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 "buffer_funcs.h"
21 #include "BufferList.h"
22 #include "BufferParams.h"
23 #include "BufferView.h"
26 #include "CutAndPaste.h"
27 #include "FuncRequest.h"
29 #include "output_latex.h"
30 #include "OutputParams.h"
31 #include "Paragraph.h"
32 #include "ParIterator.h"
37 #include "frontends/Application.h"
38 #include "frontends/alert.h"
40 #include "mathed/InsetMath.h"
41 #include "mathed/InsetMathGrid.h"
42 #include "mathed/InsetMathHull.h"
43 #include "mathed/MathData.h"
44 #include "mathed/MathStream.h"
45 #include "mathed/MathSupport.h"
47 #include "support/convert.h"
48 #include "support/debug.h"
49 #include "support/docstream.h"
50 #include "support/FileName.h"
51 #include "support/gettext.h"
52 #include "support/lassert.h"
53 #include "support/lstrings.h"
55 #include "support/regex.h"
56 #include "support/textutils.h"
60 using namespace lyx::support;
65 // Helper class for deciding what should be ignored
70 : ignoreFamily_(false), ignoreSeries_(false),
71 ignoreShape_(false), ignoreUnderline_(false),
72 ignoreMarkUp_(false), ignoreStrikeOut_(false),
73 ignoreSectioning_(false), ignoreFrontMatter_(false),
74 ignoreColor_(false), ignoreLanguage_(false) {}
76 bool getFamily() { return ignoreFamily_; }
78 bool getSeries() { return ignoreSeries_; }
80 bool getShape() { return ignoreShape_; }
82 bool getUnderline() { return ignoreUnderline_; }
84 bool getMarkUp() { return ignoreMarkUp_; }
86 bool getStrikeOut() { return ignoreStrikeOut_; }
88 bool getSectioning() { return ignoreSectioning_; }
90 bool getFrontMatter() { return ignoreFrontMatter_; }
92 bool getColor() { return ignoreColor_; }
94 bool getLanguage() { return ignoreLanguage_; }
96 void setIgnoreFormat(string const & type, bool value);
106 bool ignoreUnderline_;
110 bool ignoreStrikeOut_;
112 bool ignoreSectioning_;
114 bool ignoreFrontMatter_;
118 bool ignoreLanguage_;
122 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
124 if (type == "color") {
125 ignoreColor_ = value;
127 else if (type == "language") {
128 ignoreLanguage_ = value;
130 else if (type == "sectioning") {
131 ignoreSectioning_ = value;
132 ignoreFrontMatter_ = value;
134 else if (type == "font") {
135 ignoreSeries_ = value;
136 ignoreShape_ = value;
137 ignoreFamily_ = value;
139 else if (type == "series") {
140 ignoreSeries_ = value;
142 else if (type == "shape") {
143 ignoreShape_ = value;
145 else if (type == "family") {
146 ignoreFamily_ = value;
148 else if (type == "markup") {
149 ignoreMarkUp_ = value;
151 else if (type == "underline") {
152 ignoreUnderline_ = value;
154 else if (type == "strike") {
155 ignoreStrikeOut_ = value;
159 // The global variable that can be changed from outside
160 IgnoreFormats ignoreFormats;
163 void setIgnoreFormat(string const & type, bool value)
165 ignoreFormats.setIgnoreFormat(type, value);
171 bool parse_bool(docstring & howto)
176 howto = split(howto, var, ' ');
184 MatchString(docstring const & s, bool cs, bool mw)
185 : str(s), case_sens(cs), whole_words(mw)
188 // returns true if the specified string is at the specified position
189 // del specifies whether deleted strings in ct mode will be considered
190 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
192 return par.find(str, case_sens, whole_words, pos, del);
200 // match whole words only
205 int findForward(DocIterator & cur, MatchString const & match,
206 bool find_del = true)
208 for (; cur; cur.forwardChar())
209 if (cur.inTexted()) {
210 int len = match(cur.paragraph(), cur.pos(), find_del);
218 int findBackwards(DocIterator & cur, MatchString const & match,
219 bool find_del = true)
223 if (cur.inTexted()) {
224 int len = match(cur.paragraph(), cur.pos(), find_del);
233 bool searchAllowed(docstring const & str)
