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
12 * Full author contact details are available in file CREDITS.
20 #include "BufferList.h"
21 #include "BufferParams.h"
22 #include "BufferView.h"
25 #include "CutAndPaste.h"
26 #include "FuncRequest.h"
28 #include "output_latex.h"
29 #include "OutputParams.h"
30 #include "Paragraph.h"
34 #include "frontends/Application.h"
35 #include "frontends/alert.h"
37 #include "mathed/InsetMath.h"
38 #include "mathed/InsetMathHull.h"
39 #include "mathed/MathData.h"
40 #include "mathed/MathStream.h"
41 #include "mathed/MathSupport.h"
43 #include "support/debug.h"
44 #include "support/docstream.h"
45 #include "support/FileName.h"
46 #include "support/gettext.h"
47 #include "support/lassert.h"
48 #include "support/lstrings.h"
50 #include "support/regex.h"
51 #include "support/textutils.h"
55 using namespace lyx::support;
60 // Helper class for deciding what should be ignored
65 : ignoreFamily_(false), ignoreSeries_(false),
66 ignoreShape_(false), ignoreUnderline_(false),
67 ignoreMarkUp_(false), ignoreStrikeOut_(false),
68 ignoreSectioning_(false), ignoreFrontMatter_(false),
69 ignoreColor_(false), ignoreLanguage_(false) {}
71 bool getFamily() { return ignoreFamily_; }
73 bool getSeries() { return ignoreSeries_; }
75 bool getShape() { return ignoreShape_; }
77 bool getUnderline() { return ignoreUnderline_; }
79 bool getMarkUp() { return ignoreMarkUp_; }
81 bool getStrikeOut() { return ignoreStrikeOut_; }
83 bool getSectioning() { return ignoreSectioning_; }
85 bool getFrontMatter() { return ignoreFrontMatter_; }
87 bool getColor() { return ignoreColor_; }
89 bool getLanguage() { return ignoreLanguage_; }
91 void setIgnoreFormat(string const & type, bool value);
101 bool ignoreUnderline_;
105 bool ignoreStrikeOut_;
107 bool ignoreSectioning_;
109 bool ignoreFrontMatter_;
113 bool ignoreLanguage_;
117 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
119 if (type == "color") {
120 ignoreColor_ = value;
122 else if (type == "language") {
123 ignoreLanguage_ = value;
125 else if (type == "sectioning") {
126 ignoreSectioning_ = value;
127 ignoreFrontMatter_ = value;
129 else if (type == "font") {
130 ignoreSeries_ = value;
131 ignoreShape_ = value;
132 ignoreFamily_ = value;
134 else if (type == "series") {
135 ignoreSeries_ = value;
137 else if (type == "shape") {
138 ignoreShape_ = value;
140 else if (type == "family") {
141 ignoreFamily_ = value;
143 else if (type == "markup") {
144 ignoreMarkUp_ = value;
146 else if (type == "underline") {
147 ignoreUnderline_ = value;
149 else if (type == "strike") {
150 ignoreStrikeOut_ = value;
154 // The global variable that can be changed from outside
155 IgnoreFormats ignoreFormats;
158 void setIgnoreFormat(string const & type, bool value)
160 ignoreFormats.setIgnoreFormat(type, value);
166 bool parse_bool(docstring & howto)
171 howto = split(howto, var, ' ');
179 MatchString(docstring const & s, bool cs, bool mw)
180 : str(s), case_sens(cs), whole_words(mw)
183 // returns true if the specified string is at the specified position
184 // del specifies whether deleted strings in ct mode will be considered
185 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
187 return par.find(str, case_sens, whole_words, pos, del);
195 // match whole words only
200 int findForward(DocIterator & cur, MatchString const & match,
201 bool find_del = true)
203 for (; cur; cur.forwardChar())
204 if (cur.inTexted()) {
205 int len = match(cur.paragraph(), cur.pos(), find_del);
213 int findBackwards(DocIterator & cur, MatchString const & match,
214 bool find_del = true)
218 if (cur.inTexted()) {
219 int len = match(cur.paragraph(), cur.pos(), find_del);
228 bool searchAllowed(docstring const & str)
231 frontend::Alert::error(_("Search error"), _("Search string is empty"));
238 bool findOne(BufferView * bv, docstring const & searchstr,
239 bool case_sens, bool whole, bool forward,
240 bool find_del = true, bool check_wrap = false)
242 if (!searchAllowed(searchstr))
245 DocIterator cur = forward
246 ? bv->cursor().selectionEnd()
247 : bv->cursor().selectionBegin();
249 MatchString const match(searchstr, case_sens, whole);
251 int match_len = forward
252 ? findForward(cur, match, find_del)
253 : findBackwards(cur, match, find_del);
256 bv->putSelectionAt(cur, match_len, !forward);
257 else if (check_wrap) {
258 DocIterator cur_orig(bv->cursor());
261 q = _("End of file reached while searching forward.\n"
262 "Continue searching from the beginning?");
264 q = _("Beginning of file reached while searching backward.\n"
265 "Continue searching from the end?");
266 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
267 q, 0, 1, _("&Yes"), _("&No"));
268 if (wrap_answer == 0) {
270 bv->cursor().clear();
271 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
273 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
274 bv->cursor().backwardPos();
276 bv->clearSelection();
277 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
280 bv->cursor().setCursor(cur_orig);
284 return match_len > 0;
288 int replaceAll(BufferView * bv,
289 docstring const & searchstr, docstring const & replacestr,
290 bool case_sens, bool whole)
292 Buffer & buf = bv->buffer();
294 if (!searchAllowed(searchstr) || buf.isReadonly())
297 DocIterator cur_orig(bv->cursor());
299 MatchString const match(searchstr, case_sens, whole);
302 int const rsize = replacestr.size();
303 int const ssize = searchstr.size();
306 cur.setCursor(doc_iterator_begin(&buf));
307 int match_len = findForward(cur, match, false);
308 while (match_len > 0) {
309 // Backup current cursor position and font.
310 pos_type const pos = cur.pos();
311 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
313 int striked = ssize -
314 cur.paragraph().eraseChars(pos, pos + match_len,
315 buf.params().track_changes);
316 cur.paragraph().insert(pos, replacestr, font,
317 Change(buf.params().track_changes
319 : Change::UNCHANGED));
320 for (int i = 0; i < rsize + striked; ++i)
323 match_len = findForward(cur, match, false);
326 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
328 cur_orig.fixIfBroken();
329 bv->setCursor(cur_orig);
335 // the idea here is that we are going to replace the string that
336 // is selected IF it is the search string.
337 // if there is a selection, but it is not the search string, then
338 // we basically ignore it. (FIXME We ought to replace only within
340 // if there is no selection, then:
341 // (i) if some search string has been provided, then we find it.
342 // (think of how the dialog works when you hit "replace" the
344 // (ii) if no search string has been provided, then we treat the
345 // word the cursor is in as the search string. (why? i have no
346 // idea.) but this only works in text?
348 // returns the number of replacements made (one, if any) and
349 // whether anything at all was done.
350 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
351 docstring const & replacestr, bool case_sens,
352 bool whole, bool forward, bool findnext)
354 Cursor & cur = bv->cursor();
355 if (!cur.selection()) {
356 // no selection, non-empty search string: find it
357 if (!searchstr.empty()) {
358 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
359 return make_pair(found, 0);
361 // empty search string
364 return make_pair(false, 0);
365 // select current word and treat it as the search string.
366 // This causes a minor bug as undo will restore this selection,
367 // which the user did not create (#8986).
368 cur.innerText()->selectWord(cur, WHOLE_WORD);
369 searchstr = cur.selectionAsString(false);
372 // if we still don't have a search string, report the error
374 if (!searchAllowed(searchstr))
375 return make_pair(false, 0);
377 bool have_selection = cur.selection();
378 docstring const selected = cur.selectionAsString(false);
381 ? searchstr == selected
382 : compare_no_case(searchstr, selected) == 0;
384 // no selection or current selection is not search word:
385 // just find the search word
386 if (!have_selection || !match) {
387 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
388 return make_pair(found, 0);
391 // we're now actually ready to replace. if the buffer is
392 // read-only, we can't, though.
393 if (bv->buffer().isReadonly())
394 return make_pair(false, 0);
396 cap::replaceSelectionWithString(cur, replacestr);
398 cur.pos() += replacestr.length();
399 LASSERT(cur.pos() <= cur.lastpos(),
400 cur.pos() = cur.lastpos());
403 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
405 return make_pair(true, 1);
411 docstring const find2string(docstring const & search,
412 bool casesensitive, bool matchword, bool forward)
416 << int(casesensitive) << ' '
417 << int(matchword) << ' '
423 docstring const replace2string(docstring const & replace,
424 docstring const & search,
425 bool casesensitive, bool matchword,
426 bool all, bool forward, bool findnext)
429 ss << replace << '\n'
431 << int(casesensitive) << ' '
432 << int(matchword) << ' '
434 << int(forward) << ' '
440 bool lyxfind(BufferView * bv, FuncRequest const & ev)
442 if (!bv || ev.action() != LFUN_WORD_FIND)
445 //lyxerr << "find called, cmd: " << ev << endl;
447 // data is of the form
449 // <casesensitive> <matchword> <forward>"
451 docstring howto = split(ev.argument(), search, '\n');
453 bool casesensitive = parse_bool(howto);
454 bool matchword = parse_bool(howto);
455 bool forward = parse_bool(howto);
457 return findOne(bv, search, casesensitive, matchword, forward, true, true);
461 bool lyxreplace(BufferView * bv,
462 FuncRequest const & ev, bool has_deleted)
464 if (!bv || ev.action() != LFUN_WORD_REPLACE)
467 // data is of the form
470 // <casesensitive> <matchword> <all> <forward> <findnext>"
473 docstring howto = split(ev.argument(), rplc, '\n');
474 howto = split(howto, search, '\n');
476 bool casesensitive = parse_bool(howto);
477 bool matchword = parse_bool(howto);
478 bool all = parse_bool(howto);
479 bool forward = parse_bool(howto);
480 bool findnext = howto.empty() ? true : parse_bool(howto);
485 int replace_count = 0;
487 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
488 update = replace_count > 0;
491 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
493 replace_count = rv.second;
496 Buffer const & buf = bv->buffer();
498 // emit message signal.
