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"
54 #define USE_QT_FOR_SEARCH
55 #if defined(USE_QT_FOR_SEARCH)
56 #include <QtCore> // sets QT_VERSION
57 #if (QT_VERSION >= 0x050000)
58 #include <QRegularExpression>
68 using namespace lyx::support;
73 // Helper class for deciding what should be ignored
77 IgnoreFormats() = default;
79 bool getFamily() const { return ignoreFamily_; }
81 bool getSeries() const { return ignoreSeries_; }
83 bool getShape() const { return ignoreShape_; }
85 bool getUnderline() const { return ignoreUnderline_; }
87 bool getMarkUp() const { return ignoreMarkUp_; }
89 bool getStrikeOut() const { return ignoreStrikeOut_; }
91 bool getSectioning() const { return ignoreSectioning_; }
93 bool getFrontMatter() const { return ignoreFrontMatter_; }
95 bool getColor() const { return ignoreColor_; }
97 bool getLanguage() const { return ignoreLanguage_; }
99 void setIgnoreFormat(string const & type, bool value);
103 bool ignoreFamily_ = false;
105 bool ignoreSeries_ = false;
107 bool ignoreShape_ = false;
109 bool ignoreUnderline_ = false;
111 bool ignoreMarkUp_ = false;
113 bool ignoreStrikeOut_ = false;
115 bool ignoreSectioning_ = false;
117 bool ignoreFrontMatter_ = false;
119 bool ignoreColor_ = false;
121 bool ignoreLanguage_ = false;
125 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
127 if (type == "color") {
128 ignoreColor_ = value;
130 else if (type == "language") {
131 ignoreLanguage_ = value;
133 else if (type == "sectioning") {
134 ignoreSectioning_ = value;
135 ignoreFrontMatter_ = value;
137 else if (type == "font") {
138 ignoreSeries_ = value;
139 ignoreShape_ = value;
140 ignoreFamily_ = value;
142 else if (type == "series") {
143 ignoreSeries_ = value;
145 else if (type == "shape") {
146 ignoreShape_ = value;
148 else if (type == "family") {
149 ignoreFamily_ = value;
151 else if (type == "markup") {
152 ignoreMarkUp_ = value;
154 else if (type == "underline") {
155 ignoreUnderline_ = value;
157 else if (type == "strike") {
158 ignoreStrikeOut_ = value;
162 // The global variable that can be changed from outside
163 IgnoreFormats ignoreFormats;
166 void setIgnoreFormat(string const & type, bool value)
168 ignoreFormats.setIgnoreFormat(type, value);
174 bool parse_bool(docstring & howto)
179 howto = split(howto, var, ' ');
187 MatchString(docstring const & s, bool cs, bool mw)
188 : str(s), case_sens(cs), whole_words(mw)
191 // returns true if the specified string is at the specified position
192 // del specifies whether deleted strings in ct mode will be considered
193 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
195 return par.find(str, case_sens, whole_words, pos, del);
203 // match whole words only
208 int findForward(DocIterator & cur, MatchString const & match,
209 bool find_del = true)
211 for (; cur; cur.forwardChar())
212 if (cur.inTexted()) {
213 int len = match(cur.paragraph(), cur.pos(), find_del);
221 int findBackwards(DocIterator & cur, MatchString const & match,
222 bool find_del = true)
226 if (cur.inTexted()) {
227 int len = match(cur.paragraph(), cur.pos(), find_del);
236 bool searchAllowed(docstring const & str)
239 frontend::Alert::error(_("Search error"), _("Search string is empty"));
246 bool findOne(BufferView * bv, docstring const & searchstr,
247 bool case_sens, bool whole, bool forward,
248 bool find_del = true, bool check_wrap = false)
250 if (!searchAllowed(searchstr))
253 DocIterator cur = forward
254 ? bv->cursor().selectionEnd()
255 : bv->cursor().selectionBegin();
257 MatchString const match(searchstr, case_sens, whole);
259 int match_len = forward
260 ? findForward(cur, match, find_del)
261 : findBackwards(cur, match, find_del);
264 bv->putSelectionAt(cur, match_len, !forward);
265 else if (check_wrap) {
266 DocIterator cur_orig(bv->cursor());
269 q = _("End of file reached while searching forward.\n"
270 "Continue searching from the beginning?");
272 q = _("Beginning of file reached while searching backward.\n"
273 "Continue searching from the end?");
274 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
275 q, 0, 1, _("&Yes"), _("&No"));
276 if (wrap_answer == 0) {
278 bv->cursor().clear();
279 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
281 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
282 bv->cursor().backwardPos();
284 bv->clearSelection();
285 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
288 bv->cursor().setCursor(cur_orig);
292 return match_len > 0;
296 int replaceAll(BufferView * bv,
297 docstring const & searchstr, docstring const & replacestr,
298 bool case_sens, bool whole)
300 Buffer & buf = bv->buffer();
302 if (!searchAllowed(searchstr) || buf.isReadonly())
305 DocIterator cur_orig(bv->cursor());
307 MatchString const match(searchstr, case_sens, whole);
310 int const rsize = replacestr.size();
311 int const ssize = searchstr.size();
314 cur.setCursor(doc_iterator_begin(&buf));
315 int match_len = findForward(cur, match, false);
316 while (match_len > 0) {
317 // Backup current cursor position and font.
318 pos_type const pos = cur.pos();
319 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
321 int striked = ssize -
322 cur.paragraph().eraseChars(pos, pos + match_len,
323 buf.params().track_changes);
324 cur.paragraph().insert(pos, replacestr, font,
325 Change(buf.params().track_changes
327 : Change::UNCHANGED));
328 for (int i = 0; i < rsize + striked; ++i)
331 match_len = findForward(cur, match, false);
334 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
336 cur_orig.fixIfBroken();
337 bv->setCursor(cur_orig);
343 // the idea here is that we are going to replace the string that
344 // is selected IF it is the search string.
345 // if there is a selection, but it is not the search string, then
346 // we basically ignore it. (FIXME We ought to replace only within
348 // if there is no selection, then:
349 // (i) if some search string has been provided, then we find it.
350 // (think of how the dialog works when you hit "replace" the
352 // (ii) if no search string has been provided, then we treat the
353 // word the cursor is in as the search string. (why? i have no
354 // idea.) but this only works in text?
356 // returns the number of replacements made (one, if any) and
357 // whether anything at all was done.
358 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
359 docstring const & replacestr, bool case_sens,
360 bool whole, bool forward, bool findnext)
362 Cursor & cur = bv->cursor();
363 if (!cur.selection()) {
364 // no selection, non-empty search string: find it
365 if (!searchstr.empty()) {
366 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
367 return make_pair(found, 0);
369 // empty search string
372 return make_pair(false, 0);
373 // select current word and treat it as the search string.
374 // This causes a minor bug as undo will restore this selection,
375 // which the user did not create (#8986).
376 cur.innerText()->selectWord(cur, WHOLE_WORD);
377 searchstr = cur.selectionAsString(false);
380 // if we still don't have a search string, report the error
382 if (!searchAllowed(searchstr))
383 return make_pair(false, 0);
385 bool have_selection = cur.selection();
386 docstring const selected = cur.selectionAsString(false);
389 ? searchstr == selected
390 : compare_no_case(searchstr, selected) == 0;
392 // no selection or current selection is not search word:
393 // just find the search word
394 if (!have_selection || !match) {
395 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
396 return make_pair(found, 0);
399 // we're now actually ready to replace. if the buffer is
400 // read-only, we can't, though.
401 if (bv->buffer().isReadonly())
402 return make_pair(false, 0);
404 cap::replaceSelectionWithString(cur, replacestr);
406 cur.pos() += replacestr.length();
407 LASSERT(cur.pos() <= cur.lastpos(),
408 cur.pos() = cur.lastpos());
411 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
413 return make_pair(true, 1);
419 docstring const find2string(docstring const & search,
420 bool casesensitive, bool matchword, bool forward)
424 << int(casesensitive) << ' '
425 << int(matchword) << ' '
431 docstring const replace2string(docstring const & replace,
432 docstring const & search,
433 bool casesensitive, bool matchword,
434 bool all, bool forward, bool findnext)
437 ss << replace << '\n'
439 << int(casesensitive) << ' '
440 << int(matchword) << ' '
442 << int(forward) << ' '
448 bool lyxfind(BufferView * bv, FuncRequest const & ev)
450 if (!bv || ev.action() != LFUN_WORD_FIND)
453 //lyxerr << "find called, cmd: " << ev << endl;
455 // data is of the form
457 // <casesensitive> <matchword> <forward>"
459 docstring howto = split(ev.argument(), search, '\n');
461 bool casesensitive = parse_bool(howto);
462 bool matchword = parse_bool(howto);
463 bool forward = parse_bool(howto);
465 return findOne(bv, search, casesensitive, matchword, forward, true, true);
469 bool lyxreplace(BufferView * bv,
470 FuncRequest const & ev, bool has_deleted)
472 if (!bv || ev.action() != LFUN_WORD_REPLACE)
475 // data is of the form
478 // <casesensitive> <matchword> <all> <forward> <findnext>"
481 docstring howto = split(ev.argument(), rplc, '\n');
482 howto = split(howto, search, '\n');
484 bool casesensitive = parse_bool(howto);
485 bool matchword = parse_bool(howto);
486 bool all = parse_bool(howto);
487 bool forward = parse_bool(howto);
488 bool findnext = howto.empty() ? true : parse_bool(howto);
493 int replace_count = 0;
495 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
496 update = replace_count > 0;
499 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
501 replace_count = rv.second;
504 Buffer const & buf = bv->buffer();
506 // emit message signal.
507 buf.message(_("String not found."));
509 if (replace_count == 0) {
510 buf.message(_("String found."));
511 } else if (replace_count == 1) {
512 buf.message(_("String has been replaced."));
514 docstring const str =
515 bformat(_("%1$d strings have been replaced."), replace_count);
519 } else if (findnext) {
520 // if we have deleted characters, we do not replace at all, but
521 // rather search for the next occurence
522 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
525 bv->message(_("String not found."));
531 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
533 for (; cur; cur.forwardPos())
534 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
538 DocIterator cur_orig(bv->cursor());
539 docstring q = _("End of file reached while searching forward.\n"
540 "Continue searching from the beginning?");
541 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
542 q, 0, 1, _("&Yes"), _("&No"));
543 if (wrap_answer == 0) {
544 bv->cursor().clear();
545 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
546 bv->clearSelection();
547 cur.setCursor(bv->cursor().selectionBegin());
548 if (findNextChange(bv, cur, false))
551 bv->cursor().setCursor(cur_orig);
558 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
560 for (cur.backwardPos(); cur; cur.backwardPos()) {
561 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
566 DocIterator cur_orig(bv->cursor());
567 docstring q = _("Beginning of file reached while searching backward.\n"
568 "Continue searching from the end?");
569 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
570 q, 0, 1, _("&Yes"), _("&No"));
571 if (wrap_answer == 0) {
572 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
573 bv->cursor().backwardPos();
574 bv->clearSelection();
575 cur.setCursor(bv->cursor().selectionBegin());
576 if (findPreviousChange(bv, cur, false))
579 bv->cursor().setCursor(cur_orig);
586 bool selectChange(Cursor & cur, bool forward)
588 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
590 Change ch = cur.paragraph().lookupChange(cur.pos());
592 CursorSlice tip1 = cur.top();
593 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
594 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
595 if (!ch2.isSimilarTo(ch))
598 CursorSlice tip2 = cur.top();
599 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
601 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
602 if (!ch2.isSimilarTo(ch)) {
603 // take a step forward to correctly set the selection
611 cur.bv().mouseSetCursor(cur, false);
613 cur.bv().mouseSetCursor(cur, true);
621 bool findChange(BufferView * bv, bool forward)
624 cur.setCursor(forward ? bv->cursor().selectionEnd()
625 : bv->cursor().selectionBegin());
626 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
627 return selectChange(cur, forward);
632 bool findNextChange(BufferView * bv)
634 return findChange(bv, true);
638 bool findPreviousChange(BufferView * bv)
640 return findChange(bv, false);
647 typedef vector<pair<string, string> > Escapes;
649 /// A map of symbols and their escaped equivalent needed within a regex.