236 frontend::Alert::error(_("Search error"), _("Search string is empty"));
243 bool findOne(BufferView * bv, docstring const & searchstr,
244 bool case_sens, bool whole, bool forward,
245 bool find_del = true, bool check_wrap = false)
247 if (!searchAllowed(searchstr))
250 DocIterator cur = forward
251 ? bv->cursor().selectionEnd()
252 : bv->cursor().selectionBegin();
254 MatchString const match(searchstr, case_sens, whole);
256 int match_len = forward
257 ? findForward(cur, match, find_del)
258 : findBackwards(cur, match, find_del);
261 bv->putSelectionAt(cur, match_len, !forward);
262 else if (check_wrap) {
263 DocIterator cur_orig(bv->cursor());
266 q = _("End of file reached while searching forward.\n"
267 "Continue searching from the beginning?");
269 q = _("Beginning of file reached while searching backward.\n"
270 "Continue searching from the end?");
271 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
272 q, 0, 1, _("&Yes"), _("&No"));
273 if (wrap_answer == 0) {
275 bv->cursor().clear();
276 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
278 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
279 bv->cursor().backwardPos();
281 bv->clearSelection();
282 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
285 bv->cursor().setCursor(cur_orig);
289 return match_len > 0;
293 int replaceAll(BufferView * bv,
294 docstring const & searchstr, docstring const & replacestr,
295 bool case_sens, bool whole)
297 Buffer & buf = bv->buffer();
299 if (!searchAllowed(searchstr) || buf.isReadonly())
302 DocIterator cur_orig(bv->cursor());
304 MatchString const match(searchstr, case_sens, whole);
307 int const rsize = replacestr.size();
308 int const ssize = searchstr.size();
311 cur.setCursor(doc_iterator_begin(&buf));
312 int match_len = findForward(cur, match, false);
313 while (match_len > 0) {
314 // Backup current cursor position and font.
315 pos_type const pos = cur.pos();
316 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
318 int striked = ssize -
319 cur.paragraph().eraseChars(pos, pos + match_len,
320 buf.params().track_changes);
321 cur.paragraph().insert(pos, replacestr, font,
322 Change(buf.params().track_changes
324 : Change::UNCHANGED));
325 for (int i = 0; i < rsize + striked; ++i)
328 match_len = findForward(cur, match, false);
331 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
333 cur_orig.fixIfBroken();
334 bv->setCursor(cur_orig);
340 // the idea here is that we are going to replace the string that
341 // is selected IF it is the search string.
342 // if there is a selection, but it is not the search string, then
343 // we basically ignore it. (FIXME We ought to replace only within
345 // if there is no selection, then:
346 // (i) if some search string has been provided, then we find it.
347 // (think of how the dialog works when you hit "replace" the
349 // (ii) if no search string has been provided, then we treat the
350 // word the cursor is in as the search string. (why? i have no
351 // idea.) but this only works in text?
353 // returns the number of replacements made (one, if any) and
354 // whether anything at all was done.
355 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
356 docstring const & replacestr, bool case_sens,
357 bool whole, bool forward, bool findnext)
359 Cursor & cur = bv->cursor();
360 if (!cur.selection()) {
361 // no selection, non-empty search string: find it
362 if (!searchstr.empty()) {
363 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
364 return make_pair(found, 0);
366 // empty search string
369 return make_pair(false, 0);
370 // select current word and treat it as the search string.
371 // This causes a minor bug as undo will restore this selection,
372 // which the user did not create (#8986).