499 buf.message(_("String not found."));
501 if (replace_count == 0) {
502 buf.message(_("String found."));
503 } else if (replace_count == 1) {
504 buf.message(_("String has been replaced."));
506 docstring const str =
507 bformat(_("%1$d strings have been replaced."), replace_count);
511 } else if (findnext) {
512 // if we have deleted characters, we do not replace at all, but
513 // rather search for the next occurence
514 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
517 bv->message(_("String not found."));
523 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
525 for (; cur; cur.forwardPos())
526 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
530 DocIterator cur_orig(bv->cursor());
531 docstring q = _("End of file reached while searching forward.\n"
532 "Continue searching from the beginning?");
533 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
534 q, 0, 1, _("&Yes"), _("&No"));
535 if (wrap_answer == 0) {
536 bv->cursor().clear();
537 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
538 bv->clearSelection();
539 cur.setCursor(bv->cursor().selectionBegin());
540 if (findNextChange(bv, cur, false))
543 bv->cursor().setCursor(cur_orig);
550 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
552 for (cur.backwardPos(); cur; cur.backwardPos()) {
553 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
558 DocIterator cur_orig(bv->cursor());
559 docstring q = _("Beginning of file reached while searching backward.\n"
560 "Continue searching from the end?");
561 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
562 q, 0, 1, _("&Yes"), _("&No"));
563 if (wrap_answer == 0) {
564 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
565 bv->cursor().backwardPos();
566 bv->clearSelection();
567 cur.setCursor(bv->cursor().selectionBegin());
568 if (findPreviousChange(bv, cur, false))
571 bv->cursor().setCursor(cur_orig);
578 bool selectChange(Cursor & cur, bool forward)
580 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
582 Change ch = cur.paragraph().lookupChange(cur.pos());
584 CursorSlice tip1 = cur.top();
585 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
586 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
587 if (!ch2.isSimilarTo(ch))
590 CursorSlice tip2 = cur.top();
591 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
593 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
594 if (!ch2.isSimilarTo(ch)) {
595 // take a step forward to correctly set the selection
603 cur.bv().mouseSetCursor(cur, false);
605 cur.bv().mouseSetCursor(cur, true);
613 bool findChange(BufferView * bv, bool forward)
616 cur.setCursor(forward ? bv->cursor().selectionEnd()
617 : bv->cursor().selectionBegin());
618 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
619 return selectChange(cur, forward);
624 bool findNextChange(BufferView * bv)
626 return findChange(bv, true);
630 bool findPreviousChange(BufferView * bv)
632 return findChange(bv, false);
639 typedef vector<pair<string, string> > Escapes;
641 /// A map of symbols and their escaped equivalent needed within a regex.
642 /// @note Beware of order
643 Escapes const & get_regexp_escapes()
645 typedef std::pair<std::string, std::string> P;
647 static Escapes escape_map;
648 if (escape_map.empty()) {
649 escape_map.push_back(P("$", "_x_$"));
650 escape_map.push_back(P("{", "_x_{"));
651 escape_map.push_back(P("}", "_x_}"));
652 escape_map.push_back(P("[", "_x_["));
653 escape_map.push_back(P("]", "_x_]"));
654 escape_map.push_back(P("(", "_x_("));
655 escape_map.push_back(P(")", "_x_)"));
656 escape_map.push_back(P("+", "_x_+"));
657 escape_map.push_back(P("*", "_x_*"));
658 escape_map.push_back(P(".", "_x_."));
659 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
660 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
661 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
662 escape_map.push_back(P("_x_", "\\"));
667 /// A map of lyx escaped strings and their unescaped equivalent.
668 Escapes const & get_lyx_unescapes()
670 typedef std::pair<std::string, std::string> P;
672 static Escapes escape_map;
673 if (escape_map.empty()) {
674 escape_map.push_back(P("\\%", "%"));
675 escape_map.push_back(P("\\mathcircumflex ", "^"));
676 escape_map.push_back(P("\\mathcircumflex", "^"));
677 escape_map.push_back(P("\\backslash ", "\\"));
678 escape_map.push_back(P("\\backslash", "\\"));
679 escape_map.push_back(P("\\\\{", "_x_<"));
680 escape_map.push_back(P("\\\\}", "_x_>"));
681 escape_map.push_back(P("\\sim ", "~"));
682 escape_map.push_back(P("\\sim", "~"));
687 /// A map of escapes turning a regexp matching text to one matching latex.
688 Escapes const & get_regexp_latex_escapes()
690 typedef std::pair<std::string, std::string> P;
692 static Escapes escape_map;
693 if (escape_map.empty()) {
694 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
695 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
696 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
697 escape_map.push_back(P("\\[", "\\{\\[\\}"));
698 escape_map.push_back(P("\\]", "\\{\\]\\}"));
699 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
700 escape_map.push_back(P("%", "\\\\\\%"));
701 escape_map.push_back(P("#", "\\\\#"));
706 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
707 ** the found occurrence were escaped.
709 string apply_escapes(string s, Escapes const & escape_map)
711 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
712 Escapes::const_iterator it;
713 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
714 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
715 unsigned int pos = 0;
716 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
717 s.replace(pos, it->first.length(), it->second);
718 LYXERR(Debug::FIND, "After escape: " << s);
719 pos += it->second.length();
720 // LYXERR(Debug::FIND, "pos: " << pos);
723 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
728 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
729 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
730 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
731 string escape_for_regex(string s, bool match_latex)
734 while (pos < s.size()) {
735 size_t new_pos = s.find("\\regexp{", pos);
736 if (new_pos == string::npos)
740 LYXERR(Debug::FIND, "new_pos: " << new_pos);
741 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
742 LYXERR(Debug::FIND, "t [lyx]: " << t);
743 t = apply_escapes(t, get_regexp_escapes());
744 LYXERR(Debug::FIND, "t [rxp]: " << t);
745 s.replace(pos, new_pos - pos, t);
746 new_pos = pos + t.size();
747 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
748 LYXERR(Debug::FIND, "new_pos: " << new_pos);
749 if (new_pos == s.size())
752 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
753 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
754 LYXERR(Debug::FIND, "end_pos: " << end_pos);
755 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
756 LYXERR(Debug::FIND, "t in regexp : " << t);
757 t = apply_escapes(t, get_lyx_unescapes());
758 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
760 t = apply_escapes(t, get_regexp_latex_escapes());
761 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
763 if (end_pos == s.size()) {
764 s.replace(new_pos, end_pos - new_pos, t);
765 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
768 s.replace(new_pos, end_pos + 13 - new_pos, t);
769 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
770 pos = new_pos + t.size();
771 LYXERR(Debug::FIND, "pos: " << pos);
777 /// Wrapper for lyx::regex_replace with simpler interface
778 bool regex_replace(string const & s, string & t, string const & searchstr,
779 string const & replacestr)
781 lyx::regex e(searchstr, regex_constants::ECMAScript);
783 ostream_iterator<char, char> it(oss);
784 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
785 // tolerate t and s be references to the same variable
786 bool rv = (s != oss.str());
792 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
794 ** Verify that closed braces exactly match open braces. This avoids that, for example,
795 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
798 ** Number of open braces that must remain open at the end for the verification to succeed.
800 bool braces_match(string::const_iterator const & beg,
801 string::const_iterator const & end,
805 string::const_iterator it = beg;
806 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
807 for (; it != end; ++it) {
808 // Skip escaped braces in the count
813 } else if (*it == '{') {
815 } else if (*it == '}') {
816 if (open_pars == 0) {
817 LYXERR(Debug::FIND, "Found unmatched closed brace");
823 if (open_pars != unmatched) {
824 LYXERR(Debug::FIND, "Found " << open_pars
825 << " instead of " << unmatched
826 << " unmatched open braces at the end of count");
829 LYXERR(Debug::FIND, "Braces match as expected");
839 MatchResult(): match_len(0),match2end(0), pos(0) {};
842 /** The class performing a match between a position in the document and the FindAdvOptions.
844 class MatchStringAdv {
846 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
848 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
849 ** constructor as opt.search, under the opt.* options settings.
852 ** If set, then match is searched only against beginning of text starting at cur.
853 ** If unset, then match is searched anywhere in text starting at cur.
856 ** The length of the matching text, or zero if no match was found.
858 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
863 /// first buffer on which search was started
864 lyx::Buffer * const p_first_buf;
866 FindAndReplaceOptions const & opt;
869 /// Auxiliary find method (does not account for opt.matchword)
870 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
872 /** Normalize a stringified or latexified LyX paragraph.
876 ** <li>if search is not casesensitive, then lowercase the string;
877 ** <li>remove any newline at begin or end of the string;
878 ** <li>replace any newline in the middle of the string with a simple space;
879 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
882 ** @todo Normalization should also expand macros, if the corresponding
883 ** search option was checked.
885 string normalize(docstring const & s, bool hack_braces) const;
886 // normalized string to search
887 string par_as_string;
888 // regular expression to use for searching
890 // same as regexp, but prefixed with a ".*?"
892 // leading format material as string
893 string lead_as_string;
894 // par_as_string after removal of lead_as_string
895 string par_as_string_nolead;
896 // unmatched open braces in the search string/regexp
898 // number of (.*?) subexpressions added at end of search regexp for closing
899 // environments, math mode, styles, etc...
901 // Are we searching with regular expressions ?
906 static docstring buffer_to_latex(Buffer & buffer)
908 //OutputParams runparams(&buffer.params().encoding());
909 OutputParams runparams(encodings.fromLyXName("utf8"));
910 odocstringstream ods;
912 runparams.nice = true;
913 runparams.flavor = OutputParams::XETEX;
914 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
915 // No side effect of file copying and image conversion
916 runparams.dryrun = true;
917 runparams.for_search = true;
918 pit_type const endpit = buffer.paragraphs().size();
919 for (pit_type pit = 0; pit != endpit; ++pit) {
920 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
921 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
927 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
930 if (!opt.ignoreformat) {
931 str = buffer_to_latex(buffer);
933 // OutputParams runparams(&buffer.params().encoding());
934 OutputParams runparams(encodings.fromLyXName("utf8"));
935 runparams.nice = true;
936 runparams.flavor = OutputParams::XETEX;
937 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
938 runparams.dryrun = true;
939 runparams.for_search = true;
940 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
941 Paragraph const & par = buffer.paragraphs().at(pit);
942 LYXERR(Debug::FIND, "Adding to search string: '"
943 << par.asString(pos_type(0), par.size(),
944 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
947 str += par.asString(pos_type(0), par.size(),
948 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
956 /// Return separation pos between the leading material and the rest
957 static size_t identifyLeading(string const & s)
960 // @TODO Support \item[text]
961 // Kornel: Added textsl, textsf, textit, texttt and noun
962 // + allow to search for colored text too
963 while (regex_replace(t, t, REGEX_BOS "\\\\(((footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|emph|noun|minisec|text(bf|md|sl|sf|it|tt))|((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
964 || regex_replace(t, t, REGEX_BOS "\\$", "")
965 || regex_replace(t, t, REGEX_BOS "\\\\\\[", "")
966 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
967 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
969 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
974 * Given a latexified string, retrieve some handled features
975 * The features of the regex will later be compared with the features
976 * of the searched text. If the regex features are not a
977 * subset of the analized, then, in not format ignoring search
978 * we can early stop the search in the relevant inset.