650 /// @note Beware of order
651 Escapes const & get_regexp_escapes()
653 typedef std::pair<std::string, std::string> P;
655 static Escapes escape_map;
656 if (escape_map.empty()) {
657 escape_map.push_back(P("$", "_x_$"));
658 escape_map.push_back(P("{", "_x_{"));
659 escape_map.push_back(P("}", "_x_}"));
660 escape_map.push_back(P("[", "_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("\\", "(?:\\\\|\\\\backslash)"));
668 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
669 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
670 escape_map.push_back(P("_x_", "\\"));
675 /// A map of lyx escaped strings and their unescaped equivalent.
676 Escapes const & get_lyx_unescapes()
678 typedef std::pair<std::string, std::string> P;
680 static Escapes escape_map;
681 if (escape_map.empty()) {
682 escape_map.push_back(P("\\%", "%"));
683 escape_map.push_back(P("\\{", "{"));
684 escape_map.push_back(P("\\}", "}"));
685 escape_map.push_back(P("\\mathcircumflex ", "^"));
686 escape_map.push_back(P("\\mathcircumflex", "^"));
687 escape_map.push_back(P("\\backslash ", "\\"));
688 escape_map.push_back(P("\\backslash", "\\"));
689 escape_map.push_back(P("\\sim ", "~"));
690 escape_map.push_back(P("\\sim", "~"));
695 /// A map of escapes turning a regexp matching text to one matching latex.
696 Escapes const & get_regexp_latex_escapes()
698 typedef std::pair<std::string, std::string> P;
700 static Escapes escape_map;
701 if (escape_map.empty()) {
702 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
703 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
704 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
705 escape_map.push_back(P("\\[", "\\{\\[\\}"));
706 escape_map.push_back(P("\\]", "\\{\\]\\}"));
707 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
708 escape_map.push_back(P("%", "\\\\\\%"));
709 escape_map.push_back(P("#", "\\\\#"));
714 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
715 ** the found occurrence were escaped.
717 string apply_escapes(string s, Escapes const & escape_map)
719 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
720 Escapes::const_iterator it;
721 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
722 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
723 unsigned int pos = 0;
724 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
725 s.replace(pos, it->first.length(), it->second);
726 LYXERR(Debug::FIND, "After escape: " << s);
727 pos += it->second.length();
728 // LYXERR(Debug::FIND, "pos: " << pos);
731 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
736 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
737 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
738 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
739 string escape_for_regex(string s, bool match_latex)
742 while (pos < s.size()) {
743 size_t new_pos = s.find("\\regexp{", pos);
744 if (new_pos == string::npos)
749 LYXERR(Debug::FIND, "new_pos: " << new_pos);
750 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
751 LYXERR(Debug::FIND, "t [lyx]: " << t);
752 t = apply_escapes(t, get_regexp_escapes());
753 LYXERR(Debug::FIND, "t [rxp]: " << t);
754 s.replace(pos, new_pos - pos, t);
755 new_pos = pos + t.size();
756 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
757 LYXERR(Debug::FIND, "new_pos: " << new_pos);
758 if (new_pos == s.size())
761 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
762 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
763 LYXERR(Debug::FIND, "end_pos: " << end_pos);
764 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
765 LYXERR(Debug::FIND, "t in regexp : " << t);
766 t = apply_escapes(t, get_lyx_unescapes());
767 LYXERR(Debug::FIND, "t in regexp after unescapes [lyx]: " << t);
769 t = apply_escapes(t, get_regexp_latex_escapes());
770 LYXERR(Debug::FIND, "t in regexp after latex_escapes [ltx]: " << t);
772 if (end_pos == s.size()) {
773 s.replace(new_pos, end_pos - new_pos, t);
774 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
777 s.replace(new_pos, end_pos + 13 - new_pos, t);
778 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
779 pos = new_pos + t.size();
780 LYXERR(Debug::FIND, "pos: " << pos);
786 /// Wrapper for lyx::regex_replace with simpler interface
787 bool regex_replace(string const & s, string & t, string const & searchstr,
788 string const & replacestr)
790 regex e(searchstr, regex_constants::ECMAScript);
792 ostream_iterator<char, char> it(oss);
793 regex_replace(it, s.begin(), s.end(), e, replacestr);
794 // tolerate t and s be references to the same variable
795 bool rv = (s != oss.str());
801 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
803 ** Verify that closed braces exactly match open braces. This avoids that, for example,
804 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
807 ** Number of open braces that must remain open at the end for the verification to succeed.
810 bool braces_match(QString const & beg,
813 bool braces_match(string const & beg,
819 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg.toStdString() << "'");
821 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << beg << "'");
823 int lastidx = beg.size();
824 for (int i=0; i < lastidx; ++i) {
825 // Skip escaped braces in the count
835 } else if (c == '{') {
837 } else if (c == '}') {
838 if (open_pars == 0) {
839 LYXERR(Debug::FIND, "Found unmatched closed brace");
845 if (open_pars != unmatched) {
846 LYXERR(Debug::FIND, "Found " << open_pars
847 << " instead of " << unmatched
848 << " unmatched open braces at the end of count");
851 LYXERR(Debug::FIND, "Braces match as expected");
868 MatchResult(): match_len(0),match_prefix(0),match2end(0), pos(0),leadsize(0) {};
871 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
873 if (newres.match2end < oldres.match2end)
874 return MatchResult::newIsTooFar;
875 if (newres.match_len < oldres.match_len)
876 return MatchResult::newIsTooFar;
877 if ((newres.match_len == oldres.match_len) && (newres.match2end == oldres.match2end))
878 return MatchResult::newIsBetter;
879 if ((newres.match_len == oldres.match_len) && (newres.match2end -2 == oldres.match2end)) {
880 // The string contained for instance "\usepackage...fontenc ..."
881 // and now after moved 9 char forward contains "ge...{fontenc} ..."
882 // so we accept it as OK
883 return MatchResult::newIsBetter;
885 return MatchResult::newIsInvalid;
888 /** The class performing a match between a position in the document and the FindAdvOptions.
891 class MatchStringAdv {
893 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
895 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
896 ** constructor as opt.search, under the opt.* options settings.
899 ** If set, then match is searched only against beginning of text starting at cur.
900 ** If unset, then match is searched anywhere in text starting at cur.
903 ** The length of the matching text, or zero if no match was found.
905 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
914 /// first buffer on which search was started
915 lyx::Buffer * const p_first_buf;
917 FindAndReplaceOptions const & opt;
920 /// Auxiliary find method (does not account for opt.matchword)
921 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
923 /** Normalize a stringified or latexified LyX paragraph.
927 ** <li>if search is not casesensitive, then lowercase the string;
928 ** <li>remove any newline at begin or end of the string;
929 ** <li>replace any newline in the middle of the string with a simple space;
930 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
933 ** @todo Normalization should also expand macros, if the corresponding
934 ** search option was checked.
936 string normalize(docstring const & s, bool hack_braces) const;
937 // normalized string to search
938 string par_as_string;
939 // regular expression to use for searching
940 // regexp2 is same as regexp, but prefixed with a ".*?"
942 QRegularExpression regexp;
943 QRegularExpression regexp2;
948 // leading format material as string
949 string lead_as_string;
950 // par_as_string after removal of lead_as_string
951 string par_as_string_nolead;
952 // unmatched open braces in the search string/regexp
954 // number of (.*?) subexpressions added at end of search regexp for closing
955 // environments, math mode, styles, etc...
958 // Are we searching with regular expressions ?
963 static docstring buffer_to_latex(Buffer & buffer)
965 //OutputParams runparams(&buffer.params().encoding());
966 OutputParams runparams(encodings.fromLyXName("utf8"));
967 odocstringstream ods;
969 runparams.nice = true;
970 runparams.flavor = Flavor::XeTeX;
971 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
972 // No side effect of file copying and image conversion
973 runparams.dryrun = true;
974 runparams.for_search = true;
975 pit_type const endpit = buffer.paragraphs().size();
976 for (pit_type pit = 0; pit != endpit; ++pit) {
977 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
978 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
984 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
987 if (!opt.ignoreformat) {
988 str = buffer_to_latex(buffer);
990 // OutputParams runparams(&buffer.params().encoding());
991 OutputParams runparams(encodings.fromLyXName("utf8"));
992 runparams.nice = true;
993 runparams.flavor = Flavor::XeTeX;
994 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
995 runparams.dryrun = true;
996 runparams.for_search = true;
997 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
998 Paragraph const & par = buffer.paragraphs().at(pit);
999 LYXERR(Debug::FIND, "Adding to search string: '"
1000 << par.asString(pos_type(0), par.size(),
1001 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1004 str += par.asString(pos_type(0), par.size(),
1005 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1008 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
1009 string t = to_utf8(str);
1010 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
1017 /// Return separation pos between the leading material and the rest
1018 static size_t identifyLeading(string const & s)
1021 // @TODO Support \item[text]
1022 // Kornel: Added textsl, textsf, textit, texttt and noun
1023 // + allow to search for colored text too
1024 while (regex_replace(t, t, "^\\\\(("
1025 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
1026 "lyxaddress|lyxrightaddress|"
1027 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1028 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
1029 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
1030 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
1031 || regex_replace(t, t, "^\\$", "")
1032 || regex_replace(t, t, "^\\\\\\[", "")
1033 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
1034 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
1036 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
1041 * Given a latexified string, retrieve some handled features
1042 * The features of the regex will later be compared with the features
1043 * of the searched text. If the regex features are not a
1044 * subset of the analized, then, in not format ignoring search
1045 * we can early stop the search in the relevant inset.
1047 typedef map<string, bool> Features;
1049 static Features identifyFeatures(string const & s)
1051 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
1052 static regex const valid("^("
1054 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1055 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1056 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1057 "(u|uu)line|(s|x)out|uwave|"
1058 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1059 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1064 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1067 if (sub.str(1).compare("regexp") == 0) {
1071 string token = sub.str(1);
1073 if (regex_match(token, sub2, valid)) {
1081 if (sub.str(1).compare("endregexp") == 0) {
1091 * defines values features of a key "\\[a-z]+{"
1096 /* Char type with content discarded
1097 * like \hspace{1cm} */
1099 /* Char, like \backslash */
1101 /* replace starting backslash with '#' */
1103 /* \part, \section*, ... */
1105 /* title, author etc */
1107 /* \foreignlanguage{ngerman}, ... */
1110 * to discard language in content */
1113 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1115 /* fonts, colors, markups, ... */
1117 /* footnotesize, ... large, ...
1118 * Ignore all of them */
1121 /* inputencoding, ...