373 cur.innerText()->selectWord(cur, WHOLE_WORD);
374 searchstr = cur.selectionAsString(false);
377 // if we still don't have a search string, report the error
379 if (!searchAllowed(searchstr))
380 return make_pair(false, 0);
382 bool have_selection = cur.selection();
383 docstring const selected = cur.selectionAsString(false);
386 ? searchstr == selected
387 : compare_no_case(searchstr, selected) == 0;
389 // no selection or current selection is not search word:
390 // just find the search word
391 if (!have_selection || !match) {
392 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
393 return make_pair(found, 0);
396 // we're now actually ready to replace. if the buffer is
397 // read-only, we can't, though.
398 if (bv->buffer().isReadonly())
399 return make_pair(false, 0);
401 cap::replaceSelectionWithString(cur, replacestr);
403 cur.pos() += replacestr.length();
404 LASSERT(cur.pos() <= cur.lastpos(),
405 cur.pos() = cur.lastpos());
408 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
410 return make_pair(true, 1);
416 docstring const find2string(docstring const & search,
417 bool casesensitive, bool matchword, bool forward)
421 << int(casesensitive) << ' '
422 << int(matchword) << ' '
428 docstring const replace2string(docstring const & replace,
429 docstring const & search,
430 bool casesensitive, bool matchword,
431 bool all, bool forward, bool findnext)
434 ss << replace << '\n'
436 << int(casesensitive) << ' '
437 << int(matchword) << ' '
439 << int(forward) << ' '
445 bool lyxfind(BufferView * bv, FuncRequest const & ev)
447 if (!bv || ev.action() != LFUN_WORD_FIND)
450 //lyxerr << "find called, cmd: " << ev << endl;
452 // data is of the form
454 // <casesensitive> <matchword> <forward>"
456 docstring howto = split(ev.argument(), search, '\n');
458 bool casesensitive = parse_bool(howto);
459 bool matchword = parse_bool(howto);
460 bool forward = parse_bool(howto);
462 return findOne(bv, search, casesensitive, matchword, forward, true, true);
466 bool lyxreplace(BufferView * bv,
467 FuncRequest const & ev, bool has_deleted)
469 if (!bv || ev.action() != LFUN_WORD_REPLACE)
472 // data is of the form
475 // <casesensitive> <matchword> <all> <forward> <findnext>"
478 docstring howto = split(ev.argument(), rplc, '\n');
479 howto = split(howto, search, '\n');
481 bool casesensitive = parse_bool(howto);
482 bool matchword = parse_bool(howto);
483 bool all = parse_bool(howto);
484 bool forward = parse_bool(howto);
485 bool findnext = howto.empty() ? true : parse_bool(howto);
490 int replace_count = 0;
492 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
493 update = replace_count > 0;
496 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
498 replace_count = rv.second;
501 Buffer const & buf = bv->buffer();
503 // emit message signal.
504 buf.message(_("String not found."));
506 if (replace_count == 0) {
507 buf.message(_("String found."));
508 } else if (replace_count == 1) {
509 buf.message(_("String has been replaced."));
511 docstring const str =
512 bformat(_("%1$d strings have been replaced."), replace_count);
516 } else if (findnext) {
517 // if we have deleted characters, we do not replace at all, but
518 // rather search for the next occurence
519 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
522 bv->message(_("String not found."));
528 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
530 for (; cur; cur.forwardPos())
531 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
535 DocIterator cur_orig(bv->cursor());
536 docstring q = _("End of file reached while searching forward.\n"
537 "Continue searching from the beginning?");
538 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
539 q, 0, 1, _("&Yes"), _("&No"));
540 if (wrap_answer == 0) {
541 bv->cursor().clear();
542 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
543 bv->clearSelection();
544 cur.setCursor(bv->cursor().selectionBegin());
545 if (findNextChange(bv, cur, false))
548 bv->cursor().setCursor(cur_orig);
555 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