980 typedef map<string, bool> Features;
982 static Features identifyFeatures(string const & s)
984 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
985 static regex const valid("^("
987 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
988 "emph|noun|text(bf|md|sl|sf|it|tt)|"
989 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
990 "(u|uu)line|(s|x)out|uwave|"
991 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
992 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
997 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1000 if (sub.str(1).compare("regexp") == 0) {
1004 string token = sub.str(1);
1006 if (regex_match(token, sub2, valid)) {
1014 if (sub.str(1).compare("endregexp") == 0) {
1024 * defines values features of a key "\\[a-z]+{"
1029 /* Char type with content discarded
1030 * like \hspace{1cm} */
1032 /* Char, like \backslash */
1034 /* \part, \section*, ... */
1036 /* title, author etc */
1038 /* \foreignlanguage{ngerman}, ... */
1041 * to discard language in content */
1044 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1046 /* fonts, colors, markups, ... */
1048 /* footnotesize, ... large, ...
1049 * Ignore all of them */
1052 /* inputencoding, ...
1053 * Discard also content, because they do not help in search */
1056 * like remove, but also all arguments */
1058 /* item, listitem */
1060 /* tex, latex, ... like isChar */
1062 /* like \lettrine[lines=5]{}{} */
1064 /* End of arguments marker for lettrine,
1065 * so that they can be ignored */
1075 parenthesiscount(1),
1079 KeyInfo(KeyType type, int parcount, bool disable)
1085 parenthesiscount(parcount),
1094 int parenthesiscount;
1096 bool used; /* by pattern */
1101 Border(int l=0, int u=0) : low(l), upper(u) {};
1106 #define MAXOPENED 30
1108 bool isPatternString_;
1110 explicit Intervall(bool isPattern, string const & p) :
1111 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1112 hasTitle(false), langcount(0)
1120 static vector<Border> borders;
1121 int depts[MAXOPENED];
1122 int closes[MAXOPENED];
1123 int actualdeptindex;
1124 int previousNotIgnored(int);
1125 int nextNotIgnored(int);
1126 void handleOpenP(int i);
1127 void handleCloseP(int i, bool closingAllowed);
1128 void resetOpenedP(int openPos);
1129 void addIntervall(int upper);
1130 void addIntervall(int low, int upper); /* if explicit */
1131 void removeAccents();
1132 void setForDefaultLang(KeyInfo const & defLang);
1133 int findclosing(int start, int end, char up, char down, int repeat);
1134 void handleParentheses(int lastpos, bool closingAllowed);
1136 int langcount; // Number of disabled language specs up to current position in actual interval
1137 int isOpeningPar(int pos);
1139 void output(ostringstream &os, int lastpos);
1140 // string show(int lastpos);
1143 vector<Border> Intervall::borders = vector<Border>(30);
1145 int Intervall::isOpeningPar(int pos)
1147 if ((pos < 0) || (size_t(pos) >= par.size()))
1149 if (par[pos] != '{')
1151 if (size_t(pos) + 2 >= par.size())
1153 if (par[pos+2] != '}')
1155 if (par[pos+1] == '[' || par[pos+1] == ']')
1160 void Intervall::setForDefaultLang(KeyInfo const & defLang)
1162 // Enable the use of first token again
1163 if (ignoreidx >= 0) {
1164 int value = defLang._tokenstart + defLang._tokensize;
1166 if (borders[0].low < value)
1167 borders[0].low = value;
1168 if (borders[0].upper < value)
1169 borders[0].upper = value;
1174 static void checkDepthIndex(int val)
1176 static int maxdepthidx = MAXOPENED-2;
1177 static int lastmaxdepth = 0;
1178 if (val > lastmaxdepth) {
1179 LYXERR(Debug::INFO, "Depth reached " << val);
1182 if (val > maxdepthidx) {
1184 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1189 // Not needed, because borders are now dynamically expanded
1190 static void checkIgnoreIdx(int val)
1192 static int lastmaxignore = -1;
1193 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1194 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1195 lastmaxignore = val;
1201 * Expand the region of ignored parts of the input latex string
1202 * The region is only relevant in output()
1204 void Intervall::addIntervall(int low, int upper)
1207 if (low == upper) return;
1208 for (idx = ignoreidx+1; idx > 0; --idx) {
1209 if (low > borders[idx-1].upper) {
1213 Border br(low, upper);
1214 if (idx > ignoreidx) {
1215 if (borders.size() <= size_t(idx)) {
1216 borders.push_back(br);
1222 // checkIgnoreIdx(ignoreidx);
1226 // Expand only if one of the new bound is inside the interwall
1227 // We know here that br.low > borders[idx-1].upper
1228 if (br.upper < borders[idx].low) {
1229 // We have to insert at this pos
1230 if (size_t(ignoreidx+1) >= borders.size()) {
1231 borders.push_back(borders[ignoreidx]);
1234 borders[ignoreidx+1] = borders[ignoreidx];
1236 for (int i = ignoreidx; i > idx; --i) {
1237 borders[i] = borders[i-1];
1241 // checkIgnoreIdx(ignoreidx);
1244 // Here we know, that we are overlapping
1245 if (br.low > borders[idx].low)
1246 br.low = borders[idx].low;
1247 // check what has to be concatenated
1249 for (int i = idx; i <= ignoreidx; i++) {
1250 if (br.upper >= borders[i].low) {
1252 if (br.upper < borders[i].upper)
1253 br.upper = borders[i].upper;
1259 // count should be >= 1 here
1262 for (int i = idx + count; i <= ignoreidx; i++) {
1263 borders[i-count+1] = borders[i];
1265 ignoreidx -= count - 1;
1271 typedef map<string, string> AccentsMap;
1272 static AccentsMap accents = map<string, string>();
1274 static void buildaccent(string n, string param, string values)
1278 const char delim = '|';
1279 while (getline(s, name, delim)) {
1281 for (char c : param) {
1282 string key = name + "{" + c + "}";
1283 // get the corresponding utf8-value
1284 if ((values[start] & 0xc0) != 0xc0) {
1285 // should not happen, utf8 encoding starts at least with 11xxxxxx
1286 // but value for '\dot{i}' is 'i', which is ascii
1287 if ((values[start] & 0x80) == 0) {
1289 accents[key] = values.substr(start, 1);
1290 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1295 for (int j = 1; ;j++) {
1296 if (start + j >= values.size()) {
1297 accents[key] = values.substr(start, j);
1298 start = values.size() - 1;
1301 else if ((values[start+j] & 0xc0) != 0x80) {
1302 // This is the first byte of following utf8 char
1303 accents[key] = values.substr(start, j);
1305 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1313 static void buildAccentsMap()
1315 accents["imath"] = "ı";
1317 accents["jmath"] = "ȷ";
1318 accents["lyxmathsym{ß}"] = "ß";
1319 accents["text{ß}"] = "ß";
1320 accents["ddot{\\imath}"] = "ï";
1321 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1322 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1323 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1324 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1325 accents["acute{\\imath}"] = "í";
1326 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1327 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1328 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1329 buildaccent("mathring|r", "aAuUwy",
1331 accents["check{\\imath}"] = "ǐ";
1332 accents["check{\\jmath}"] = "ǰ";
1333 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1334 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1335 accents["hat{\\imath}"] = "î";
1336 accents["hat{\\jmath}"] = "ĵ";
1337 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1338 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1339 accents["bar{\\imath}"] = "ī";
1340 buildaccent("bar|=", "aAeEiIoOuUyY",
1341 "āĀēĒīĪōŌūŪȳȲ"); // macron
1342 accents["tilde{\\imath}"] = "ĩ";
1343 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1344 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1345 accents["breve{\\imath}"] = "ĭ";
1346 buildaccent("breve|u", "aAeEgGiIoOuU",
1347 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1348 accents["grave{\\imath}"] = "ì";
1349 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1350 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1351 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1352 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1353 buildaccent("ogonek|k", "AaEeIiUuOo",
1354 "ĄąĘęĮįŲųǪǫ"); // ogonek
1355 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1356 "ÇçĢĢĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1357 buildaccent("subring|textsubring", "Aa",
1359 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1360 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1361 buildaccent("subtilde|textsubtilde", "EeIiUu",
1362 "ḚḛḬḭṴṵ"); // subtilde
1363 accents["dgrave{\\imath}"] = "ȉ";
1364 accents["textdoublegrave{\\i}"] = "ȉ";
1365 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1366 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1367 accents["rcap{\\imath}"] = "ȉ";
1368 accents["textroundcap{\\i}"] = "ȉ";
1369 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1370 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1371 buildaccent("slashed", "oO",
1376 * Created accents in math or regexp environment
1377 * are macros, but we need the utf8 equivalent
1379 void Intervall::removeAccents()
1381 if (accents.empty())
1383 static regex const accre("\\\\(([\\S]|grave|breve|lyxmathsym|text|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}|(i|imath|jmath)(?![a-zA-Z]))");
1385 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1387 string key = sub.str(1);
1388 if (accents.find(key) != accents.end()) {
1389 string val = accents[key];
1390 size_t pos = sub.position(size_t(0));
1391 for (size_t i = 0; i < val.size(); i++) {
1392 par[pos+i] = val[i];
1394 addIntervall(pos+val.size(), pos + sub.str(0).size());
1395 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1396 // remove traces of any remaining chars
1401 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1406 void Intervall::handleOpenP(int i)
1409 depts[actualdeptindex] = i+1;
1410 closes[actualdeptindex] = -1;
1411 checkDepthIndex(actualdeptindex);
1414 void Intervall::handleCloseP(int i, bool closingAllowed)
1416 if (actualdeptindex <= 0) {
1417 if (! closingAllowed)
1418 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1419 // if we are at the very end
1420 addIntervall(i, i+1);
1423 closes[actualdeptindex] = i+1;
1428 void Intervall::resetOpenedP(int openPos)
1430 // Used as initializer for foreignlanguage entry
1431 actualdeptindex = 1;
1432 depts[1] = openPos+1;
1436 int Intervall::previousNotIgnored(int start)
1438 int idx = 0; /* int intervalls */
1439 for (idx = ignoreidx; idx >= 0; --idx) {
1440 if (start > borders[idx].upper)
1442 if (start >= borders[idx].low)
1443 start = borders[idx].low-1;
1448 int Intervall::nextNotIgnored(int start)
1450 int idx = 0; /* int intervalls */
1451 for (idx = 0; idx <= ignoreidx; idx++) {
1452 if (start < borders[idx].low)
1454 if (start < borders[idx].upper)
1455 start = borders[idx].upper;
1460 typedef map<string, KeyInfo> KeysMap;
1461 typedef vector< KeyInfo> Entries;
1462 static KeysMap keys = map<string, KeyInfo>();
1468 Intervall interval_;
1469 void buildKeys(bool);
1470 void buildEntries(bool);