1122 * Discard also content, because they do not help in search */
1125 * like remove, but also all arguments */
1127 /* item, listitem */
1129 /* tex, latex, ... like isChar */
1131 /* like \lettrine[lines=5]{}{} */
1133 // like isStandard, but always remove head
1135 /* End of arguments marker for lettrine,
1136 * so that they can be ignored */
1139 KeyInfo() = default;
1140 KeyInfo(KeyType type, int parcount, bool disable)
1142 parenthesiscount(parcount),
1143 disabled(disable) {}
1144 KeyType keytype = invalid;
1146 int _tokensize = -1;
1147 int _tokenstart = -1;
1148 int _dataStart = -1;
1150 int parenthesiscount = 1;
1151 bool disabled = false;
1152 bool used = false; /* by pattern */
1157 Border(int l=0, int u=0) : low(l), upper(u) {};
1162 #define MAXOPENED 30
1164 bool isPatternString_;
1166 explicit Intervall(bool isPattern, string const & p) :
1167 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1168 hasTitle(false), langcount(0)
1176 static vector<Border> borders;
1177 int depts[MAXOPENED];
1178 int closes[MAXOPENED];
1179 int actualdeptindex;
1180 int previousNotIgnored(int) const;
1181 int nextNotIgnored(int) const;
1182 void handleOpenP(int i);
1183 void handleCloseP(int i, bool closingAllowed);
1184 void resetOpenedP(int openPos);
1185 void addIntervall(int upper);
1186 void addIntervall(int low, int upper); /* if explicit */
1187 void removeAccents();
1188 void setForDefaultLang(KeyInfo const & defLang) const;
1189 int findclosing(int start, int end, char up, char down, int repeat);
1190 void handleParentheses(int lastpos, bool closingAllowed);
1192 int langcount; // Number of disabled language specs up to current position in actual interval
1193 int isOpeningPar(int pos) const;
1195 void output(ostringstream &os, int lastpos);
1196 // string show(int lastpos);
1199 vector<Border> Intervall::borders = vector<Border>(30);
1201 int Intervall::isOpeningPar(int pos) const
1203 if ((pos < 0) || (size_t(pos) >= par.size()))
1205 if (par[pos] != '{')
1207 if (size_t(pos) + 2 >= par.size())
1209 if (par[pos+2] != '}')
1211 if (par[pos+1] == '[' || par[pos+1] == ']')
1216 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1218 // Enable the use of first token again
1219 if (ignoreidx >= 0) {
1220 int value = defLang._tokenstart + defLang._tokensize;
1226 if (borders[borderidx].low < value)
1227 borders[borderidx].low = value;
1228 if (borders[borderidx].upper < value)
1229 borders[borderidx].upper = value;
1234 static void checkDepthIndex(int val)
1236 static int maxdepthidx = MAXOPENED-2;
1237 static int lastmaxdepth = 0;
1238 if (val > lastmaxdepth) {
1239 LYXERR(Debug::INFO, "Depth reached " << val);
1242 if (val > maxdepthidx) {
1244 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1249 // Not needed, because borders are now dynamically expanded
1250 static void checkIgnoreIdx(int val)
1252 static int lastmaxignore = -1;
1253 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1254 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1255 lastmaxignore = val;
1261 * Expand the region of ignored parts of the input latex string
1262 * The region is only relevant in output()
1264 void Intervall::addIntervall(int low, int upper)
1267 if (low == upper) return;
1268 for (idx = ignoreidx+1; idx > 0; --idx) {
1269 if (low > borders[idx-1].upper) {
1273 Border br(low, upper);
1274 if (idx > ignoreidx) {
1275 if (borders.size() <= size_t(idx)) {
1276 borders.push_back(br);
1282 // checkIgnoreIdx(ignoreidx);
1286 // Expand only if one of the new bound is inside the interwall
1287 // We know here that br.low > borders[idx-1].upper
1288 if (br.upper < borders[idx].low) {
1289 // We have to insert at this pos
1290 if (size_t(ignoreidx+1) >= borders.size()) {
1291 borders.push_back(borders[ignoreidx]);
1294 borders[ignoreidx+1] = borders[ignoreidx];
1296 for (int i = ignoreidx; i > idx; --i) {
1297 borders[i] = borders[i-1];
1301 // checkIgnoreIdx(ignoreidx);
1304 // Here we know, that we are overlapping
1305 if (br.low > borders[idx].low)
1306 br.low = borders[idx].low;
1307 // check what has to be concatenated
1309 for (int i = idx; i <= ignoreidx; i++) {
1310 if (br.upper >= borders[i].low) {
1312 if (br.upper < borders[i].upper)
1313 br.upper = borders[i].upper;
1319 // count should be >= 1 here
1322 for (int i = idx + count; i <= ignoreidx; i++) {
1323 borders[i-count+1] = borders[i];
1325 ignoreidx -= count - 1;
1331 typedef map<string, string> AccentsMap;
1332 static AccentsMap accents = map<string, string>();
1334 static void buildaccent(string n, string param, string values)
1338 const char delim = '|';
1339 while (getline(s, name, delim)) {
1341 for (char c : param) {
1342 string key = name + "{" + c + "}";
1343 // get the corresponding utf8-value
1344 if ((values[start] & 0xc0) != 0xc0) {
1345 // should not happen, utf8 encoding starts at least with 11xxxxxx
1346 // but value for '\dot{i}' is 'i', which is ascii
1347 if ((values[start] & 0x80) == 0) {
1349 accents[key] = values.substr(start, 1);
1350 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1355 for (int j = 1; ;j++) {
1356 if (start + j >= values.size()) {
1357 accents[key] = values.substr(start, j);
1358 start = values.size() - 1;
1361 else if ((values[start+j] & 0xc0) != 0x80) {
1362 // This is the first byte of following utf8 char
1363 accents[key] = values.substr(start, j);
1365 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1374 static string getutf8(unsigned uchar)
1377 string ret = string();
1379 if (uchar <= 0x7f) {
1380 c[maxc-1] = uchar & 0x7f;
1383 unsigned char rest = 0x40;
1384 unsigned char first = 0x80;
1386 for (int i = start; i >=0; --i) {
1388 c[i] = first + uchar;
1391 c[i] = 0x80 | (uchar & 0x3f);
1398 for (int i = 0; i < maxc; i++) {
1399 if (c[i] == 0) continue;
1405 static void buildAccentsMap()
1407 accents["imath"] = "ı";
1409 accents["jmath"] = "ȷ";
1410 accents["cdot"] = "·";
1411 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1412 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1413 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1414 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1415 accents["negmedspace"] = getutf8(0xf0005);
1416 accents["thickspace"] = getutf8(0xf0006);
1417 accents["negthickspace"] = getutf8(0xf0007);
1418 accents["lyx"] = getutf8(0xf0010); // Used logos
1419 accents["LyX"] = getutf8(0xf0010);
1420 accents["tex"] = getutf8(0xf0011);
1421 accents["TeX"] = getutf8(0xf0011);
1422 accents["latex"] = getutf8(0xf0012);
1423 accents["LaTeX"] = getutf8(0xf0012);
1424 accents["latexe"] = getutf8(0xf0013);
1425 accents["LaTeXe"] = getutf8(0xf0013);
1426 accents["lyxarrow"] = getutf8(0xf0020);
1427 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1428 accents["backslash LyX"] = getutf8(0xf0010);
1429 accents["backslash tex"] = getutf8(0xf0011);
1430 accents["backslash TeX"] = getutf8(0xf0011);
1431 accents["backslash latex"] = getutf8(0xf0012);
1432 accents["backslash LaTeX"] = getutf8(0xf0012);
1433 accents["backslash latexe"] = getutf8(0xf0013);
1434 accents["backslash LaTeXe"] = getutf8(0xf0013);
1435 accents["backslash lyxarrow"] = getutf8(0xf0020);
1436 accents["ddot{\\imath}"] = "ï";
1437 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1438 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1439 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1440 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1441 accents["acute{\\imath}"] = "í";
1442 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1443 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1444 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1445 buildaccent("mathring|r", "aAuUwy",
1447 accents["check{\\imath}"] = "ǐ";
1448 accents["check{\\jmath}"] = "ǰ";
1449 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1450 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1451 accents["hat{\\imath}"] = "î";
1452 accents["hat{\\jmath}"] = "ĵ";
1453 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1454 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1455 accents["bar{\\imath}"] = "ī";
1456 buildaccent("bar|=", "aAeEiIoOuUyY",
1457 "āĀēĒīĪōŌūŪȳȲ"); // macron
1458 accents["tilde{\\imath}"] = "ĩ";
1459 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1460 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1461 accents["breve{\\imath}"] = "ĭ";
1462 buildaccent("breve|u", "aAeEgGiIoOuU",
1463 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1464 accents["grave{\\imath}"] = "ì";
1465 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1466 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1467 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1468 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1469 buildaccent("ogonek|k", "AaEeIiUuOo",
1470 "ĄąĘęĮįŲųǪǫ"); // ogonek
1471 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1472 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1473 buildaccent("subring|textsubring", "Aa",
1475 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1476 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1477 buildaccent("subtilde|textsubtilde", "EeIiUu",
1478 "ḚḛḬḭṴṵ"); // subtilde
1479 accents["dgrave{\\imath}"] = "ȉ";
1480 accents["textdoublegrave{\\i}"] = "ȉ";
1481 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1482 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1483 accents["rcap{\\imath}"] = "ȉ";
1484 accents["textroundcap{\\i}"] = "ȉ";
1485 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1486 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1487 buildaccent("slashed", "oO",
1492 * Created accents in math or regexp environment
1493 * are macros, but we need the utf8 equivalent
1495 void Intervall::removeAccents()
1497 if (accents.empty())
1499 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1500 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1501 "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow)))(?![a-zA-Z]))");
1503 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1505 string key = sub.str(1);
1506 if (accents.find(key) != accents.end()) {
1507 string val = accents[key];
1508 size_t pos = sub.position(size_t(0));
1509 for (size_t i = 0; i < val.size(); i++) {
1510 par[pos+i] = val[i];
1512 // Remove possibly following space too
1513 if (par[pos+sub.str(0).size()] == ' ')
1514 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1516 addIntervall(pos+val.size(), pos + sub.str(0).size());
1517 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1518 // remove traces of any remaining chars
1523 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1528 void Intervall::handleOpenP(int i)
1531 depts[actualdeptindex] = i+1;
1532 closes[actualdeptindex] = -1;
1533 checkDepthIndex(actualdeptindex);
1536 void Intervall::handleCloseP(int i, bool closingAllowed)
1538 if (actualdeptindex <= 0) {
1539 if (! closingAllowed)
1540 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1541 // if we are at the very end
1542 addIntervall(i, i+1);
1545 closes[actualdeptindex] = i+1;
1550 void Intervall::resetOpenedP(int openPos)
1552 // Used as initializer for foreignlanguage entry
1553 actualdeptindex = 1;
1554 depts[1] = openPos+1;
1558 int Intervall::previousNotIgnored(int start) const
1560 int idx = 0; /* int intervalls */
1561 for (idx = ignoreidx; idx >= 0; --idx) {
1562 if (start > borders[idx].upper)
1564 if (start >= borders[idx].low)
1565 start = borders[idx].low-1;
1570 int Intervall::nextNotIgnored(int start) const
1572 int idx = 0; /* int intervalls */
1573 for (idx = 0; idx <= ignoreidx; idx++) {
1574 if (start < borders[idx].low)
1576 if (start < borders[idx].upper)
1577 start = borders[idx].upper;
1582 typedef map<string, KeyInfo> KeysMap;
1583 typedef vector< KeyInfo> Entries;
1584 static KeysMap keys = map<string, KeyInfo>();
1590 Intervall interval_;
1591 void buildKeys(bool);
1592 void buildEntries(bool);
1593 void makeKey(const string &, KeyInfo, bool isPatternString);
1594 void processRegion(int start, int region_end); /* remove {} parts */