557 for (cur.backwardPos(); cur; cur.backwardPos()) {
558 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
563 DocIterator cur_orig(bv->cursor());
564 docstring q = _("Beginning of file reached while searching backward.\n"
565 "Continue searching from the end?");
566 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
567 q, 0, 1, _("&Yes"), _("&No"));
568 if (wrap_answer == 0) {
569 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
570 bv->cursor().backwardPos();
571 bv->clearSelection();
572 cur.setCursor(bv->cursor().selectionBegin());
573 if (findPreviousChange(bv, cur, false))
576 bv->cursor().setCursor(cur_orig);
583 bool selectChange(Cursor & cur, bool forward)
585 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
587 Change ch = cur.paragraph().lookupChange(cur.pos());
589 CursorSlice tip1 = cur.top();
590 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
591 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
592 if (!ch2.isSimilarTo(ch))
595 CursorSlice tip2 = cur.top();
596 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
598 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
599 if (!ch2.isSimilarTo(ch)) {
600 // take a step forward to correctly set the selection
608 cur.bv().mouseSetCursor(cur, false);
610 cur.bv().mouseSetCursor(cur, true);
618 bool findChange(BufferView * bv, bool forward)
621 cur.setCursor(forward ? bv->cursor().selectionEnd()
622 : bv->cursor().selectionBegin());
623 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
624 return selectChange(cur, forward);
629 bool findNextChange(BufferView * bv)
631 return findChange(bv, true);
635 bool findPreviousChange(BufferView * bv)
637 return findChange(bv, false);
644 typedef vector<pair<string, string> > Escapes;
646 /// A map of symbols and their escaped equivalent needed within a regex.
647 /// @note Beware of order
648 Escapes const & get_regexp_escapes()
650 typedef std::pair<std::string, std::string> P;
652 static Escapes escape_map;
653 if (escape_map.empty()) {
654 escape_map.push_back(P("$", "_x_$"));
655 escape_map.push_back(P("{", "_x_{"));
656 escape_map.push_back(P("}", "_x_}"));
657 escape_map.push_back(P("[", "_x_["));
658 escape_map.push_back(P("]", "_x_]"));
659 escape_map.push_back(P("(", "_x_("));
660 escape_map.push_back(P(")", "_x_)"));
661 escape_map.push_back(P("+", "_x_+"));
662 escape_map.push_back(P("*", "_x_*"));
663 escape_map.push_back(P(".", "_x_."));
664 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
665 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
666 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
667 escape_map.push_back(P("_x_", "\\"));
672 /// A map of lyx escaped strings and their unescaped equivalent.
673 Escapes const & get_lyx_unescapes()
675 typedef std::pair<std::string, std::string> P;
677 static Escapes escape_map;
678 if (escape_map.empty()) {
679 escape_map.push_back(P("\\%", "%"));
680 escape_map.push_back(P("\\mathcircumflex ", "^"));
681 escape_map.push_back(P("\\mathcircumflex", "^"));
682 escape_map.push_back(P("\\backslash ", "\\"));
683 escape_map.push_back(P("\\backslash", "\\"));
684 escape_map.push_back(P("\\\\{", "_x_<"));
685 escape_map.push_back(P("\\\\}", "_x_>"));
686 escape_map.push_back(P("\\sim ", "~"));
687 escape_map.push_back(P("\\sim", "~"));
692 /// A map of escapes turning a regexp matching text to one matching latex.
693 Escapes const & get_regexp_latex_escapes()
695 typedef std::pair<std::string, std::string> P;
697 static Escapes escape_map;
698 if (escape_map.empty()) {
699 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
700 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
701 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
702 escape_map.push_back(P("\\[", "\\{\\[\\}"));
703 escape_map.push_back(P("\\]", "\\{\\]\\}"));
704 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
705 escape_map.push_back(P("%", "\\\\\\%"));
706 escape_map.push_back(P("#", "\\\\#"));
711 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
712 ** the found occurrence were escaped.