1471 void makeKey(const string &, KeyInfo, bool isPatternString);
1472 void processRegion(int start, int region_end); /* remove {} parts */
1473 void removeHead(KeyInfo const &, int count=0);
1476 LatexInfo(string const & par, bool isPatternString)
1477 : entidx_(-1), interval_(isPatternString, par)
1479 buildKeys(isPatternString);
1480 entries_ = vector<KeyInfo>();
1481 buildEntries(isPatternString);
1485 if (entries_.empty()) {
1488 if (entries_[0].keytype == KeyInfo::isTitle) {
1489 if (! entries_[0].disabled) {
1490 interval_.hasTitle = true;
1491 interval_.titleValue = entries_[0].head;
1494 interval_.hasTitle = false;
1495 interval_.titleValue = "";
1497 removeHead(entries_[0]);
1498 if (entries_.size() > 1)
1507 if (int(entries_.size()) > entidx_) {
1514 bool setNextKey(int idx) {
1515 if ((idx == entidx_) && (entidx_ >= 0)) {
1522 int find(int start, KeyInfo::KeyType keytype) {
1526 while (tmpIdx < int(entries_.size())) {
1527 if (entries_[tmpIdx].keytype == keytype)
1533 int process(ostringstream & os, KeyInfo const & actual);
1534 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1535 // string show(int lastpos) { return interval.show(lastpos);};
1536 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1537 KeyInfo &getKeyInfo(int keyinfo) {
1538 static KeyInfo invalidInfo = KeyInfo();
1539 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1542 return entries_[keyinfo];
1544 void setForDefaultLang(KeyInfo &defLang) {interval_.setForDefaultLang(defLang);};
1545 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1549 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1553 for (int i = start; i < end; i += 1 + skip) {
1557 if (c == '\\') skip = 1;
1561 else if (c == down) {
1564 if ((repeat <= 0) || (par[i+1] != up))
1582 vector<MathEntry> entries_;
1587 void insert(string const & wait, size_t start, size_t end) {
1588 MathEntry m = MathEntry();
1590 m.mathStart = start;
1592 m.mathSize = end - start;
1593 entries_.push_back(m);
1595 bool empty() { return entries_.empty(); };
1596 size_t getEndPos() {
1597 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1600 return entries_[actualIdx_].mathEnd;
1602 size_t getStartPos() {
1603 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1604 return 100000; /* definitely enough? */
1606 return entries_[actualIdx_].mathStart;
1608 size_t getFirstPos() {
1610 return getStartPos();
1613 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1616 return entries_[actualIdx_].mathSize;
1618 void incrEntry() { actualIdx_++; };
1621 void LatexInfo::buildEntries(bool isPatternString)
1623 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1624 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1625 static bool disableLanguageOverride = false;
1626 smatch sub, submath;
1627 bool evaluatingRegexp = false;
1629 bool evaluatingMath = false;
1630 bool evaluatingCode = false;
1632 bool evaluatingOptional = false;
1633 size_t optionalEnd = 0;
1636 bool math_end_waiting = false;
1637 size_t math_pos = 10000;
1640 interval_.removeAccents();
1642 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1644 if (math_end_waiting) {
1645 size_t pos = submath.position(size_t(0));
1646 if ((math_end == "$") &&
1647 (submath.str(0) == "$") &&
1648 (interval_.par[pos-1] != '\\')) {
1649 mi.insert("$", math_pos, pos + 1);
1650 math_end_waiting = false;
1652 else if ((math_end == "\\]") &&
1653 (submath.str(0) == "\\]")) {
1654 mi.insert("\\]", math_pos, pos + 2);
1655 math_end_waiting = false;
1657 else if ((submath.str(1).compare("end") == 0) &&
1658 (submath.str(2).compare(math_end) == 0)) {
1659 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1660 math_end_waiting = false;
1666 if (submath.str(1).compare("begin") == 0) {
1667 math_end_waiting = true;
1668 math_end = submath.str(2);
1669 math_pos = submath.position(size_t(0));
1671 else if (submath.str(0).compare("\\[") == 0) {
1672 math_end_waiting = true;
1674 math_pos = submath.position(size_t(0));
1676 else if (submath.str(0) == "$") {
1677 size_t pos = submath.position(size_t(0));
1678 if ((pos == 0) || (interval_.par[pos-1] != '\\')) {
1679 math_end_waiting = true;
1686 // Ignore language if there is math somewhere in pattern-string
1687 if (isPatternString) {
1690 keys["foreignlanguage"].disabled = true;
1691 disableLanguageOverride = true;
1694 disableLanguageOverride = false;
1697 if (disableLanguageOverride) {
1698 keys["foreignlanguage"].disabled = true;
1701 math_pos = mi.getFirstPos();
1702 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1704 string key = sub.str(3);
1706 if (sub.str(0)[0] == '\\')
1707 key = sub.str(0)[1];
1711 size_t k_pos = sub.position(size_t(0));
1712 if ((k_pos > 0) && (interval_.par[k_pos - 1] == '\\')) {
1713 // Escaped '$', ignoring
1719 if (evaluatingRegexp) {
1720 if (sub.str(1).compare("endregexp") == 0) {
1721 evaluatingRegexp = false;
1722 // found._tokenstart already set
1723 found._dataEnd = sub.position(size_t(0)) + 13;
1724 found._dataStart = found._dataEnd;
1725 found._tokensize = found._dataEnd - found._tokenstart;
1726 found.parenthesiscount = 0;
1727 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1734 if (evaluatingMath) {
1735 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1737 evaluatingMath = false;
1739 math_pos = mi.getStartPos();
1741 if (keys.find(key) == keys.end()) {
1742 found = KeyInfo(KeyInfo::isStandard, 0, true);
1743 if (isPatternString) {
1744 found.keytype = KeyInfo::isChar;
1745 found.disabled = false;
1752 if (key.compare("regexp") == 0) {
1753 evaluatingRegexp = true;
1754 found._tokenstart = sub.position(size_t(0));
1755 found._tokensize = 0;
1759 // Handle the other params of key
1760 if (found.keytype == KeyInfo::isIgnored)
1762 else if (found.keytype == KeyInfo::isMath) {
1763 if (size_t(sub.position(size_t(0))) == math_pos) {
1765 found._tokenstart = sub.position(size_t(0));
1766 found._tokensize = mi.getSize();
1767 found._dataEnd = found._tokenstart + found._tokensize;
1768 found._dataStart = found._dataEnd;
1769 found.parenthesiscount = 0;
1770 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1771 evaluatingMath = true;
1774 // begin|end of unknown env, discard
1775 // First handle tables
1776 // longtable|tabular
1777 bool discardComment;
1779 found.keytype = KeyInfo::doRemove;
1780 if ((sub.str(5).compare("longtable") == 0) ||
1781 (sub.str(5).compare("tabular") == 0)) {
1782 discardComment = true; /* '%' */
1785 discardComment = false;
1786 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1788 string token = sub.str(5);
1789 if (regex_match(token, sub2, removeArgs)) {
1790 found.keytype = KeyInfo::removeWithArg;
1793 // discard spaces before pos(0)
1794 int pos = sub.position(size_t(0));
1796 for (count = 0; pos - count > 0; count++) {
1797 char c = interval_.par[pos-count-1];
1798 if (discardComment) {
1799 if ((c != ' ') && (c != '%'))
1805 found._tokenstart = pos - count;
1806 if (sub.str(1).compare(0, 5, "begin") == 0) {
1807 size_t pos1 = pos + sub.str(0).length();
1808 if (sub.str(5).compare("cjk") == 0) {
1809 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1810 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1812 found.keytype = KeyInfo::isMain;
1813 found._dataStart = pos1;
1814 found._dataEnd = interval_.par.length();
1815 found.disabled = keys["foreignlanguage"].disabled;
1816 found.used = keys["foreignlanguage"].used;
1817 found._tokensize = pos1 - found._tokenstart;
1818 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1821 // Swallow possible optional params
1822 while (interval_.par[pos1] == '[') {
1823 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1825 // Swallow also the eventual parameter
1826 if (interval_.par[pos1] == '{') {
1827 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1830 found._dataEnd = pos1;
1832 found._dataStart = found._dataEnd;
1833 found._tokensize = count + found._dataEnd - pos;
1834 found.parenthesiscount = 0;
1835 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1836 found.disabled = true;
1840 // Handle "\end{...}"
1841 found._dataStart = pos + sub.str(0).length();
1842 found._dataEnd = found._dataStart;
1843 found._tokensize = count + found._dataEnd - pos;
1844 found.parenthesiscount = 0;
1845 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1846 found.disabled = true;
1850 else if (found.keytype != KeyInfo::isRegex) {
1851 found._tokenstart = sub.position(size_t(0));
1852 if (found.parenthesiscount == 0) {
1853 // Probably to be discarded
1854 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1855 char following = interval_.par[following_pos];
1856 if (following == ' ')
1857 found.head = "\\" + sub.str(3) + " ";
1858 else if (following == '=') {
1859 // like \uldepth=1000pt
1860 found.head = sub.str(0);
1863 found.head = "\\" + key;
1864 found._tokensize = found.head.length();
1865 found._dataEnd = found._tokenstart + found._tokensize;
1866 found._dataStart = found._dataEnd;
1869 int params = found._tokenstart + key.length() + 1;
1870 if (evaluatingOptional) {
1871 if (size_t(found._tokenstart) > optionalEnd) {
1872 evaluatingOptional = false;
1875 found.disabled = true;
1878 int optend = params;
1879 while (interval_.par[optend] == '[') {
1880 // discard optional parameters
1881 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
1883 if (optend > params) {
1884 key += interval_.par.substr(params, optend-params);
1885 evaluatingOptional = true;
1886 optionalEnd = optend;
1888 string token = sub.str(5);
1889 int closings = found.parenthesiscount;
1890 if (found.parenthesiscount == 1) {
1891 found.head = "\\" + key + "{";
1893 else if (found.parenthesiscount > 1) {
1895 found.head = sub.str(0) + "{";
1896 closings = found.parenthesiscount - 1;
1899 found.head = "\\" + key + "{";
1902 found._tokensize = found.head.length();
1903 found._dataStart = found._tokenstart + found.head.length();
1904 if (found.keytype == KeyInfo::doRemove) {
1905 int endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1906 found._dataStart = endpar;
1907 found._tokensize = found._dataStart - found._tokenstart;
1910 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1911 found._dataStart += 15;
1915 endpos = found._dataStart - 1;
1917 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1918 if (found.keytype == KeyInfo::isList) {
1919 // Check if it really is list env
1920 static regex const listre("^([a-z]+)$");
1922 if (!regex_match(token, sub2, listre)) {
1923 // Change the key of this entry. It is not in a list/item environment
1924 found.keytype = KeyInfo::endArguments;
1927 if (found.keytype == KeyInfo::noMain) {
1928 evaluatingCode = true;
1930 codeStart = found._dataStart;
1932 else if (evaluatingCode) {
1933 if (size_t(found._dataStart) > codeEnd)
1934 evaluatingCode = false;
1935 else if (found.keytype == KeyInfo::isMain) {
1936 // Disable this key, treate it as standard
1937 found.keytype = KeyInfo::isStandard;
1938 found.disabled = true;
1939 if ((codeEnd == interval_.par.length()) &&
1940 (found._tokenstart == codeStart)) {
1941 // trickery, because the code inset starts
1942 // with \selectlanguage ...