1595 void removeHead(KeyInfo const &, int count=0);
1598 LatexInfo(string const & par, bool isPatternString)
1599 : entidx_(-1), interval_(isPatternString, par)
1601 buildKeys(isPatternString);
1602 entries_ = vector<KeyInfo>();
1603 buildEntries(isPatternString);
1607 if (entries_.empty()) {
1610 if (entries_[0].keytype == KeyInfo::isTitle) {
1611 interval_.hasTitle = true;
1612 if (! entries_[0].disabled) {
1613 interval_.titleValue = entries_[0].head;
1616 interval_.titleValue = "";
1618 removeHead(entries_[0]);
1619 if (entries_.size() > 1)
1628 if (int(entries_.size()) > entidx_) {
1635 bool setNextKey(int idx) {
1636 if ((idx == entidx_) && (entidx_ >= 0)) {
1643 int find(int start, KeyInfo::KeyType keytype) const {
1647 while (tmpIdx < int(entries_.size())) {
1648 if (entries_[tmpIdx].keytype == keytype)
1654 int process(ostringstream & os, KeyInfo const & actual);
1655 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1656 // string show(int lastpos) { return interval.show(lastpos);};
1657 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1658 KeyInfo &getKeyInfo(int keyinfo) {
1659 static KeyInfo invalidInfo = KeyInfo();
1660 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1663 return entries_[keyinfo];
1665 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1666 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1670 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1674 for (int i = start; i < end; i += 1 + skip) {
1678 if (c == '\\') skip = 1;
1682 else if (c == down) {
1685 if ((repeat <= 0) || (par[i+1] != up))
1703 vector<MathEntry> entries_;
1708 void insert(string const & wait, size_t start, size_t end) {
1709 MathEntry m = MathEntry();
1711 m.mathStart = start;
1713 m.mathSize = end - start;
1714 entries_.push_back(m);
1716 bool empty() const { return entries_.empty(); };
1717 size_t getEndPos() const {
1718 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1721 return entries_[actualIdx_].mathEnd;
1723 size_t getStartPos() const {
1724 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1725 return 100000; /* definitely enough? */
1727 return entries_[actualIdx_].mathStart;
1729 size_t getFirstPos() {
1731 return getStartPos();
1733 size_t getSize() const {
1734 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1737 return entries_[actualIdx_].mathSize;
1739 void incrEntry() { actualIdx_++; };
1742 void LatexInfo::buildEntries(bool isPatternString)
1744 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1745 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1746 static bool disableLanguageOverride = false;
1747 smatch sub, submath;
1748 bool evaluatingRegexp = false;
1750 bool evaluatingMath = false;
1751 bool evaluatingCode = false;
1753 bool evaluatingOptional = false;
1754 size_t optionalEnd = 0;
1757 bool math_end_waiting = false;
1758 size_t math_pos = 10000;
1760 static vector<string> usedText = vector<string>();
1762 interval_.removeAccents();
1764 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1766 if (math_end_waiting) {
1767 size_t pos = submath.position(size_t(0));
1768 if ((math_end == "$") &&
1769 (submath.str(0) == "$") &&
1770 (interval_.par[pos-1] != '\\')) {
1771 mi.insert("$", math_pos, pos + 1);
1772 math_end_waiting = false;
1774 else if ((math_end == "\\]") &&
1775 (submath.str(0) == "\\]")) {
1776 mi.insert("\\]", math_pos, pos + 2);
1777 math_end_waiting = false;
1779 else if ((submath.str(1).compare("end") == 0) &&
1780 (submath.str(2).compare(math_end) == 0)) {
1781 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1782 math_end_waiting = false;
1788 if (submath.str(1).compare("begin") == 0) {
1789 math_end_waiting = true;
1790 math_end = submath.str(2);
1791 math_pos = submath.position(size_t(0));
1793 else if (submath.str(0).compare("\\[") == 0) {
1794 math_end_waiting = true;
1796 math_pos = submath.position(size_t(0));
1798 else if (submath.str(0) == "$") {
1799 size_t pos = submath.position(size_t(0));
1800 if ((pos == 0) || (interval_.par[pos-1] != '\\')) {
1801 math_end_waiting = true;
1808 // Ignore language if there is math somewhere in pattern-string
1809 if (isPatternString) {
1810 for (auto s: usedText) {
1811 // Remove entries created in previous search runs
1814 usedText = vector<string>();
1817 keys["foreignlanguage"].disabled = true;
1818 disableLanguageOverride = true;
1821 disableLanguageOverride = false;
1824 if (disableLanguageOverride) {
1825 keys["foreignlanguage"].disabled = true;
1828 math_pos = mi.getFirstPos();
1829 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1831 string key = sub.str(3);
1833 if (sub.str(0)[0] == '\\')
1834 key = sub.str(0)[1];
1838 size_t k_pos = sub.position(size_t(0));
1839 if ((k_pos > 0) && (interval_.par[k_pos - 1] == '\\')) {
1840 // Escaped '$', ignoring
1846 if (keys.find(key) != keys.end()) {
1847 if (keys[key].keytype == KeyInfo::headRemove) {
1848 KeyInfo found1 = keys[key];
1849 found1.disabled = true;
1850 found1.head = "\\" + key + "{";
1851 found1._tokenstart = sub.position(size_t(0));
1852 found1._tokensize = found1.head.length();
1853 found1._dataStart = found1._tokenstart + found1.head.length();
1854 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1855 found1._dataEnd = endpos;
1860 if (evaluatingRegexp) {
1861 if (sub.str(1).compare("endregexp") == 0) {
1862 evaluatingRegexp = false;
1863 // found._tokenstart already set
1864 found._dataEnd = sub.position(size_t(0)) + 13;
1865 found._dataStart = found._dataEnd;
1866 found._tokensize = found._dataEnd - found._tokenstart;
1867 found.parenthesiscount = 0;
1868 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1875 if (evaluatingMath) {
1876 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1878 evaluatingMath = false;
1880 math_pos = mi.getStartPos();
1882 if (keys.find(key) == keys.end()) {
1883 found = KeyInfo(KeyInfo::isStandard, 0, true);
1884 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1885 found = KeyInfo(KeyInfo::isText, 0, false);
1886 if (isPatternString) {
1887 found.keytype = KeyInfo::isChar;
1888 found.disabled = false;
1892 usedText.push_back(key);
1896 if (key.compare("regexp") == 0) {
1897 evaluatingRegexp = true;
1898 found._tokenstart = sub.position(size_t(0));
1899 found._tokensize = 0;
1903 // Handle the other params of key
1904 if (found.keytype == KeyInfo::isIgnored)
1906 else if (found.keytype == KeyInfo::isMath) {
1907 if (size_t(sub.position(size_t(0))) == math_pos) {
1909 found._tokenstart = sub.position(size_t(0));
1910 found._tokensize = mi.getSize();
1911 found._dataEnd = found._tokenstart + found._tokensize;
1912 found._dataStart = found._dataEnd;
1913 found.parenthesiscount = 0;
1914 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1915 evaluatingMath = true;
1918 // begin|end of unknown env, discard
1919 // First handle tables
1920 // longtable|tabular
1921 bool discardComment;
1923 found.keytype = KeyInfo::doRemove;
1924 if ((sub.str(5).compare("longtable") == 0) ||
1925 (sub.str(5).compare("tabular") == 0)) {
1926 discardComment = true; /* '%' */
1929 discardComment = false;
1930 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1932 string token = sub.str(5);
1933 if (regex_match(token, sub2, removeArgs)) {
1934 found.keytype = KeyInfo::removeWithArg;
1937 // discard spaces before pos(0)
1938 int pos = sub.position(size_t(0));
1940 for (count = 0; pos - count > 0; count++) {
1941 char c = interval_.par[pos-count-1];
1942 if (discardComment) {
1943 if ((c != ' ') && (c != '%'))
1949 found._tokenstart = pos - count;
1950 if (sub.str(1).compare(0, 5, "begin") == 0) {
1951 size_t pos1 = pos + sub.str(0).length();
1952 if (sub.str(5).compare("cjk") == 0) {
1953 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1954 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1956 found.keytype = KeyInfo::isMain;
1957 found._dataStart = pos1;
1958 found._dataEnd = interval_.par.length();
1959 found.disabled = keys["foreignlanguage"].disabled;
1960 found.used = keys["foreignlanguage"].used;
1961 found._tokensize = pos1 - found._tokenstart;
1962 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1965 // Swallow possible optional params
1966 while (interval_.par[pos1] == '[') {
1967 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1969 // Swallow also the eventual parameter
1970 if (interval_.par[pos1] == '{') {
1971 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1974 found._dataEnd = pos1;
1976 found._dataStart = found._dataEnd;
1977 found._tokensize = count + found._dataEnd - pos;
1978 found.parenthesiscount = 0;
1979 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1980 found.disabled = true;
1984 // Handle "\end{...}"
1985 found._dataStart = pos + sub.str(0).length();
1986 found._dataEnd = found._dataStart;
1987 found._tokensize = count + found._dataEnd - pos;
1988 found.parenthesiscount = 0;
1989 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1990 found.disabled = true;
1994 else if (found.keytype != KeyInfo::isRegex) {
1995 found._tokenstart = sub.position(size_t(0));
1996 if (found.parenthesiscount == 0) {
1997 // Probably to be discarded
1998 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1999 char following = interval_.par[following_pos];
2000 if (following == ' ')
2001 found.head = "\\" + sub.str(3) + " ";
2002 else if (following == '=') {
2003 // like \uldepth=1000pt
2004 found.head = sub.str(0);
2007 found.head = "\\" + key;
2008 found._tokensize = found.head.length();
2009 found._dataEnd = found._tokenstart + found._tokensize;
2010 found._dataStart = found._dataEnd;
2013 int params = found._tokenstart + key.length() + 1;
2014 if (evaluatingOptional) {
2015 if (size_t(found._tokenstart) > optionalEnd) {
2016 evaluatingOptional = false;
2019 found.disabled = true;
2022 int optend = params;
2023 while (interval_.par[optend] == '[') {
2024 // discard optional parameters
2025 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
2027 if (optend > params) {
2028 key += interval_.par.substr(params, optend-params);
2029 evaluatingOptional = true;
2030 optionalEnd = optend;
2032 string token = sub.str(5);
2034 if (interval_.par[optend] != '{') {
2036 found.parenthesiscount = 0;
2037 found.head = "\\" + key;
2040 closings = found.parenthesiscount;
2041 if (found.parenthesiscount == 1) {
2042 found.head = "\\" + key + "{";
2044 else if (found.parenthesiscount > 1) {
2046 found.head = sub.str(0) + "{";
2047 closings = found.parenthesiscount - 1;
2050 found.head = "\\" + key + "{";
2053 found._tokensize = found.head.length();
2054 found._dataStart = found._tokenstart + found.head.length();
2055 if (found.keytype == KeyInfo::doRemove) {
2057 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2058 if (endpar >= interval_.par.length())
2059 found._dataStart = interval_.par.length();
2061 found._dataStart = endpar;
2062 found._tokensize = found._dataStart - found._tokenstart;
2065 found._dataStart = found._tokenstart + found._tokensize;
2069 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
2070 found._dataStart += 15;
2074 endpos = found._dataStart - 1;
2076 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2077 if (found.keytype == KeyInfo::isList) {
2078 // Check if it really is list env
2079 static regex const listre("^([a-z]+)$");
2081 if (!regex_match(token, sub2, listre)) {
2082 // Change the key of this entry. It is not in a list/item environment
2083 found.keytype = KeyInfo::endArguments;
2086 if (found.keytype == KeyInfo::noMain) {
2087 evaluatingCode = true;
2089 codeStart = found._dataStart;
2091 else if (evaluatingCode) {
2092 if (size_t(found._dataStart) > codeEnd)
2093 evaluatingCode = false;
2094 else if (found.keytype == KeyInfo::isMain) {
2095 // Disable this key, treate it as standard
2096 found.keytype = KeyInfo::isStandard;
2097 found.disabled = true;
2098 if ((codeEnd +1 >= interval_.par.length()) &&
2099 (found._tokenstart == codeStart)) {
2100 // trickery, because the code inset starts
2101 // with \selectlanguage ...