714 string apply_escapes(string s, Escapes const & escape_map)
716 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
717 Escapes::const_iterator it;
718 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
719 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
720 unsigned int pos = 0;
721 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
722 s.replace(pos, it->first.length(), it->second);
723 LYXERR(Debug::FIND, "After escape: " << s);
724 pos += it->second.length();
725 // LYXERR(Debug::FIND, "pos: " << pos);
728 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
733 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
734 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
735 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
736 string escape_for_regex(string s, bool match_latex)
739 while (pos < s.size()) {
740 size_t new_pos = s.find("\\regexp{", pos);
741 if (new_pos == string::npos)
745 LYXERR(Debug::FIND, "new_pos: " << new_pos);
746 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
747 LYXERR(Debug::FIND, "t [lyx]: " << t);
748 t = apply_escapes(t, get_regexp_escapes());
749 LYXERR(Debug::FIND, "t [rxp]: " << t);
750 s.replace(pos, new_pos - pos, t);
751 new_pos = pos + t.size();
752 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
753 LYXERR(Debug::FIND, "new_pos: " << new_pos);
754 if (new_pos == s.size())
757 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
758 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
759 LYXERR(Debug::FIND, "end_pos: " << end_pos);
760 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
761 LYXERR(Debug::FIND, "t in regexp : " << t);
762 t = apply_escapes(t, get_lyx_unescapes());
763 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
765 t = apply_escapes(t, get_regexp_latex_escapes());
766 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
768 if (end_pos == s.size()) {
769 s.replace(new_pos, end_pos - new_pos, t);
770 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
773 s.replace(new_pos, end_pos + 13 - new_pos, t);
774 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
775 pos = new_pos + t.size();
776 LYXERR(Debug::FIND, "pos: " << pos);
782 /// Wrapper for lyx::regex_replace with simpler interface
783 bool regex_replace(string const & s, string & t, string const & searchstr,
784 string const & replacestr)
786 lyx::regex e(searchstr, regex_constants::ECMAScript);
788 ostream_iterator<char, char> it(oss);
789 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
790 // tolerate t and s be references to the same variable
791 bool rv = (s != oss.str());
797 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
799 ** Verify that closed braces exactly match open braces. This avoids that, for example,
800 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
803 ** Number of open braces that must remain open at the end for the verification to succeed.
805 bool braces_match(string::const_iterator const & beg,
806 string::const_iterator const & end,
810 string::const_iterator it = beg;
811 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
812 for (; it != end; ++it) {
813 // Skip escaped braces in the count
818 } else if (*it == '{') {
820 } else if (*it == '}') {
821 if (open_pars == 0) {
822 LYXERR(Debug::FIND, "Found unmatched closed brace");
828 if (open_pars != unmatched) {
829 LYXERR(Debug::FIND, "Found " << open_pars
830 << " instead of " << unmatched
831 << " unmatched open braces at the end of count");
834 LYXERR(Debug::FIND, "Braces match as expected");
844 MatchResult(): match_len(0),match2end(0), pos(0) {};
847 /** The class performing a match between a position in the document and the FindAdvOptions.
849 class MatchStringAdv {
851 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
853 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
854 ** constructor as opt.search, under the opt.* options settings.
857 ** If set, then match is searched only against beginning of text starting at cur.
858 ** If unset, then match is searched anywhere in text starting at cur.
861 ** The length of the matching text, or zero if no match was found.
863 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
868 /// first buffer on which search was started
869 lyx::Buffer * const p_first_buf;
871 FindAndReplaceOptions const & opt;
874 /// Auxiliary find method (does not account for opt.matchword)
875 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
877 /** Normalize a stringified or latexified LyX paragraph.
881 ** <li>if search is not casesensitive, then lowercase the string;
882 ** <li>remove any newline at begin or end of the string;
883 ** <li>replace any newline in the middle of the string with a simple space;
884 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
887 ** @todo Normalization should also expand macros, if the corresponding
888 ** search option was checked.
890 string normalize(docstring const & s, bool hack_braces) const;
891 // normalized string to search
892 string par_as_string;
893 // regular expression to use for searching
895 // same as regexp, but prefixed with a ".*?"