1944 if (entries_.size() > 1) {
1945 entries_[entries_.size()-1]._dataEnd = codeEnd;
1950 if ((endpos == interval_.par.length()) &&
1951 (found.keytype == KeyInfo::doRemove)) {
1952 // Missing closing => error in latex-input?
1953 // therefore do not delete remaining data
1954 found._dataStart -= 1;
1955 found._dataEnd = found._dataStart;
1958 found._dataEnd = endpos;
1960 if (isPatternString) {
1961 keys[key].used = true;
1964 entries_.push_back(found);
1968 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1970 stringstream s(keysstring);
1972 const char delim = '|';
1973 while (getline(s, key, delim)) {
1974 KeyInfo keyII(keyI);
1975 if (isPatternString) {
1978 else if ( !keys[key].used)
1979 keyII.disabled = true;
1984 void LatexInfo::buildKeys(bool isPatternString)
1987 static bool keysBuilt = false;
1988 if (keysBuilt && !isPatternString) return;
1990 // Known standard keys with 1 parameter.
1991 // Split is done, if not at start of region
1992 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1993 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1994 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1995 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1996 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1997 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
1999 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2000 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2001 makeKey("section*|subsection*|subsubsection*|paragraph*",
2002 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2003 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2004 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2006 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2008 // Split is done, if not at start of region
2009 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2010 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2012 // Split is done always.
2013 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2017 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2018 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2019 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2020 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2022 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2023 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2024 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2026 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2027 // Custom space/skip, remove the content (== length value)
2028 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2029 // Found in fr/UserGuide.lyx
2030 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2032 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2033 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2034 // Known macros to remove (including their parameter)
2036 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2037 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2038 // handle like standard keys with 1 parameter.
2039 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2041 // Ignore deleted text
2042 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2043 // but preserve added text
2044 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2046 // Macros to remove, but let the parameter survive
2048 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2050 // Remove language spec from content of these insets
2051 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2053 // Same effect as previous, parameter will survive (because there is no one anyway)
2055 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2056 // Remove table decorations
2057 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2058 // Discard shape-header.
2059 // For footnote or shortcut too, because of lang settings
2060 // and wrong handling if used 'KeyInfo::noMain'
2061 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2062 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2063 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2064 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2065 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2066 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2067 // like ('tiny{}' or '\tiny ' ... )
2068 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2070 // Survives, like known character
2071 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2072 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2074 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2075 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2076 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2078 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2079 // Remove RTL/LTR marker
2080 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2081 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2082 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2083 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2084 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2085 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2086 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2087 if (isPatternString) {
2088 // Allow the first searched string to rebuild the keys too
2092 // no need to rebuild again
2098 * Keep the list of actual opened parentheses actual
2099 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2101 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2104 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2108 if (c == '\\') skip = 1;
2109 else if (c == '{') {
2112 else if (c == '}') {
2113 handleCloseP(i, closingAllowed);
2119 string Intervall::show(int lastpos)
2121 int idx = 0; /* int intervalls */
2124 for (idx = 0; idx <= ignoreidx; idx++) {
2125 while (i < lastpos) {
2127 if (i <= borders[idx].low) {
2128 if (borders[idx].low > lastpos)
2129 printsize = lastpos - i;
2131 printsize = borders[idx].low - i;
2132 s += par.substr(i, printsize);
2134 if (i >= borders[idx].low)
2135 i = borders[idx].upper;
2138 i = borders[idx].upper;
2144 s += par.substr(i, lastpos-i);
2150 void Intervall::output(ostringstream &os, int lastpos)
2152 // get number of chars to output
2153 int idx = 0; /* int intervalls */
2156 string startTitle = titleValue;
2157 for (idx = 0; idx <= ignoreidx; idx++) {
2159 if (i <= borders[idx].low) {
2161 if (borders[idx].low > lastpos)
2162 printsize = lastpos - i;
2164 printsize = borders[idx].low - i;
2165 if (printsize > 0) {
2166 os << startTitle << par.substr(i, printsize);
2168 printed += printsize;
2171 handleParentheses(i, false);
2172 if (i >= borders[idx].low)
2173 i = borders[idx].upper;
2176 i = borders[idx].upper;
2183 os << startTitle << par.substr(i, lastpos-i);
2184 printed += lastpos-i;
2186 handleParentheses(lastpos, false);
2188 if (keys["foreignlanguage"].disabled)
2189 startindex = actualdeptindex-langcount;
2191 startindex = actualdeptindex;
2192 for (int i = startindex; i > 0; --i) {
2195 if (hasTitle && (printed > 0))
2197 if (! isPatternString_)
2199 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2202 void LatexInfo::processRegion(int start, int region_end)
2204 while (start < region_end) { /* Let {[} and {]} survive */
2205 int cnt = interval_.isOpeningPar(start);
2207 // Closing is allowed past the region
2208 int closing = interval_.findclosing(start+1, interval_.par.length());
2209 interval_.addIntervall(start, start+1);
2210 interval_.addIntervall(closing, closing+1);
2214 start = interval_.nextNotIgnored(start+1);
2218 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2220 if (actual.parenthesiscount == 0) {
2221 // "{\tiny{} ...}" ==> "{{} ...}"
2222 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2225 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2226 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2227 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2231 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2234 switch (actual.keytype)
2236 case KeyInfo::isTitle: {
2238 nextKeyIdx = getNextKey();
2241 case KeyInfo::cleanToStart: {
2242 actual._dataEnd = actual._dataStart;
2243 nextKeyIdx = getNextKey();
2244 // Search for end of arguments
2245 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2247 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2248 entries_[i].disabled = true;
2250 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2252 while (interval_.par[actual._dataEnd] == ' ')
2254 interval_.addIntervall(0, actual._dataEnd+1);
2255 interval_.actualdeptindex = 0;
2256 interval_.depts[0] = actual._dataEnd+1;
2257 interval_.closes[0] = -1;
2260 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2261 if (actual.disabled)
2262 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2264 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2267 case KeyInfo::isChar: {
2268 nextKeyIdx = getNextKey();
2271 case KeyInfo::isSize: {
2272 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2273 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2274 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2275 nextKeyIdx = getNextKey();
2277 // Here _dataStart points to '{', so correct it
2278 actual._dataStart += 1;
2279 actual._tokensize += 1;
2280 actual.parenthesiscount = 1;
2281 if (interval_.par[actual._dataStart] == '}') {
2282 // Determine the end if used like '{\tiny{}...}'
2283 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2284 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2287 // Determine the end if used like '\tiny{...}'
2288 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2290 // Split on this key if not at start
2291 int start = interval_.nextNotIgnored(previousStart);
2292 if (start < actual._tokenstart) {
2293 interval_.output(os, actual._tokenstart);
2294 interval_.addIntervall(start, actual._tokenstart);
2296 // discard entry if at end of actual
2297 nextKeyIdx = process(os, actual);
2301 case KeyInfo::endArguments:
2302 // Remove trailing '{}' too
2303 actual._dataStart += 1;
2304 actual._dataEnd += 1;
2305 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2306 nextKeyIdx = getNextKey();
2308 case KeyInfo::noMain:
2310 case KeyInfo::isStandard: {
2311 if (actual.disabled) {
2313 processRegion(actual._dataStart, actual._dataStart+1);
2314 nextKeyIdx = getNextKey();
2316 // Split on this key if not at datastart of calling entry
2317 int start = interval_.nextNotIgnored(previousStart);
2318 if (start < actual._tokenstart) {
2319 interval_.output(os, actual._tokenstart);
2320 interval_.addIntervall(start, actual._tokenstart);
2322 // discard entry if at end of actual
2323 nextKeyIdx = process(os, actual);
2327 case KeyInfo::removeWithArg: {
2328 nextKeyIdx = getNextKey();
2329 // Search for end of arguments
2330 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2332 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2333 entries_[i].disabled = true;
2335 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2337 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2340 case KeyInfo::doRemove: {
2341 // Remove the key with all parameters and following spaces
2343 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2344 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2347 // Remove also enclosing parentheses [] and {}
2350 while (actual._tokenstart > numpars) {
2351 if (pos+numpars >= interval_.par.size())
2353 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2355 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2361 if (interval_.par[pos+numpars] == ' ')
2365 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2366 nextKeyIdx = getNextKey();
2369 case KeyInfo::isList: {
2370 // Discard space before _tokenstart
2372 for (count = 0; count < actual._tokenstart; count++) {
2373 if (interval_.par[actual._tokenstart-count-1] != ' ')
2376 nextKeyIdx = getNextKey();
2377 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2379 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2381 // How else can we catch this one?