2103 if (entries_.size() > 1) {
2104 entries_[entries_.size()-1]._dataEnd = codeEnd;
2109 if ((endpos == interval_.par.length()) &&
2110 (found.keytype == KeyInfo::doRemove)) {
2111 // Missing closing => error in latex-input?
2112 // therefore do not delete remaining data
2113 found._dataStart -= 1;
2114 found._dataEnd = found._dataStart;
2117 found._dataEnd = endpos;
2119 if (isPatternString) {
2120 keys[key].used = true;
2123 entries_.push_back(found);
2127 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2129 stringstream s(keysstring);
2131 const char delim = '|';
2132 while (getline(s, key, delim)) {
2133 KeyInfo keyII(keyI);
2134 if (isPatternString) {
2137 else if ( !keys[key].used)
2138 keyII.disabled = true;
2143 void LatexInfo::buildKeys(bool isPatternString)
2146 static bool keysBuilt = false;
2147 if (keysBuilt && !isPatternString) return;
2149 // Keys to ignore in any case
2150 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2151 // Known standard keys with 1 parameter.
2152 // Split is done, if not at start of region
2153 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2154 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2155 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2156 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2157 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2158 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2160 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2161 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2162 makeKey("section*|subsection*|subsubsection*|paragraph*",
2163 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2164 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2165 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2167 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2169 // Split is done, if not at start of region
2170 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2171 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2173 // Split is done always.
2174 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2178 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2179 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2180 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2181 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2182 makeKey("guillemotright|guillemotleft", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2184 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2185 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2186 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2187 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2189 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2190 // Custom space/skip, remove the content (== length value)
2191 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2192 // Found in fr/UserGuide.lyx
2193 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2195 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2196 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2197 // Known macros to remove (including their parameter)
2199 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2200 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2201 // handle like standard keys with 1 parameter.
2202 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2204 // Ignore deleted text
2205 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2206 // but preserve added text
2207 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2209 // Macros to remove, but let the parameter survive
2211 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2213 // Remove language spec from content of these insets
2214 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2216 // Same effect as previous, parameter will survive (because there is no one anyway)
2218 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2219 // Remove table decorations
2220 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2221 // Discard shape-header.
2222 // For footnote or shortcut too, because of lang settings
2223 // and wrong handling if used 'KeyInfo::noMain'
2224 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2225 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2226 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2227 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2228 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2229 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2230 // like ('tiny{}' or '\tiny ' ... )
2231 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2233 // Survives, like known character
2234 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2235 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2236 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2238 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2239 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2240 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2242 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2243 // Remove RTL/LTR marker
2244 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2245 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2246 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2247 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2248 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2249 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2250 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2251 if (isPatternString) {
2252 // Allow the first searched string to rebuild the keys too
2256 // no need to rebuild again
2262 * Keep the list of actual opened parentheses actual
2263 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2265 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2268 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2272 if (c == '\\') skip = 1;
2273 else if (c == '{') {
2276 else if (c == '}') {
2277 handleCloseP(i, closingAllowed);
2283 string Intervall::show(int lastpos)
2285 int idx = 0; /* int intervalls */
2288 for (idx = 0; idx <= ignoreidx; idx++) {
2289 while (i < lastpos) {
2291 if (i <= borders[idx].low) {
2292 if (borders[idx].low > lastpos)
2293 printsize = lastpos - i;
2295 printsize = borders[idx].low - i;
2296 s += par.substr(i, printsize);
2298 if (i >= borders[idx].low)
2299 i = borders[idx].upper;
2302 i = borders[idx].upper;
2308 s += par.substr(i, lastpos-i);
2314 void Intervall::output(ostringstream &os, int lastpos)
2316 // get number of chars to output
2317 int idx = 0; /* int intervalls */
2320 string startTitle = titleValue;
2321 for (idx = 0; idx <= ignoreidx; idx++) {
2323 if (i <= borders[idx].low) {
2325 if (borders[idx].low > lastpos)
2326 printsize = lastpos - i;
2328 printsize = borders[idx].low - i;
2329 if (printsize > 0) {
2330 os << startTitle << par.substr(i, printsize);
2332 printed += printsize;
2335 handleParentheses(i, false);
2336 if (i >= borders[idx].low)
2337 i = borders[idx].upper;
2340 i = borders[idx].upper;
2347 os << startTitle << par.substr(i, lastpos-i);
2348 printed += lastpos-i;
2350 handleParentheses(lastpos, false);
2352 if (keys["foreignlanguage"].disabled)
2353 startindex = actualdeptindex-langcount;
2355 startindex = actualdeptindex;
2356 for (int i = startindex; i > 0; --i) {
2359 if (hasTitle && (printed > 0))
2361 if (! isPatternString_)
2363 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2366 void LatexInfo::processRegion(int start, int region_end)
2368 while (start < region_end) { /* Let {[} and {]} survive */
2369 int cnt = interval_.isOpeningPar(start);
2371 // Closing is allowed past the region
2372 int closing = interval_.findclosing(start+1, interval_.par.length());
2373 interval_.addIntervall(start, start+1);
2374 interval_.addIntervall(closing, closing+1);
2378 start = interval_.nextNotIgnored(start+1);
2382 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2384 if (actual.parenthesiscount == 0) {
2385 // "{\tiny{} ...}" ==> "{{} ...}"
2386 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2389 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2390 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2391 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2395 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2398 switch (actual.keytype)
2400 case KeyInfo::isTitle: {
2402 nextKeyIdx = getNextKey();
2405 case KeyInfo::cleanToStart: {
2406 actual._dataEnd = actual._dataStart;
2407 nextKeyIdx = getNextKey();
2408 // Search for end of arguments
2409 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2411 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2412 entries_[i].disabled = true;
2414 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2416 while (interval_.par[actual._dataEnd] == ' ')
2418 interval_.addIntervall(0, actual._dataEnd+1);
2419 interval_.actualdeptindex = 0;
2420 interval_.depts[0] = actual._dataEnd+1;
2421 interval_.closes[0] = -1;
2424 case KeyInfo::isText:
2425 interval_.par[actual._tokenstart] = '#';
2426 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2427 nextKeyIdx = getNextKey();
2429 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2430 if (actual.disabled)
2431 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2433 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2436 case KeyInfo::isChar: {
2437 nextKeyIdx = getNextKey();
2440 case KeyInfo::isSize: {
2441 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2442 if (actual.parenthesiscount == 0)
2443 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2445 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2447 nextKeyIdx = getNextKey();
2449 // Here _dataStart points to '{', so correct it
2450 actual._dataStart += 1;
2451 actual._tokensize += 1;
2452 actual.parenthesiscount = 1;
2453 if (interval_.par[actual._dataStart] == '}') {
2454 // Determine the end if used like '{\tiny{}...}'
2455 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2456 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2459 // Determine the end if used like '\tiny{...}'
2460 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2462 // Split on this key if not at start
2463 int start = interval_.nextNotIgnored(previousStart);
2464 if (start < actual._tokenstart) {
2465 interval_.output(os, actual._tokenstart);
2466 interval_.addIntervall(start, actual._tokenstart);
2468 // discard entry if at end of actual
2469 nextKeyIdx = process(os, actual);
2473 case KeyInfo::endArguments: {
2474 // Remove trailing '{}' too
2475 actual._dataStart += 1;
2476 actual._dataEnd += 1;
2477 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2478 nextKeyIdx = getNextKey();
2481 case KeyInfo::noMain:
2483 case KeyInfo::isStandard: {
2484 if (actual.disabled) {
2486 processRegion(actual._dataStart, actual._dataStart+1);
2487 nextKeyIdx = getNextKey();
2489 // Split on this key if not at datastart of calling entry
2490 int start = interval_.nextNotIgnored(previousStart);
2491 if (start < actual._tokenstart) {
2492 interval_.output(os, actual._tokenstart);
2493 interval_.addIntervall(start, actual._tokenstart);
2495 // discard entry if at end of actual
2496 nextKeyIdx = process(os, actual);
2500 case KeyInfo::removeWithArg: {
2501 nextKeyIdx = getNextKey();
2502 // Search for end of arguments
2503 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2505 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2506 entries_[i].disabled = true;
2508 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2510 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2513 case KeyInfo::doRemove: {
2514 // Remove the key with all parameters and following spaces
2516 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2517 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2520 // Remove also enclosing parentheses [] and {}
2523 while (actual._tokenstart > numpars) {
2524 if (pos+numpars >= interval_.par.size())
2526 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2528 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2534 if (interval_.par[pos+numpars] == ' ')
2538 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2539 nextKeyIdx = getNextKey();
2542 case KeyInfo::isList: {
2543 // Discard space before _tokenstart
2545 for (count = 0; count < actual._tokenstart; count++) {
2546 if (interval_.par[actual._tokenstart-count-1] != ' ')
2549 nextKeyIdx = getNextKey();
2550 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2552 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2554 // How else can we catch this one?
2555 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2556 entries_[i].disabled = true;
2558 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2560 else if (nextKeyIdx > 0) {
2561 // Ignore any lang entries inside data region
2562 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2563 if (entries_[i].keytype == KeyInfo::isMain)
2564 entries_[i].disabled = true;
2567 if (actual.disabled) {
2568 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2571 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2573 if (interval_.par[actual._dataEnd+1] == '[') {
2574 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2575 if ((interval_.par[actual._dataEnd+2] == '{') &&
2576 (interval_.par[posdown-1] == '}')) {
2577 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2578 interval_.addIntervall(posdown-1, posdown+1);
2581 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2582 interval_.addIntervall(posdown, posdown+1);
2584 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2585 if (blk > posdown) {
2586 // Discard at most 1 space after empty item
2588 for (count = 0; count < 1; count++) {
2589 if (interval_.par[blk+count] != ' ')
2593 interval_.addIntervall(blk, blk+count);
2598 case KeyInfo::isSectioning: {
2599 // Discard spaces before _tokenstart
2601 int val = actual._tokenstart;
2602 for (count = 0; count < actual._tokenstart;) {
2603 val = interval_.previousNotIgnored(val-1);
2604 if (val < 0 || interval_.par[val] != ' ')
2607 count = actual._tokenstart - val;
2610 if (actual.disabled) {
2611 removeHead(actual, count);
2612 nextKeyIdx = getNextKey();
2614 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2615 nextKeyIdx = process(os, actual);
2619 case KeyInfo::isMath: {
2620 // Same as regex, use the content unchanged
2621 nextKeyIdx = getNextKey();
2624 case KeyInfo::isRegex: {
2625 // DO NOT SPLIT ON REGEX
2627 nextKeyIdx = getNextKey();
2630 case KeyInfo::isIgnored: {
2631 // Treat like a character for now
2632 nextKeyIdx = getNextKey();
2635 case KeyInfo::isMain: {
2636 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2637 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2638 if (actual._tokenstart > 0) {
2639 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2640 if ((prev >= 0) && interval_.par[prev] == '%')
2641 interval_.addIntervall(prev, prev+1);
2643 if (actual.disabled) {
2645 interval_.langcount++;
2646 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2647 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2648 // Discard also the space before math-equation
2649 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2651 nextKeyIdx = getNextKey();
2652 // interval.resetOpenedP(actual._dataStart-1);
2655 if (actual._tokenstart < 26) {
2656 // for the first (and maybe dummy) language
2657 interval_.setForDefaultLang(actual);
2659 interval_.resetOpenedP(actual._dataStart-1);
2663 case KeyInfo::invalid:
2664 case KeyInfo::headRemove:
2665 // These two cases cannot happen, already handled
2668 // LYXERR(Debug::INFO, "Unhandled keytype");
2669 nextKeyIdx = getNextKey();
2676 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2678 int end = interval_.nextNotIgnored(actual._dataEnd);
2679 int oldStart = actual._dataStart;
2680 int nextKeyIdx = getNextKey();
2682 if ((nextKeyIdx < 0) ||
2683 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2684 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2685 if (oldStart <= end) {
2686 processRegion(oldStart, end);
2691 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2693 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2694 (void) dispatch(os, actual._dataStart, nextKey);
2695 end = nextKey._tokenstart;
2698 processRegion(oldStart, nextKey._tokenstart);
2699 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2701 oldStart = nextKey._dataEnd+1;
2703 // now nextKey is either invalid or is outside of actual._dataEnd
2704 // output the remaining and discard myself
2705 if (oldStart <= end) {
2706 processRegion(oldStart, end);
2708 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2710 // This is the normal case.