897 // leading format material as string
898 string lead_as_string;
899 // par_as_string after removal of lead_as_string
900 string par_as_string_nolead;
901 // unmatched open braces in the search string/regexp
903 // number of (.*?) subexpressions added at end of search regexp for closing
904 // environments, math mode, styles, etc...
906 // Are we searching with regular expressions ?
911 static docstring buffer_to_latex(Buffer & buffer)
913 //OutputParams runparams(&buffer.params().encoding());
914 OutputParams runparams(encodings.fromLyXName("utf8"));
915 odocstringstream ods;
917 runparams.nice = true;
918 runparams.flavor = OutputParams::XETEX;
919 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
920 // No side effect of file copying and image conversion
921 runparams.dryrun = true;
922 runparams.for_search = true;
923 pit_type const endpit = buffer.paragraphs().size();
924 for (pit_type pit = 0; pit != endpit; ++pit) {
925 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
926 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
932 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
935 if (!opt.ignoreformat) {
936 str = buffer_to_latex(buffer);
938 // OutputParams runparams(&buffer.params().encoding());
939 OutputParams runparams(encodings.fromLyXName("utf8"));
940 runparams.nice = true;
941 runparams.flavor = OutputParams::XETEX;
942 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
943 runparams.dryrun = true;
944 runparams.for_search = true;
945 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
946 Paragraph const & par = buffer.paragraphs().at(pit);
947 LYXERR(Debug::FIND, "Adding to search string: '"
948 << par.asString(pos_type(0), par.size(),
949 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
952 str += par.asString(pos_type(0), par.size(),
953 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
961 /// Return separation pos between the leading material and the rest
962 static size_t identifyLeading(string const & s)
965 // @TODO Support \item[text]
966 // Kornel: Added textsl, textsf, textit, texttt and noun
967 // + allow to search for colored text too
968 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)\\*?)\\{", "")
969 || regex_replace(t, t, REGEX_BOS "\\$", "")
970 || regex_replace(t, t, REGEX_BOS "\\\\\\[", "")
971 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
972 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
974 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
979 * Given a latexified string, retrieve some handled features
980 * The features of the regex will later be compared with the features
981 * of the searched text. If the regex features are not a
982 * subset of the analized, then, in not format ignoring search
983 * we can early stop the search in the relevant inset.
985 typedef map<string, bool> Features;
987 static Features identifyFeatures(string const & s)
989 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
990 static regex const valid("^("
992 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
993 "emph|noun|text(bf|md|sl|sf|it|tt)|"
994 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
995 "(u|uu)line|(s|x)out|uwave|"
996 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
997 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1002 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1005 if (sub.str(1).compare("regexp") == 0) {
1009 string token = sub.str(1);
1011 if (regex_match(token, sub2, valid)) {
1019 if (sub.str(1).compare("endregexp") == 0) {
1029 * defines values features of a key "\\[a-z]+{"
1034 /* Char type with content discarded
1035 * like \hspace{1cm} */
1037 /* Char, like \backslash */
1039 /* \part, \section*, ... */
1041 /* title, author etc */
1043 /* \foreignlanguage{ngerman}, ... */
1046 * to discard language in content */
1049 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1051 /* fonts, colors, markups, ... */
1053 /* footnotesize, ... large, ...
1054 * Ignore all of them */
1057 /* inputencoding, ...