2382 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2383 entries_[i].disabled = true;
2385 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2387 else if (nextKeyIdx > 0) {
2388 // Ignore any lang entries inside data region
2389 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2390 if (entries_[i].keytype == KeyInfo::isMain)
2391 entries_[i].disabled = true;
2394 if (actual.disabled) {
2395 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2398 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2400 if (interval_.par[actual._dataEnd+1] == '[') {
2401 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2402 if ((interval_.par[actual._dataEnd+2] == '{') &&
2403 (interval_.par[posdown-1] == '}')) {
2404 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2405 interval_.addIntervall(posdown-1, posdown+1);
2408 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2409 interval_.addIntervall(posdown, posdown+1);
2411 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2412 if (blk > posdown) {
2413 // Discard at most 1 space after empty item
2415 for (count = 0; count < 1; count++) {
2416 if (interval_.par[blk+count] != ' ')
2420 interval_.addIntervall(blk, blk+count);
2425 case KeyInfo::isSectioning: {
2426 // Discard spaces before _tokenstart
2428 int val = actual._tokenstart;
2429 for (count = 0; count < actual._tokenstart;) {
2430 val = interval_.previousNotIgnored(val-1);
2431 if (val < 0 || interval_.par[val] != ' ')
2434 count = actual._tokenstart - val;
2437 if (actual.disabled) {
2438 removeHead(actual, count);
2439 nextKeyIdx = getNextKey();
2441 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2442 nextKeyIdx = process(os, actual);
2446 case KeyInfo::isMath: {
2447 // Same as regex, use the content unchanged
2448 nextKeyIdx = getNextKey();
2451 case KeyInfo::isRegex: {
2452 // DO NOT SPLIT ON REGEX
2454 nextKeyIdx = getNextKey();
2457 case KeyInfo::isIgnored: {
2458 // Treat like a character for now
2459 nextKeyIdx = getNextKey();
2462 case KeyInfo::isMain: {
2463 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2464 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2465 if (actual._tokenstart > 0) {
2466 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2467 if ((prev >= 0) && interval_.par[prev] == '%')
2468 interval_.addIntervall(prev, prev+1);
2470 if (actual.disabled) {
2472 interval_.langcount++;
2473 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2474 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2475 // Discard also the space before math-equation
2476 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2478 nextKeyIdx = getNextKey();
2479 // interval.resetOpenedP(actual._dataStart-1);
2482 if (actual._tokenstart < 26) {
2483 // for the first (and maybe dummy) language
2484 interval_.setForDefaultLang(actual);
2486 interval_.resetOpenedP(actual._dataStart-1);
2490 case KeyInfo::invalid:
2491 // This cannot happen, already handled
2494 // LYXERR(Debug::INFO, "Unhandled keytype");
2495 nextKeyIdx = getNextKey();
2502 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2504 int end = interval_.nextNotIgnored(actual._dataEnd);
2505 int oldStart = actual._dataStart;
2506 int nextKeyIdx = getNextKey();
2508 if ((nextKeyIdx < 0) ||
2509 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2510 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2511 if (oldStart <= end) {
2512 processRegion(oldStart, end);
2517 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2519 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2520 (void) dispatch(os, actual._dataStart, nextKey);
2521 end = nextKey._tokenstart;
2524 processRegion(oldStart, nextKey._tokenstart);
2525 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2527 oldStart = nextKey._dataEnd+1;
2529 // now nextKey is either invalid or is outside of actual._dataEnd
2530 // output the remaining and discard myself
2531 if (oldStart <= end) {
2532 processRegion(oldStart, end);
2534 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2536 // This is the normal case.
2537 // But if using the firstlanguage, the closing may be missing
2539 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2541 if (actual._dataEnd < end)
2542 output_end = interval_.nextNotIgnored(actual._dataEnd);
2544 output_end = interval_.nextNotIgnored(end);
2545 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2546 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2548 // Remove possible empty data
2549 int dstart = interval_.nextNotIgnored(actual._dataStart);
2550 while (interval_.isOpeningPar(dstart) == 1) {
2551 interval_.addIntervall(dstart, dstart+1);
2552 int dend = interval_.findclosing(dstart+1, output_end);
2553 interval_.addIntervall(dend, dend+1);
2554 dstart = interval_.nextNotIgnored(dstart+1);
2556 if (dstart < output_end)
2557 interval_.output(os, output_end);
2558 interval_.addIntervall(actual._tokenstart, end);
2562 string splitOnKnownMacros(string par, bool isPatternString)
2565 LatexInfo li(par, isPatternString);
2566 // LYXERR(Debug::INFO, "Berfore split: " << par);
2567 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2569 DummyKey._tokensize = 0;
2570 DummyKey._dataStart = 0;
2571 DummyKey._dataEnd = par.length();
2572 DummyKey.disabled = true;
2573 int firstkeyIdx = li.getFirstKey();
2575 if (firstkeyIdx >= 0) {
2576 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2577 DummyKey._tokenstart = firstKey._tokenstart;
2579 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2580 // Use dummy firstKey
2581 firstKey = DummyKey;
2582 (void) li.setNextKey(firstkeyIdx);
2585 if (par.substr(firstKey._dataStart, 2) == "% ")
2586 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2588 nextkeyIdx = li.process(os, firstKey);
2589 while (nextkeyIdx >= 0) {
2590 // Check for a possible gap between the last
2591 // entry and this one
2592 int datastart = li.nextNotIgnored(firstKey._dataStart);
2593 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2594 if ((nextKey._tokenstart > datastart)) {
2596 firstKey._dataStart = datastart;
2597 firstKey._dataEnd = par.length();
2598 (void) li.setNextKey(nextkeyIdx);
2599 // Fake the last opened parenthesis
2600 li.setForDefaultLang(firstKey);
2601 nextkeyIdx = li.process(os, firstKey);
2604 if (nextKey.keytype != KeyInfo::isMain) {
2605 firstKey._dataStart = datastart;
2606 firstKey._dataEnd = nextKey._dataEnd+1;
2607 (void) li.setNextKey(nextkeyIdx);
2608 li.setForDefaultLang(firstKey);
2609 nextkeyIdx = li.process(os, firstKey);
2612 nextkeyIdx = li.process(os, nextKey);
2616 // Handle the remaining
2617 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2618 firstKey._dataEnd = par.length();
2620 if ((firstKey._dataStart < firstKey._dataEnd) &&
2621 (par[firstKey._dataStart] != '}')) {
2622 li.setForDefaultLang(firstKey);
2623 (void) li.process(os, firstKey);
2627 // return string definitelly impossible to match
2628 s = "\\foreignlanguage{ignore}{ }";
2632 s = par; /* no known macros found */
2633 // LYXERR(Debug::INFO, "After split: " << s);
2638 * Try to unify the language specs in the latexified text.
2639 * Resulting modified string is set to "", if
2640 * the searched tex does not contain all the features in the search pattern
2642 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2644 static Features regex_f;
2645 static int missed = 0;
2646 static bool regex_with_format = false;
2648 int parlen = par.length();
2650 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2653 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2654 // Happens to be there in case of description or labeling environment
2659 // Split the latex input into pieces which
2660 // can be digested by our search engine
2661 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2662 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2663 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2666 result = par.substr(0, parlen);
2667 if (isPatternString) {
2670 regex_f = identifyFeatures(result);
2671 string features = "";
2672 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2673 string a = it->first;
2674 regex_with_format = true;
2675 features += " " + a;
2676 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2678 LYXERR(Debug::FIND, "Identified Features" << features);
2681 } else if (regex_with_format) {
2682 Features info = identifyFeatures(result);
2683 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2684 string a = it->first;
2685 bool b = it->second;
2686 if (b && ! info[a]) {
2688 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2694 // LYXERR(Debug::INFO, "No regex formats");
2700 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2701 static int identifyClosing(string & t)
2703 int open_braces = 0;
2705 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2706 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2708 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]" REGEX_EOS, "$1"))
2710 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2712 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2721 static int num_replaced = 0;
2722 static bool previous_single_replace = true;
2724 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2725 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2727 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2728 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2729 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2730 if (opt.replace_all && previous_single_replace) {
2731 previous_single_replace = false;
2734 else if (!opt.replace_all) {
2735 num_replaced = 0; // count number of replaced strings
2736 previous_single_replace = true;
2738 // When using regexp, braces are hacked already by escape_for_regex()
2739 par_as_string = normalize(ds, !use_regexp);
2741 close_wildcards = 0;
2743 size_t lead_size = 0;
2744 // correct the language settings
2745 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2746 if (opt.ignoreformat) {
2748 // if par_as_string_nolead were emty,
2749 // the following call to findAux will always *find* the string
2750 // in the checked data, and thus always using the slow
2751 // examining of the current text part.