2711 // But if using the firstlanguage, the closing may be missing
2713 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2715 if (actual._dataEnd < end)
2716 output_end = interval_.nextNotIgnored(actual._dataEnd);
2717 else if (interval_.par.size() > (size_t) end)
2718 output_end = interval_.nextNotIgnored(end);
2720 output_end = interval_.par.size();
2721 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2722 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2724 // Remove possible empty data
2725 int dstart = interval_.nextNotIgnored(actual._dataStart);
2726 while (interval_.isOpeningPar(dstart) == 1) {
2727 interval_.addIntervall(dstart, dstart+1);
2728 int dend = interval_.findclosing(dstart+1, output_end);
2729 interval_.addIntervall(dend, dend+1);
2730 dstart = interval_.nextNotIgnored(dstart+1);
2732 if (dstart < output_end)
2733 interval_.output(os, output_end);
2734 interval_.addIntervall(actual._tokenstart, end);
2738 string splitOnKnownMacros(string par, bool isPatternString)
2741 LatexInfo li(par, isPatternString);
2742 // LYXERR(Debug::INFO, "Berfore split: " << par);
2743 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2745 DummyKey._tokensize = 0;
2746 DummyKey._dataStart = 0;
2747 DummyKey._dataEnd = par.length();
2748 DummyKey.disabled = true;
2749 int firstkeyIdx = li.getFirstKey();
2751 if (firstkeyIdx >= 0) {
2752 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2753 DummyKey._tokenstart = firstKey._tokenstart;
2755 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2756 // Use dummy firstKey
2757 firstKey = DummyKey;
2758 (void) li.setNextKey(firstkeyIdx);
2761 if (par.substr(firstKey._dataStart, 2) == "% ")
2762 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2764 nextkeyIdx = li.process(os, firstKey);
2765 while (nextkeyIdx >= 0) {
2766 // Check for a possible gap between the last
2767 // entry and this one
2768 int datastart = li.nextNotIgnored(firstKey._dataStart);
2769 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2770 if ((nextKey._tokenstart > datastart)) {
2772 firstKey._dataStart = datastart;
2773 firstKey._dataEnd = par.length();
2774 (void) li.setNextKey(nextkeyIdx);
2775 // Fake the last opened parenthesis
2776 li.setForDefaultLang(firstKey);
2777 nextkeyIdx = li.process(os, firstKey);
2780 if (nextKey.keytype != KeyInfo::isMain) {
2781 firstKey._dataStart = datastart;
2782 firstKey._dataEnd = nextKey._dataEnd+1;
2783 (void) li.setNextKey(nextkeyIdx);
2784 li.setForDefaultLang(firstKey);
2785 nextkeyIdx = li.process(os, firstKey);
2788 nextkeyIdx = li.process(os, nextKey);
2792 // Handle the remaining
2793 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2794 firstKey._dataEnd = par.length();
2796 if ((firstKey._dataStart < firstKey._dataEnd) &&
2797 (par[firstKey._dataStart] != '}')) {
2798 li.setForDefaultLang(firstKey);
2799 (void) li.process(os, firstKey);
2803 // return string definitelly impossible to match
2804 s = "\\foreignlanguage{ignore}{ }";
2808 s = par; /* no known macros found */
2809 // LYXERR(Debug::INFO, "After split: " << s);
2814 * Try to unify the language specs in the latexified text.
2815 * Resulting modified string is set to "", if
2816 * the searched tex does not contain all the features in the search pattern
2818 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2820 static Features regex_f;
2821 static int missed = 0;
2822 static bool regex_with_format = false;
2824 int parlen = par.length();
2826 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2829 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2830 // Happens to be there in case of description or labeling environment
2835 // Split the latex input into pieces which
2836 // can be digested by our search engine
2837 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2838 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2839 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2842 result = par.substr(0, parlen);
2843 if (isPatternString) {
2846 regex_f = identifyFeatures(result);
2847 string features = "";
2848 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2849 string a = it->first;
2850 regex_with_format = true;
2851 features += " " + a;
2852 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2854 LYXERR(Debug::FIND, "Identified Features" << features);
2857 } else if (regex_with_format) {
2858 Features info = identifyFeatures(result);
2859 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2860 string a = it->first;
2861 bool b = it->second;
2862 if (b && ! info[a]) {
2864 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2870 // LYXERR(Debug::INFO, "No regex formats");
2876 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2877 static int identifyClosing(string & t)
2879 int open_braces = 0;
2881 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2882 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2884 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2886 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2888 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2897 static int num_replaced = 0;
2898 static bool previous_single_replace = true;
2900 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2901 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2903 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2904 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2905 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2906 if (opt.replace_all && previous_single_replace) {
2907 previous_single_replace = false;
2910 else if (!opt.replace_all) {
2911 num_replaced = 0; // count number of replaced strings
2912 previous_single_replace = true;
2914 // When using regexp, braces are hacked already by escape_for_regex()
2915 par_as_string = normalize(ds, !use_regexp);
2917 close_wildcards = 0;
2919 size_t lead_size = 0;
2920 // correct the language settings
2921 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2922 if (opt.ignoreformat) {
2924 // if par_as_string_nolead were emty,
2925 // the following call to findAux will always *find* the string
2926 // in the checked data, and thus always using the slow
2927 // examining of the current text part.
2928 par_as_string_nolead = par_as_string;
2931 lead_size = identifyLeading(par_as_string);
2932 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2933 lead_as_string = par_as_string.substr(0, lead_size);
2934 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2938 open_braces = identifyClosing(par_as_string);
2939 identifyClosing(par_as_string_nolead);
2940 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2941 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2943 string lead_as_regexp;
2944 if (lead_size > 0) {
2945 // @todo No need to search for \regexp{} insets in leading material
2946 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2947 par_as_string = par_as_string_nolead;
2948 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2949 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2951 LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
2952 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2953 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2954 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2956 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2957 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2958 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2959 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2960 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2961 || regex_replace(par_as_string, par_as_string,
2962 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2963 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2964 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2968 if (!opt.ignoreformat) {
2969 // Remove extra '\}' at end if not part of \{\.\}
2970 size_t lng = par_as_string.size();
2972 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2974 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2983 if (lng < par_as_string.size())
2984 par_as_string = par_as_string.substr(0,lng);
2987 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2988 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2989 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2990 // replace '[^...]' with '[^...\}\{\\]'
2991 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2992 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2993 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2995 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2998 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2999 LYXERR(Debug::FIND, "Open braces: " << open_braces);
3000 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
3001 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3003 // If entered regexp must match at begin of searched string buffer
3004 // Kornel: Added parentheses to use $1 for size of the leading string
3008 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3009 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3010 // so the convert has no effect in that case
3011 for (int i = 7; i > 0; --i) {
3012 string orig = "\\\\" + std::to_string(i);
3013 string dest = "\\" + std::to_string(i+2);
3014 while (regex_replace(par_as_string, par_as_string, orig, dest));
3016 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3017 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3019 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3020 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3022 // Handle \w properly
3023 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
3024 if (! opt.casesensitive) {
3025 popts |= QRegularExpression::CaseInsensitiveOption;
3027 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
3028 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
3030 if (regexp.isValid() && regexp2.isValid()) {
3031 regexIsValid = true;
3032 // Check '{', '}' pairs inside the regex
3035 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
3036 char c = par_as_string[i];
3043 else if (c == '}') {
3050 if (balanced != 0) {
3051 regexIsValid = false;
3052 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
3056 regexIsValid = false;
3057 if (!regexp.isValid())
3058 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
3059 if (!regexp2.isValid())
3060 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
3063 if (opt.casesensitive) {
3064 regexp = regex(regexp_str);
3065 regexp2 = regex(regexp2_str);
3068 regexp = regex(regexp_str, std::regex_constants::icase);
3069 regexp2 = regex(regexp2_str, std::regex_constants::icase);
3076 // Count number of characters in string
3080 // \\[a-zA-Z]+ ==> 1
3082 static int computeSize(QStringRef s, int len)
3083 #define isLyxAlpha(arg) arg.isLetter()
3085 static int computeSize(string s, int len)
3086 #define isLyxAlpha(arg) isalpha(arg)
3093 for (int i = 0; i < len; i += skip, count++) {
3094 if (s.at(i) == '\\') {
3096 if (i + 1 < len && isLyxAlpha(s.at(i+1))) {
3097 for (int j = 2; i+j < len; j++) {
3098 if (! isLyxAlpha(s.at(i+j))) {
3099 if (s.at(i+j) == ' ')
3101 else if (s.at(i+j) == '{') {
3102 if (i+j+1 < len && s.at(i+j+1) == '}')
3104 else if (i + j + 1 >= len)
3113 else if (s.at(i) == '{') {
3114 if (i + 1 < len && s.at(i+1) == '}')
3119 else if (s.at(i) == '-') {
3120 if (i+1 < len && s.at(i+1) == '-') {
3121 if (i + 2 < len && s.at(i+2) == '-')
3137 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3142 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3145 docstring docstr = stringifyFromForSearch(opt, cur, len);
3147 if (use_regexp || opt.casesensitive)
3148 str = normalize(docstr, true);
3150 str = normalize(lowercase(docstr), true);
3151 if (!opt.ignoreformat) {
3152 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3155 mres.match_len = -1;
3158 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
3159 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
3162 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3164 QString qstr = QString::fromStdString(str);
3165 QRegularExpression const *p_regexp;
3166 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3170 p_regexp = ®exp2;
3172 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3173 if (!match.hasMatch())
3175 // Check braces on segments that matched all (.*?) subexpressions,
3176 // except the last "padding" one inserted by lyx.
3177 for (int i = 3; i < match.lastCapturedIndex(); ++i)
3178 if (!braces_match(match.captured(i), open_braces))
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;
3194 // Check braces on segments that matched all (.*?) subexpressions,
3195 // except the last "padding" one inserted by lyx.
