030 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
4031 right = pit->size();
4032 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4038 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4040 Cursor & cur = bv->cursor();
4041 if (opt.repl_buf_name.empty()
4042 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4043 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4046 DocIterator sel_beg = cur.selectionBegin();
4047 DocIterator sel_end = cur.selectionEnd();
4048 if (&sel_beg.inset() != &sel_end.inset()
4049 || sel_beg.pit() != sel_end.pit()
4050 || sel_beg.idx() != sel_end.idx())
4052 int sel_len = sel_end.pos() - sel_beg.pos();
4053 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4054 << ", sel_len: " << sel_len << endl);
4057 LASSERT(sel_len > 0, return 0);
4059 if (!matchAdv(sel_beg, sel_len).match_len)
4062 // Build a copy of the replace buffer, adapted to the KeepCase option
4063 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4065 repl_buffer_orig.write(oss);
4066 string lyx = oss.str();
4067 Buffer repl_buffer("", false);
4068 repl_buffer.setUnnamed(true);
4069 LASSERT(repl_buffer.readString(lyx), return 0);
4070 if (opt.keep_case && sel_len >= 2) {
4071 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4072 if (cur.inTexted()) {
4073 if (firstUppercase(cur))
4074 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4075 else if (allNonLowercase(cur, sel_len))
4076 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4079 cap::cutSelection(cur, false);
4080 if (cur.inTexted()) {
4081 repl_buffer.changeLanguage(
4082 repl_buffer.language(),
4083 cur.getFont().language());
4084 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4085 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4086 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4087 repl_buffer.params().documentClassPtr(),
4088 bv->buffer().errorList("Paste"));
4089 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4090 sel_len = repl_buffer.paragraphs().begin()->size();
4091 } else if (cur.inMathed()) {
4092 odocstringstream ods;
4094 // OutputParams runparams(&repl_buffer.params().encoding());
4095 OutputParams runparams(encodings.fromLyXName("utf8"));
4096 runparams.nice = false;
4097 runparams.flavor = Flavor::XeTeX;
4098 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4099 runparams.dryrun = true;
4100 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4101 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4102 docstring repl_latex = ods.str();
4103 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4105 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4106 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4107 repl_latex = from_utf8(s);
4108 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4109 MathData ar(cur.buffer());
4110 asArray(repl_latex, ar, Parse::NORMAL);
4112 sel_len = ar.size();
4113 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4115 if (cur.pos() >= sel_len)
4116 cur.pos() -= sel_len;
4119 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4120 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4121 bv->processUpdateFlags(Update::Force);
4126 /// Perform a FindAdv operation.
4127 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
4132 // e.g., when invoking word-findadv from mini-buffer wither with
4133 // wrong options syntax or before ever opening advanced F&R pane
4134 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4138 MatchStringAdv matchAdv(bv->buffer(), opt);
4140 if (!matchAdv.regexIsValid) {
4141 bv->message(lyx::from_utf8(matchAdv.regexError));
4145 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4147 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4148 num_replaced += findAdvReplace(bv, opt, matchAdv);
4151 match_len = findForwardAdv(cur, matchAdv);
4153 match_len = findBackwardsAdv(cur, matchAdv);
4154 } catch (exception & ex) {
4155 bv->message(from_utf8(ex.what()));
4159 if (match_len == 0) {
4160 if (num_replaced > 0) {
4161 switch (num_replaced)
4164 bv->message(_("One match has been replaced."));
4167 bv->message(_("Two matches have been replaced."));
4170 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4176 bv->message(_("Match not found."));
4181 if (num_replaced > 0)
4182 bv->message(_("Match has been replaced."));
4184 bv->message(_("Match found."));
4186 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
4187 bv->putSelectionAt(cur, match_len, !opt.forward);
4193 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4195 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4196 << opt.casesensitive << ' '
4197 << opt.matchword << ' '
4198 << opt.forward << ' '
4199 << opt.expandmacros << ' '
4200 << opt.ignoreformat << ' '
4201 << opt.replace_all << ' '
4202 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4203 << opt.keep_case << ' '
4204 << int(opt.scope) << ' '
4207 LYXERR(Debug::FIND, "built: " << os.str());
4213 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4215 LYXERR(Debug::FIND, "parsing");
4219 while (line != "EOSS") {
4223 if (is.eof()) // Tolerate malformed request
4227 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4228 opt.find_buf_name = from_utf8(s);
4229 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4230 is.get(); // Waste space before replace string
4233 while (line != "EOSS") {
4237 if (is.eof()) // Tolerate malformed request
4241 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4242 opt.repl_buf_name = from_utf8(s);
4243 is >> opt.keep_case;
4246 opt.scope = FindAndReplaceOptions::SearchScope(i);
4248 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4250 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4251 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4252 << opt.scope << ' ' << opt.restr);