2752 par_as_string_nolead = par_as_string;
2755 lead_size = identifyLeading(par_as_string);
2756 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2757 lead_as_string = par_as_string.substr(0, lead_size);
2758 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2762 open_braces = identifyClosing(par_as_string);
2763 identifyClosing(par_as_string_nolead);
2764 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2765 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2767 string lead_as_regexp;
2768 if (lead_size > 0) {
2769 // @todo No need to search for \regexp{} insets in leading material
2770 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2771 par_as_string = par_as_string_nolead;
2772 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2773 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2775 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2776 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2777 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2779 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2780 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2781 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2782 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2783 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2784 || regex_replace(par_as_string, par_as_string,
2785 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2786 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2787 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2791 if (!opt.ignoreformat) {
2792 // Remove extra '\}' at end if not part of \{\.\}
2793 size_t lng = par_as_string.size();
2795 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2797 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2806 if (lng < par_as_string.size())
2807 par_as_string = par_as_string.substr(0,lng);
2810 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2811 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2812 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2813 // replace '[^...]' with '[^...\}\{\\]'
2814 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2815 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2816 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2818 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2821 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2822 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2823 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2824 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2826 // If entered regexp must match at begin of searched string buffer
2827 // Kornel: Added parentheses to use $1 for size of the leading string
2831 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2832 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2833 // so the convert has no effect in that case
2834 for (int i = 8; i > 0; --i) {
2835 string orig = "\\\\" + std::to_string(i);
2836 string dest = "\\" + std::to_string(i+1);
2837 while (regex_replace(par_as_string, par_as_string, orig, dest));
2839 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2840 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2842 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2843 regexp = lyx::regex(regexp_str);
2845 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2846 regexp2 = lyx::regex(regexp2_str);
2851 // Count number of characters in string
2855 // \\[a-zA-Z]+ ==> 1
2856 static int computeSize(string s, int len)
2862 for (int i = 0; i < len; i += skip, count++) {
2865 if (isalpha(s[i+1])) {
2866 for (int j = 2; i+j < len; j++) {
2867 if (! isalpha(s[i+j])) {
2870 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2872 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2880 else if (s[i] == '{') {
2886 else if (s[i] == '-') {
2887 if (s[i+1] == '-') {
2903 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2908 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2911 docstring docstr = stringifyFromForSearch(opt, cur, len);
2912 string str = normalize(docstr, true);
2913 if (!opt.ignoreformat) {
2914 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2917 mres.match_len = -1;
2920 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2921 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2924 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2925 regex const *p_regexp;
2926 regex_constants::match_flag_type flags;
2928 flags = regex_constants::match_continuous;
2931 flags = regex_constants::match_default;
2932 p_regexp = ®exp2;
2934 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2935 if (re_it == sregex_iterator())
2937 match_results<string::const_iterator> const & m = *re_it;
2939 if (0) { // Kornel Benko: DO NOT CHECKK
2940 // Check braces on the segment that matched the entire regexp expression,
2941 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2942 if (!braces_match(m[0].first, m[0].second, open_braces))
2946 // Check braces on segments that matched all (.*?) subexpressions,
2947 // except the last "padding" one inserted by lyx.
2948 for (size_t i = 1; i < m.size() - 1; ++i)
2949 if (!braces_match(m[i].first, m[i].second, open_braces))
2952 // Exclude from the returned match length any length
2953 // due to close wildcards added at end of regexp
2954 // and also the length of the leading (e.g. '\emph{}')
2956 // Whole found string, including the leading: m[0].second - m[0].first
2957 // Size of the leading string: m[1].second - m[1].first
2958 int leadingsize = 0;
2960 leadingsize = m[1].second - m[1].first;
2962 for (size_t i = 0; i < m.size(); i++) {
2963 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2965 if (close_wildcards == 0)
2966 result = m[0].second - m[0].first;
2969 result = m[m.size() - close_wildcards].first - m[0].first;
2971 size_t pos = m.position(size_t(0));
2972 // Ignore last closing characters
2973 while (result > 0) {
2974 if (str[pos+result-1] == '}')
2979 if (result > leadingsize)
2980 result -= leadingsize;
2983 mres.match_len = computeSize(str.substr(pos+leadingsize,result), result);
2984 mres.match2end = str.size() - pos - leadingsize;
2985 mres.pos = pos+leadingsize;
2989 // else !use_regexp: but all code paths above return
2990 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2991 << par_as_string << "', str='" << str << "'");
2992 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2993 << lead_as_string << "', par_as_string_nolead='"
2994 << par_as_string_nolead << "'");
2997 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2998 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2999 if (str.substr(0, par_as_string.size()) == par_as_string) {
3000 mres.match_len = par_as_string.size();
3001 mres.match2end = str.size();
3006 // Start the search _after_ the leading part
3007 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3008 if (pos != string::npos) {
3009 mres.match_len = par_as_string.size();
3010 mres.match2end = str.size() - pos;
3019 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3021 MatchResult mres = findAux(cur, len, at_begin);
3022 int res = mres.match_len;
3024 "res=" << res << ", at_begin=" << at_begin
3025 << ", matchword=" << opt.matchword
3026 << ", inTexted=" << cur.inTexted());
3027 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3029 if ((len > 0) && (res < len)) {
3033 Paragraph const & par = cur.paragraph();
3034 bool ws_left = (cur.pos() > 0)
3035 ? par.isWordSeparator(cur.pos() - 1)
3041 ws_right = (cur.pos() + len < par.size())
3042 ? par.isWordSeparator(cur.pos() + len)
3046 "cur.pos()=" << cur.pos() << ", res=" << res
3047 << ", separ: " << ws_left << ", " << ws_right
3050 if (ws_left && ws_right) {
3051 // Check for word separators inside the found 'word'
3052 for (int i = 0; i < len; i++) {
3053 if (par.isWordSeparator(cur.pos() + i)) {
3065 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
3068 if (! opt.casesensitive)
3069 t = lyx::to_utf8(lowercase(s));
3071 t = lyx::to_utf8(s);
3072 // Remove \n at begin
3073 while (!t.empty() && t[0] == '\n')
3076 while (!t.empty() && t[t.size() - 1] == '\n')
3077 t = t.substr(0, t.size() - 1);
3079 // Handle all other '\n'
3080 while ((pos = t.find("\n")) != string::npos) {
3081 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3083 if (isAlnumASCII(t[pos+1])) {
3084 t.replace(pos-2, 3, " ");
3087 t.replace(pos-2, 3, "");
3090 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3091 // '\n' adjacent to non-alpha-numerics, discard
3092 t.replace(pos, 1, "");
3095 // Replace all other \n with spaces
3096 t.replace(pos, 1, " ");
3099 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3100 // Kornel: Added textsl, textsf, textit, texttt and noun
3101 // + allow to seach for colored text too
3102 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3103 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3104 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3105 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3106 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3108 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3109 // FIXME - check what preceeds the brace
3111 if (opt.ignoreformat)
3112 while (regex_replace(t, t, "\\{", "_x_<")
3113 || regex_replace(t, t, "\\}", "_x_>"))
3114 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3116 while (regex_replace(t, t, "\\\\\\{", "_x_<")
3117 || regex_replace(t, t, "\\\\\\}", "_x_>"))
3118 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3125 docstring stringifyFromCursor(DocIterator const & cur, int len)
3127 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3128 if (cur.inTexted()) {
3129 Paragraph const & par = cur.paragraph();
3130 // TODO what about searching beyond/across paragraph breaks ?
3131 // TODO Try adding a AS_STR_INSERTS as last arg
3132 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3133 int(par.size()) : cur.pos() + len;
3134 // OutputParams runparams(&cur.buffer()->params().encoding());
3135 OutputParams runparams(encodings.fromLyXName("utf8"));
3136 runparams.nice = true;
3137 runparams.flavor = OutputParams::XETEX;
3138 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3139 // No side effect of file copying and image conversion
3140 runparams.dryrun = true;
3141 runparams.for_search = true;
3142 LYXERR(Debug::FIND, "Stringifying with cur: "
3143 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3144 return par.asString(cur.pos(), end,
3145 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3147 } else if (cur.inMathed()) {
3148 CursorSlice cs = cur.top();
3149 MathData md = cs.cell();
3150 MathData::const_iterator it_end =
3151 (( len == -1 || cs.pos() + len > int(md.size()))
3153 : md.begin() + cs.pos() + len );
3155 for (MathData::const_iterator it = md.begin() + cs.pos();
3158 docstring s = asString(md2);
3159 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3162 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3167 /** Computes the LaTeX export of buf starting from cur and ending len positions
3168 * after cur, if len is positive, or at the paragraph or innermost inset end
3171 docstring latexifyFromCursor(DocIterator const & cur, int len)
3173 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3174 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3175 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3176 Buffer const & buf = *cur.buffer();
3178 odocstringstream ods;
3180 //OutputParams runparams(&buf.params().encoding());
3181 OutputParams runparams(encodings.fromLyXName("utf8"));
3182 runparams.nice = false;
3183 runparams.flavor = OutputParams::XETEX;
3184 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3185 // No side effect of file copying and image conversion
3186 runparams.dryrun = true;
3187 runparams.for_search = true;
3189 if (cur.inTexted()) {
3190 // @TODO what about searching beyond/across paragraph breaks ?
3191 pos_type endpos = cur.paragraph().size();
3192 if (len != -1 && endpos > cur.pos() + len)
3193 endpos = cur.pos() + len;
3194 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3195 string(), cur.pos(), endpos);
3196 string s = lyx::to_utf8(ods.str());
3197 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3198 return(lyx::from_utf8(s));
3199 } else if (cur.inMathed()) {
3200 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3201 for (int s = cur.depth() - 1; s >= 0; --s) {
3202 CursorSlice const & cs = cur[s];
3203 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3205 cs.asInsetMath()->asHullInset()->header_write(ws);
3210 CursorSlice const & cs = cur.top();
3211 MathData md = cs.cell();
3212 MathData::const_iterator it_end =
3213 ((len == -1 || cs.pos() + len > int(md.size()))
3215 : md.begin() + cs.pos() + len);
3217 for (MathData::const_iterator it = md.begin() + cs.pos();
3221 ods << asString(md2);
3222 // Retrieve the math environment type, and add '$' or '$]'
3223 // or others (\end{equation}) accordingly
3224 for (int s = cur.depth() - 1; s >= 0; --s) {
3225 CursorSlice const & cs2 = cur[s];
3226 InsetMath * inset = cs2.asInsetMath();
3227 if (inset && inset->asHullInset()) {
3229 inset->asHullInset()->footer_write(ws);
3233 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3235 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3241 /** Finalize an advanced find operation, advancing the cursor to the innermost
3242 ** position that matches, plus computing the length of the matching text to
3245 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
3247 // Search the foremost position that matches (avoids find of entire math
3248 // inset when match at start of it)
3250 DocIterator old_cur(cur.buffer());
3252 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3256 } while (cur && cur.depth() > d && match(cur).match_len > 0);
3258 int max_match = match(cur).match_len; /* match valid only if not searching whole words */
3259 if (max_match <= 0) return 0;
3260 LYXERR(Debug::FIND, "Ok");
3262 // Compute the match length
3264 if (cur.pos() + len > cur.lastpos())
3266 if (match.opt.matchword) {
3267 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3268 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3270 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3272 // Length of matched text (different from len param)
3273 int old_match = match(cur, len).match_len;
3277 // Greedy behaviour while matching regexps
3278 while ((new_match = match(cur, len + 1).match_len) > old_match) {
3280 old_match = new_match;
3281 LYXERR(Debug::FIND, "verifying match with len = " << len);
3288 int maxl = cur.lastpos() - cur.pos();
3289 // Greedy behaviour while matching regexps
3290 while (maxl > minl) {
3291 int actual_match = match(cur, len).match_len;
3292 if (actual_match >= max_match) {
3293 // actual_match > max_match _can_ happen,
3294 // if the search area splits
3295 // some following word so that the regex
3296 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3298 // This means, the len value is too big
3300 len = (int)((maxl + minl)/2);
3303 // (actual_match < max_match)
3305 len = (int)((maxl + minl)/2);
3309 // Search for real start of matched characters
3314 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3315 if (cur.depth() < old_cur.depth()) {
3317 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3320 if (cur.pos() != old_cur.pos()) {
3321 // OK, forwarded 1 pos in actual inset
3322 actual_match = match(cur, len-1).match_len;
3323 if (actual_match == max_match) {
3324 // Ha, got it! The shorter selection has the same match length
3329 // OK, the shorter selection matches less chars, revert to previous value
3335 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3336 actual_match = match(cur, len).match_len;
3337 if (actual_match == max_match)
3347 int findForwardAdv(DocIterator & cur, MatchStringAdv const & match)
3351 while (!theApp()->longOperationCancelled() && cur) {
3352 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3353 MatchResult mres = match(cur, -1, false);
3354 int match_len = mres.match_len;
3355 LYXERR(Debug::FIND, "match_len: " << match_len);
3356 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3357 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3360 if (match_len > 0) {
3361 // Try to find the begin of searched string
3362 int increment = mres.pos/2;
3363 while (mres.pos > 5 && (increment > 5)) {
3364 DocIterator old_cur = cur;
3365 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3367 if (! cur || (cur.pit() > old_cur.pit())) {
3368 // Are we outside of the paragraph?