3196 for (size_t i = 3; i < m.size() - 1; ++i)
3197 if (!braces_match(m[i], open_braces))
3200 // Exclude from the returned match length any length
3201 // due to close wildcards added at end of regexp
3202 // and also the length of the leading (e.g. '\emph{}')
3204 // Whole found string, including the leading: m[0].second - m[0].first
3205 // Size of the leading string: m[1].second - m[1].first
3206 int leadingsize = 0;
3209 if (match.lastCapturedIndex() > 0) {
3210 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3212 int lastidx = match.lastCapturedIndex();
3213 for (int i = 0; i <= lastidx; i++) {
3214 LYXERR(Debug::FIND, "Match " << i << " is " << match.capturedEnd(i) - match.capturedStart(i) << " long");
3216 if (close_wildcards == 0)
3217 result = match.capturedEnd(0) - match.capturedStart(0);
3219 result = match.capturedStart(lastidx + 1 - close_wildcards) - match.capturedStart(0);
3223 leadingsize = m[1].second - m[1].first;
3225 for (size_t i = 0; i < m.size(); i++) {
3226 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
3228 if (close_wildcards == 0)
3229 result = m[0].second - m[0].first;
3231 result = m[m.size() - close_wildcards].first - m[0].first;
3233 if (result > leadingsize)
3234 result -= leadingsize;
3238 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3239 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3240 // because of different number of closing at end of string
3241 // we have to 'unify' the length of the post-match.
3242 // Done by ignoring closing parenthesis and linefeeds at string end
3243 int matchend = match.capturedEnd(0);
3244 while (mres.match_len > 0) {
3245 QChar c = qstr.at(matchend - 1);
3246 if ((c == '\n') || (c == '}') || (c == '{')) {
3253 size_t strsize = qstr.size();
3254 while (strsize > (size_t) match.capturedEnd(0)) {
3255 QChar c = qstr.at(strsize-1);
3256 if ((c == '\n') || (c == '}')) {
3262 // LYXERR0(qstr.toStdString());
3263 mres.match2end = strsize - matchend;
3264 mres.pos = match.capturedStart(2);
3266 mres.match_prefix = m[2].second - m[2].first;
3267 mres.match_len = m[0].second - m[2].second;
3268 // ignore closing parenthesis and linefeeds at string end
3269 size_t strend = m[0].second - m[0].first;
3270 int matchend = strend;
3271 while (mres.match_len > 0) {
3272 char c = str.at(matchend - 1);
3273 if ((c == '\n') || (c == '}') || (c == '{')) {
3280 size_t strsize = str.size();
3281 while (strsize > strend) {
3282 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3289 mres.match2end = strsize - matchend;
3290 mres.pos = m[2].first - m[0].first;;
3292 if (mres.match2end < 0)
3294 mres.leadsize = leadingsize;
3298 // else !use_regexp: but all code paths above return
3299 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
3300 << par_as_string << "', str='" << str << "'");
3301 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
3302 << lead_as_string << "', par_as_string_nolead='"
3303 << par_as_string_nolead << "'");
3306 LYXERR(Debug::FIND, "size=" << par_as_string.size()
3307 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
3308 if (str.substr(0, par_as_string.size()) == par_as_string) {
3309 mres.match_len = par_as_string.size();
3310 mres.match2end = str.size();
3315 // Start the search _after_ the leading part
3316 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3317 if (pos != string::npos) {
3318 mres.match_len = par_as_string.size();
3319 mres.match2end = str.size() - pos;
3328 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3330 MatchResult mres = findAux(cur, len, at_begin);
3331 int res = mres.match_len;
3333 "res=" << res << ", at_begin=" << at_begin
3334 << ", matchword=" << opt.matchword
3335 << ", inTexted=" << cur.inTexted());
3336 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3338 if ((len > 0) && (res < len)) {
3342 Paragraph const & par = cur.paragraph();
3343 bool ws_left = (cur.pos() > 0)
3344 ? par.isWordSeparator(cur.pos() - 1)
3350 ws_right = (cur.pos() + len < par.size())
3351 ? par.isWordSeparator(cur.pos() + len)
3355 "cur.pos()=" << cur.pos() << ", res=" << res
3356 << ", separ: " << ws_left << ", " << ws_right
3359 if (ws_left && ws_right) {
3360 // Check for word separators inside the found 'word'
3361 for (int i = 0; i < len; i++) {
3362 if (par.isWordSeparator(cur.pos() + i)) {
3374 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
3377 t = lyx::to_utf8(s);
3378 // Remove \n at begin
3379 while (!t.empty() && t[0] == '\n')
3382 while (!t.empty() && t[t.size() - 1] == '\n')
3383 t = t.substr(0, t.size() - 1);
3385 // Handle all other '\n'
3386 while ((pos = t.find("\n")) != string::npos) {
3387 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3389 if (isAlnumASCII(t[pos+1])) {
3390 t.replace(pos-2, 3, " ");
3393 t.replace(pos-2, 3, "");
3396 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3397 // '\n' adjacent to non-alpha-numerics, discard
3398 t.replace(pos, 1, "");
3401 // Replace all other \n with spaces
3402 t.replace(pos, 1, " ");
3405 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3406 // Kornel: Added textsl, textsf, textit, texttt and noun
3407 // + allow to seach for colored text too
3408 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3409 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3410 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3411 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3412 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3413 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3414 // FIXME - check what preceeds the brace
3416 if (opt.ignoreformat)
3417 while (regex_replace(t, t, "\\{", "_x_<")
3418 || regex_replace(t, t, "\\}", "_x_>"))
3419 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3421 while (regex_replace(t, t, "\\\\\\{", "_x_<")
3422 || regex_replace(t, t, "\\\\\\}", "_x_>"))
3423 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3430 docstring stringifyFromCursor(DocIterator const & cur, int len)
3432 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3433 if (cur.inTexted()) {
3434 Paragraph const & par = cur.paragraph();
3435 // TODO what about searching beyond/across paragraph breaks ?
3436 // TODO Try adding a AS_STR_INSERTS as last arg
3437 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3438 int(par.size()) : cur.pos() + len;
3439 // OutputParams runparams(&cur.buffer()->params().encoding());
3440 OutputParams runparams(encodings.fromLyXName("utf8"));
3441 runparams.nice = true;
3442 runparams.flavor = Flavor::XeTeX;
3443 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3444 // No side effect of file copying and image conversion
3445 runparams.dryrun = true;
3446 runparams.for_search = true;
3447 LYXERR(Debug::FIND, "Stringifying with cur: "
3448 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3449 return par.asString(cur.pos(), end,
3450 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3452 } else if (cur.inMathed()) {
3453 CursorSlice cs = cur.top();
3454 MathData md = cs.cell();
3455 MathData::const_iterator it_end =
3456 (( len == -1 || cs.pos() + len > int(md.size()))
3458 : md.begin() + cs.pos() + len );
3460 for (MathData::const_iterator it = md.begin() + cs.pos();
3463 docstring s = asString(md2);
3464 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3467 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3472 /** Computes the LaTeX export of buf starting from cur and ending len positions
3473 * after cur, if len is positive, or at the paragraph or innermost inset end
3476 docstring latexifyFromCursor(DocIterator const & cur, int len)
3478 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3479 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3480 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3481 Buffer const & buf = *cur.buffer();
3483 odocstringstream ods;
3485 //OutputParams runparams(&buf.params().encoding());
3486 OutputParams runparams(encodings.fromLyXName("utf8"));
3487 runparams.nice = false;
3488 runparams.flavor = Flavor::XeTeX;
3489 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3490 // No side effect of file copying and image conversion
3491 runparams.dryrun = true;
3492 runparams.for_search = true;
3494 if (cur.inTexted()) {
3495 // @TODO what about searching beyond/across paragraph breaks ?
3496 pos_type endpos = cur.paragraph().size();
3497 if (len != -1 && endpos > cur.pos() + len)
3498 endpos = cur.pos() + len;
3499 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3500 string(), cur.pos(), endpos);
3501 string s = lyx::to_utf8(ods.str());
3502 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3503 return(lyx::from_utf8(s));
3504 } else if (cur.inMathed()) {
3505 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3506 for (int s = cur.depth() - 1; s >= 0; --s) {
3507 CursorSlice const & cs = cur[s];
3508 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3509 TeXMathStream ws(os);
3510 cs.asInsetMath()->asHullInset()->header_write(ws);
3515 CursorSlice const & cs = cur.top();
3516 MathData md = cs.cell();
3517 MathData::const_iterator it_end =
3518 ((len == -1 || cs.pos() + len > int(md.size()))
3520 : md.begin() + cs.pos() + len);
3522 for (MathData::const_iterator it = md.begin() + cs.pos();
3526 ods << asString(md2);
3527 // Retrieve the math environment type, and add '$' or '$]'
3528 // or others (\end{equation}) accordingly
3529 for (int s = cur.depth() - 1; s >= 0; --s) {
3530 CursorSlice const & cs2 = cur[s];
3531 InsetMath * inset = cs2.asInsetMath();
3532 if (inset && inset->asHullInset()) {
3533 TeXMathStream ws(os);
3534 inset->asHullInset()->footer_write(ws);
3538 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3540 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3547 static void displayMResult(MatchResult &mres, int increment)
3549 LYXERR0( "pos: " << mres.pos << " increment " << increment);
3550 LYXERR0( "leadsize: " << mres.leadsize);
3551 LYXERR0( "match_len: " << mres.match_len);
3552 LYXERR0( "match_prefix: " << mres.match_prefix);
3553 LYXERR0( "match2end: " << mres.match2end);
3555 #define displayMres(s,i) displayMResult(s,i);
3557 #define displayMres(s,i)
3560 /** Finalize an advanced find operation, advancing the cursor to the innermost
3561 ** position that matches, plus computing the length of the matching text to
3564 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, int expected_len)
3566 // Search the foremost position that matches (avoids find of entire math
3567 // inset when match at start of it)
3569 DocIterator old_cur(cur.buffer());
3572 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3578 if (cur.depth() > d)
3580 if (cur.depth() == d)
3583 displayMres(mres, 1);
3584 if (expected_len > 0) {
3585 if (mres.match_len < expected_len)
3589 if (mres.match_len <= 0)
3594 mres = match(cur); /* match valid only if not searching whole words */
3595 int max_match = mres.match_len;
3596 if (max_match <= 0) return 0;
3597 LYXERR(Debug::FIND, "Ok");
3599 // Compute the match length
3601 if (cur.pos() + len > cur.lastpos())
3603 // regexp should use \w+, \S+, or \b(some string)\b
3604 // to search for whole words
3605 if (match.opt.matchword && !match.use_regexp) {
3606 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3607 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3609 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3611 // Length of matched text (different from len param)
3612 int old_match = match(cur, len).match_len;
3616 // Greedy behaviour while matching regexps
3617 while ((new_match = match(cur, len + 1).match_len) > old_match) {
3619 old_match = new_match;
3620 LYXERR(Debug::FIND, "verifying match with len = " << len);
3627 int maxl = cur.lastpos() - cur.pos();
3628 // Greedy behaviour while matching regexps
3629 while (maxl > minl) {
3630 int actual_match = match(cur, len).match_len;
3631 if (actual_match >= max_match) {
3632 // actual_match > max_match _can_ happen,
3633 // if the search area splits
3634 // some following word so that the regex
3635 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3637 // This means, the len value is too big
3639 if (maxl - minl < 4)
3640 len = (int)((maxl + minl)/2);
3642 len = (int)(minl + (maxl - minl + 3)/4);
3645 // (actual_match < max_match)
3647 len = (int)((maxl + minl)/2);
3651 // Search for real start of matched characters
3656 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3657 if (cur.depth() < old_cur.depth()) {
3659 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3662 if (cur.pos() != old_cur.pos()) {
3663 // OK, forwarded 1 pos in actual inset
3664 actual_match = match(cur, len-1).match_len;
3665 if (actual_match == max_match) {
3666 // Ha, got it! The shorter selection has the same match length
3671 // OK, the shorter selection matches less chars, revert to previous value
3677 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3678 actual_match = match(cur, len).match_len;
3679 if (actual_match == max_match)
3688 int findForwardAdv(DocIterator & cur, MatchStringAdv const & match)
3692 while (!theApp()->longOperationCancelled() && cur) {
3696 DocIterator old_cur(cur.buffer());
3703 if (cur.depth() > d)
3705 if (cur.depth() == d)
3711 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3712 MatchResult mres = match(cur, -1, false);
3713 displayMres(mres,-1)
3714 int match_len = mres.match_len;
3715 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3716 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3719 if (match_len > 0) {
3720 // Try to find the begin of searched string
3722 int firstInvalid = 100000;
3723 if (mres.match_prefix + mres.pos - mres.leadsize > 1)
3724 increment = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3727 LYXERR(Debug::FIND, "Set increment to " << increment);
3728 while (increment > 0) {
3729 DocIterator old_cur = cur;
3730 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3732 if (! cur || (cur.pit() > old_cur.pit())) {
3733 // Are we outside of the paragraph?