3369 // This can happen if moving past some UTF8-encoded chars
3374 MatchResult mres2 = match(cur, -1, false);
3375 if ((mres2.match2end < mres.match2end) ||
3376 (mres2.match_len < mres.match_len)) {
3383 if (increment > mres.pos/2)
3384 increment = mres.pos/2;
3388 int match_len_zero_count = 0;
3389 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3391 int remaining_len = match(cur, -1, false).match_len;
3392 if (remaining_len <= 0) {
3393 // Apparently the searched string is not in the remaining part
3400 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3401 int match_len3 = match(cur, 1).match_len;
3404 int match_len2 = match(cur).match_len;
3405 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3406 if (match_len2 > 0) {
3407 // Sometimes in finalize we understand it wasn't a match
3408 // and we need to continue the outest loop
3409 int len = findAdvFinalize(cur, match);
3414 if (match_len2 >= 0) {
3415 if (match_len2 == 0)
3416 match_len_zero_count++;
3418 match_len_zero_count = 0;
3421 if (++match_len_zero_count > 3) {
3422 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3430 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3431 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3434 // This should exit nested insets, if any, or otherwise undefine the currsor.
3435 cur.pos() = cur.lastpos();
3436 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3444 /// Find the most backward consecutive match within same paragraph while searching backwards.
3445 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3447 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3448 DocIterator tmp_cur = cur;
3449 int len = findAdvFinalize(tmp_cur, match);
3450 Inset & inset = cur.inset();
3451 for (; cur != cur_begin; cur.backwardPos()) {
3452 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3453 DocIterator new_cur = cur;
3454 new_cur.backwardPos();
3455 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3457 int new_len = findAdvFinalize(new_cur, match);
3462 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3468 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3472 // Backup of original position
3473 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3474 if (cur == cur_begin)
3477 DocIterator cur_orig(cur);
3478 bool pit_changed = false;
3481 bool found_match = (match(cur, -1, false).match_len > 0);
3485 cur.pos() = cur.lastpos();
3487 cur.pos() = cur_orig.pos();
3488 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3489 DocIterator cur_prev_iter;
3491 found_match = (match(cur).match_len > 0);
3492 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3493 << found_match << ", cur: " << cur);
3495 return findMostBackwards(cur, match);
3497 // Stop if begin of document reached
3498 if (cur == cur_begin)
3500 cur_prev_iter = cur;
3504 if (cur == cur_begin)
3511 } while (!theApp()->longOperationCancelled());
3519 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3520 DocIterator const & cur, int len)
3522 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3524 if (!opt.ignoreformat)
3525 return latexifyFromCursor(cur, len);
3527 return stringifyFromCursor(cur, len);
3531 FindAndReplaceOptions::FindAndReplaceOptions(
3532 docstring const & _find_buf_name, bool _casesensitive,
3533 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3534 docstring const & _repl_buf_name, bool _keep_case,
3535 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3536 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3537 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3538 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3546 /** Check if 'len' letters following cursor are all non-lowercase */
3547 static bool allNonLowercase(Cursor const & cur, int len)
3549 pos_type beg_pos = cur.selectionBegin().pos();
3550 pos_type end_pos = cur.selectionBegin().pos() + len;
3551 if (len > cur.lastpos() + 1 - beg_pos) {
3552 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3553 len = cur.lastpos() + 1 - beg_pos;
3554 end_pos = beg_pos + len;
3556 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3557 if (isLowerCase(cur.paragraph().getChar(pos)))
3563 /** Check if first letter is upper case and second one is lower case */
3564 static bool firstUppercase(Cursor const & cur)
3567 pos_type pos = cur.selectionBegin().pos();
3568 if (pos >= cur.lastpos() - 1) {
3569 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3572 ch1 = cur.paragraph().getChar(pos);
3573 ch2 = cur.paragraph().getChar(pos + 1);
3574 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3575 LYXERR(Debug::FIND, "firstUppercase(): "
3576 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3577 << ch2 << "(" << char(ch2) << ")"
3578 << ", result=" << result << ", cur=" << cur);
3583 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3585 ** \fixme What to do with possible further paragraphs in replace buffer ?
3587 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3589 ParagraphList::iterator pit = buffer.paragraphs().begin();
3590 LASSERT(!pit->empty(), /**/);
3591 pos_type right = pos_type(1);
3592 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3593 right = pit->size();
3594 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3600 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3602 Cursor & cur = bv->cursor();
3603 if (opt.repl_buf_name.empty()
3604 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3605 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3608 DocIterator sel_beg = cur.selectionBegin();
3609 DocIterator sel_end = cur.selectionEnd();
3610 if (&sel_beg.inset() != &sel_end.inset()
3611 || sel_beg.pit() != sel_end.pit()
3612 || sel_beg.idx() != sel_end.idx())
3614 int sel_len = sel_end.pos() - sel_beg.pos();
3615 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3616 << ", sel_len: " << sel_len << endl);
3619 LASSERT(sel_len > 0, return 0);
3621 if (!matchAdv(sel_beg, sel_len).match_len)
3624 // Build a copy of the replace buffer, adapted to the KeepCase option
3625 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3627 repl_buffer_orig.write(oss);
3628 string lyx = oss.str();
3629 Buffer repl_buffer("", false);
3630 repl_buffer.setUnnamed(true);
3631 LASSERT(repl_buffer.readString(lyx), return 0);
3632 if (opt.keep_case && sel_len >= 2) {
3633 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3634 if (cur.inTexted()) {
3635 if (firstUppercase(cur))
3636 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3637 else if (allNonLowercase(cur, sel_len))
3638 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3641 cap::cutSelection(cur, false);
3642 if (cur.inTexted()) {
3643 repl_buffer.changeLanguage(
3644 repl_buffer.language(),
3645 cur.getFont().language());
3646 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3647 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3648 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3649 repl_buffer.params().documentClassPtr(),
3650 bv->buffer().errorList("Paste"));
3651 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3652 sel_len = repl_buffer.paragraphs().begin()->size();
3653 } else if (cur.inMathed()) {
3654 odocstringstream ods;
3656 // OutputParams runparams(&repl_buffer.params().encoding());
3657 OutputParams runparams(encodings.fromLyXName("utf8"));
3658 runparams.nice = false;
3659 runparams.flavor = OutputParams::XETEX;
3660 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3661 runparams.dryrun = true;
3662 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3663 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3664 docstring repl_latex = ods.str();
3665 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3667 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3668 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3669 repl_latex = from_utf8(s);
3670 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3671 MathData ar(cur.buffer());
3672 asArray(repl_latex, ar, Parse::NORMAL);
3674 sel_len = ar.size();
3675 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3677 if (cur.pos() >= sel_len)
3678 cur.pos() -= sel_len;
3681 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3682 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3683 bv->processUpdateFlags(Update::Force);
3688 /// Perform a FindAdv operation.
3689 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3694 // e.g., when invoking word-findadv from mini-buffer wither with
3695 // wrong options syntax or before ever opening advanced F&R pane
3696 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3700 MatchStringAdv matchAdv(bv->buffer(), opt);
3701 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3703 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3704 num_replaced += findAdvReplace(bv, opt, matchAdv);
3707 match_len = findForwardAdv(cur, matchAdv);
3709 match_len = findBackwardsAdv(cur, matchAdv);
3711 // This may only be raised by lyx::regex()
3712 bv->message(_("Invalid regular expression!"));
3716 if (match_len == 0) {
3717 if (num_replaced > 0) {
3718 switch (num_replaced)
3721 bv->message(_("One match has been replaced."));
3724 bv->message(_("Two matches have been replaced."));
3727 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
3733 bv->message(_("Match not found."));
3738 if (num_replaced > 0)
3739 bv->message(_("Match has been replaced."));
3741 bv->message(_("Match found."));
3743 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3744 bv->putSelectionAt(cur, match_len, !opt.forward);
3750 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3752 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3753 << opt.casesensitive << ' '
3754 << opt.matchword << ' '
3755 << opt.forward << ' '
3756 << opt.expandmacros << ' '
3757 << opt.ignoreformat << ' '
3758 << opt.replace_all << ' '
3759 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3760 << opt.keep_case << ' '
3761 << int(opt.scope) << ' '
3764 LYXERR(Debug::FIND, "built: " << os.str());
3770 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3772 LYXERR(Debug::FIND, "parsing");
3776 while (line != "EOSS") {
3780 if (is.eof()) // Tolerate malformed request
3784 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3785 opt.find_buf_name = from_utf8(s);
3786 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
3787 is.get(); // Waste space before replace string
3790 while (line != "EOSS") {
3794 if (is.eof()) // Tolerate malformed request
3798 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3799 opt.repl_buf_name = from_utf8(s);
3800 is >> opt.keep_case;
3803 opt.scope = FindAndReplaceOptions::SearchScope(i);
3805 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3807 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3808 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3809 << opt.scope << ' ' << opt.restr);