3734 // This can happen if moving past some UTF8-encoded chars
3739 MatchResult mres2 = match(cur, -1, false);
3740 displayMres(mres2,increment)
3741 switch (interpretMatch(mres, mres2)) {
3742 case MatchResult::newIsTooFar:
3743 // behind the expected match
3744 firstInvalid = increment;
3748 case MatchResult::newIsBetter:
3749 // not reached ye, but cur.pos()+increment is bettert
3751 firstInvalid -= increment;
3752 if (increment > firstInvalid*3/4)
3753 increment = firstInvalid*3/4;
3754 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3755 if (increment >= mres2.match_prefix)
3756 increment = (mres2.match_prefix+1)*3/4;
3761 // Handle not like MatchResult::newIsTooFar
3762 // LYXERR0( "Something is wrong: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3764 increment = increment*3/4;
3770 // LYXERR0("Leaving first loop");
3772 LYXERR(Debug::FIND, "Finalizing 1");
3773 int len = findAdvFinalize(cur, match, mres.match_len);
3777 // try next possible match
3782 // The following code is newer reached
3783 // but parts of it may be needed in future
3784 int match_len_zero_count = 0;
3786 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3788 mres3 = match(cur, -1, false);
3789 displayMres(mres3, 1)
3790 int remaining_len = mres3.match_len;
3791 if (remaining_len <= 0) {
3792 // Apparently the searched string is not in the remaining part
3799 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3800 mres3 = match(cur, 1);
3801 displayMres(mres3, 1)
3802 int match_len3 = mres3.match_len;
3806 displayMres(mres3, 1)
3807 int match_len2 = mres3.match_len;
3808 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3809 if (match_len2 > 0) {
3810 // Sometimes in finalize we understand it wasn't a match
3811 // and we need to continue the outest loop
3812 LYXERR(Debug::FIND, "Finalizing 2");
3813 int len = findAdvFinalize(cur, match, mres.match_len);
3819 match_len_zero_count = 0;
3820 else if (match_len2 == 0)
3821 match_len_zero_count++;
3822 if (match_len2 < 0) {
3823 if (++match_len_zero_count > 3) {
3824 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3832 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3833 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3836 // This should exit nested insets, if any, or otherwise undefine the currsor.
3837 cur.pos() = cur.lastpos();
3838 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3846 /// Find the most backward consecutive match within same paragraph while searching backwards.
3847 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3849 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3850 DocIterator tmp_cur = cur;
3851 int len = findAdvFinalize(tmp_cur, match, -1);
3852 Inset & inset = cur.inset();
3853 for (; cur != cur_begin; cur.backwardPos()) {
3854 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3855 DocIterator new_cur = cur;
3856 new_cur.backwardPos();
3857 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3859 int new_len = findAdvFinalize(new_cur, match, -1);
3864 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3870 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3874 // Backup of original position
3875 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3876 if (cur == cur_begin)
3879 DocIterator cur_orig(cur);
3880 bool pit_changed = false;
3883 bool found_match = (match(cur, -1, false).match_len > 0);
3887 cur.pos() = cur.lastpos();
3889 cur.pos() = cur_orig.pos();
3890 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3891 DocIterator cur_prev_iter;
3893 found_match = (match(cur).match_len > 0);
3894 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3895 << found_match << ", cur: " << cur);
3897 return findMostBackwards(cur, match);
3899 // Stop if begin of document reached
3900 if (cur == cur_begin)
3902 cur_prev_iter = cur;
3906 if (cur == cur_begin)
3913 } while (!theApp()->longOperationCancelled());
3921 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3922 DocIterator const & cur, int len)
3924 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3926 if (!opt.ignoreformat)
3927 return latexifyFromCursor(cur, len);
3929 return stringifyFromCursor(cur, len);
3933 FindAndReplaceOptions::FindAndReplaceOptions(
3934 docstring const & _find_buf_name, bool _casesensitive,
3935 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3936 docstring const & _repl_buf_name, bool _keep_case,
3937 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3938 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3939 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3940 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3948 /** Check if 'len' letters following cursor are all non-lowercase */
3949 static bool allNonLowercase(Cursor const & cur, int len)
3951 pos_type beg_pos = cur.selectionBegin().pos();
3952 pos_type end_pos = cur.selectionBegin().pos() + len;
3953 if (len > cur.lastpos() + 1 - beg_pos) {
3954 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3955 len = cur.lastpos() + 1 - beg_pos;
3956 end_pos = beg_pos + len;
3958 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3959 if (isLowerCase(cur.paragraph().getChar(pos)))
3965 /** Check if first letter is upper case and second one is lower case */
3966 static bool firstUppercase(Cursor const & cur)
3969 pos_type pos = cur.selectionBegin().pos();
3970 if (pos >= cur.lastpos() - 1) {
3971 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3974 ch1 = cur.paragraph().getChar(pos);
3975 ch2 = cur.paragraph().getChar(pos + 1);
3976 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3977 LYXERR(Debug::FIND, "firstUppercase(): "
3978 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3979 << ch2 << "(" << char(ch2) << ")"
3980 << ", result=" << result << ", cur=" << cur);
3985 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3987 ** \fixme What to do with possible further paragraphs in replace buffer ?
3989 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3991 ParagraphList::iterator pit = buffer.paragraphs().begin();
3992 LASSERT(!pit->empty(), /**/);
3993 pos_type right = pos_type(1);
3994 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3995 right = pit->size();
3996 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
4002 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
4004 Cursor & cur = bv->cursor();
4005 if (opt.repl_buf_name.empty()
4006 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
4007 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4010 DocIterator sel_beg = cur.selectionBegin();
4011 DocIterator sel_end = cur.selectionEnd();
4012 if (&sel_beg.inset() != &sel_end.inset()
4013 || sel_beg.pit() != sel_end.pit()
4014 || sel_beg.idx() != sel_end.idx())
4016 int sel_len = sel_end.pos() - sel_beg.pos();
4017 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
4018 << ", sel_len: " << sel_len << endl);
4021 LASSERT(sel_len > 0, return 0);
4023 if (!matchAdv(sel_beg, sel_len).match_len)
4026 // Build a copy of the replace buffer, adapted to the KeepCase option
4027 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
4029 repl_buffer_orig.write(oss);
4030 string lyx = oss.str();
4031 Buffer repl_buffer("", false);
4032 repl_buffer.setUnnamed(true);
4033 LASSERT(repl_buffer.readString(lyx), return 0);
4034 if (opt.keep_case && sel_len >= 2) {
4035 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
4036 if (cur.inTexted()) {
4037 if (firstUppercase(cur))
4038 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
4039 else if (allNonLowercase(cur, sel_len))
4040 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
4043 cap::cutSelection(cur, false);
4044 if (cur.inTexted()) {
4045 repl_buffer.changeLanguage(
4046 repl_buffer.language(),
4047 cur.getFont().language());
4048 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
4049 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
4050 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
4051 repl_buffer.params().documentClassPtr(),
4052 bv->buffer().errorList("Paste"));
4053 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4054 sel_len = repl_buffer.paragraphs().begin()->size();
4055 } else if (cur.inMathed()) {
4056 odocstringstream ods;
4058 // OutputParams runparams(&repl_buffer.params().encoding());
4059 OutputParams runparams(encodings.fromLyXName("utf8"));
4060 runparams.nice = false;
4061 runparams.flavor = Flavor::XeTeX;
4062 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4063 runparams.dryrun = true;
4064 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4065 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4066 docstring repl_latex = ods.str();
4067 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4069 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4070 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4071 repl_latex = from_utf8(s);
4072 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4073 MathData ar(cur.buffer());
4074 asArray(repl_latex, ar, Parse::NORMAL);
4076 sel_len = ar.size();
4077 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4079 if (cur.pos() >= sel_len)
4080 cur.pos() -= sel_len;
4083 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4084 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4085 bv->processUpdateFlags(Update::Force);
4090 /// Perform a FindAdv operation.
4091 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
4096 // e.g., when invoking word-findadv from mini-buffer wither with
4097 // wrong options syntax or before ever opening advanced F&R pane
4098 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4102 MatchStringAdv matchAdv(bv->buffer(), opt);
4104 if (!matchAdv.regexIsValid) {
4105 bv->message(lyx::from_utf8(matchAdv.regexError));
4109 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4111 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4112 num_replaced += findAdvReplace(bv, opt, matchAdv);
4115 match_len = findForwardAdv(cur, matchAdv);
4117 match_len = findBackwardsAdv(cur, matchAdv);
4118 } catch (exception & ex) {
4119 bv->message(from_ascii(ex.what()));
4123 if (match_len == 0) {
4124 if (num_replaced > 0) {
4125 switch (num_replaced)
4128 bv->message(_("One match has been replaced."));
4131 bv->message(_("Two matches have been replaced."));
4134 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4140 bv->message(_("Match not found."));
4145 if (num_replaced > 0)
4146 bv->message(_("Match has been replaced."));
4148 bv->message(_("Match found."));
4150 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
4151 bv->putSelectionAt(cur, match_len, !opt.forward);
4157 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4159 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4160 << opt.casesensitive << ' '
4161 << opt.matchword << ' '
4162 << opt.forward << ' '
4163 << opt.expandmacros << ' '
4164 << opt.ignoreformat << ' '
4165 << opt.replace_all << ' '
4166 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4167 << opt.keep_case << ' '
4168 << int(opt.scope) << ' '
4171 LYXERR(Debug::FIND, "built: " << os.str());
4177 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4179 LYXERR(Debug::FIND, "parsing");
4183 while (line != "EOSS") {
4187 if (is.eof()) // Tolerate malformed request
4191 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4192 opt.find_buf_name = from_utf8(s);
4193 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4194 is.get(); // Waste space before replace string
4197 while (line != "EOSS") {
4201 if (is.eof()) // Tolerate malformed request
4205 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4206 opt.repl_buf_name = from_utf8(s);
4207 is >> opt.keep_case;
4210 opt.scope = FindAndReplaceOptions::SearchScope(i);
4212 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4214 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4215 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4216 << opt.scope << ' ' << opt.restr);