1058 * Discard also content, because they do not help in search */
1061 * like remove, but also all arguments */
1063 /* item, listitem */
1065 /* tex, latex, ... like isChar */
1067 /* like \lettrine[lines=5]{}{} */
1069 /* End of arguments marker for lettrine,
1070 * so that they can be ignored */
1080 parenthesiscount(1),
1084 KeyInfo(KeyType type, int parcount, bool disable)
1090 parenthesiscount(parcount),
1099 int parenthesiscount;
1101 bool used; /* by pattern */
1106 Border(int l=0, int u=0) : low(l), upper(u) {};
1111 #define MAXOPENED 30
1113 bool isPatternString_;
1115 explicit Intervall(bool isPattern, string const & p) :
1116 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1117 hasTitle(false), langcount(0)
1125 static vector<Border> borders;
1126 int depts[MAXOPENED];
1127 int closes[MAXOPENED];
1128 int actualdeptindex;
1129 int previousNotIgnored(int);
1130 int nextNotIgnored(int);
1131 void handleOpenP(int i);
1132 void handleCloseP(int i, bool closingAllowed);
1133 void resetOpenedP(int openPos);
1134 void addIntervall(int upper);
1135 void addIntervall(int low, int upper); /* if explicit */
1136 void removeAccents();
1137 void setForDefaultLang(KeyInfo &defLang);
1138 int findclosing(int start, int end, char up, char down, int repeat);
1139 void handleParentheses(int lastpos, bool closingAllowed);
1141 int langcount; // Number of disabled language specs up to current position in actual interval
1142 int isOpeningPar(int pos);
1144 void output(ostringstream &os, int lastpos);
1145 // string show(int lastpos);
1148 vector<Border> Intervall::borders = vector<Border>(30);
1150 int Intervall::isOpeningPar(int pos)
1152 if ((pos < 0) || (size_t(pos) >= par.size()))
1154 if (par[pos] != '{')
1156 if (size_t(pos) + 2 >= par.size())
1158 if (par[pos+2] != '}')
1160 if (par[pos+1] == '[' || par[pos+1] == ']')
1165 void Intervall::setForDefaultLang(KeyInfo &defLang)
1167 // Enable the use of first token again
1168 if (ignoreidx >= 0) {
1169 int value = defLang._tokenstart + defLang._tokensize;
1171 if (borders[0].low < value)
1172 borders[0].low = value;
1173 if (borders[0].upper < value)
1174 borders[0].upper = value;
1179 static void checkDepthIndex(int val)
1181 static int maxdepthidx = MAXOPENED-2;
1182 static int lastmaxdepth = 0;
1183 if (val > lastmaxdepth) {
1184 LYXERR(Debug::INFO, "Depth reached " << val);
1187 if (val > maxdepthidx) {
1189 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1194 // Not needed, because borders are now dynamically expanded
1195 static void checkIgnoreIdx(int val)
1197 static int lastmaxignore = -1;
1198 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1199 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1200 lastmaxignore = val;
1206 * Expand the region of ignored parts of the input latex string
1207 * The region is only relevant in output()
1209 void Intervall::addIntervall(int low, int upper)
1212 if (low == upper) return;
1213 for (idx = ignoreidx+1; idx > 0; --idx) {
1214 if (low > borders[idx-1].upper) {
1218 Border br(low, upper);
1219 if (idx > ignoreidx) {
1220 if (borders.size() <= size_t(idx)) {
1221 borders.push_back(br);
1227 // checkIgnoreIdx(ignoreidx);
1231 // Expand only if one of the new bound is inside the interwall
1232 // We know here that br.low > borders[idx-1].upper
1233 if (br.upper < borders[idx].low) {
1234 // We have to insert at this pos
1235 if (size_t(ignoreidx+1) >= borders.size()) {
1236 borders.push_back(borders[ignoreidx]);
1239 borders[ignoreidx+1] = borders[ignoreidx];
1241 for (int i = ignoreidx; i > idx; --i) {
1242 borders[i] = borders[i-1];
1246 // checkIgnoreIdx(ignoreidx);
1249 // Here we know, that we are overlapping
1250 if (br.low > borders[idx].low)
1251 br.low = borders[idx].low;
1252 // check what has to be concatenated
1254 for (int i = idx; i <= ignoreidx; i++) {
1255 if (br.upper >= borders[i].low) {
1257 if (br.upper < borders[i].upper)
1258 br.upper = borders[i].upper;
1264 // count should be >= 1 here
1267 for (int i = idx + count; i <= ignoreidx; i++) {
1268 borders[i-count+1] = borders[i];
1270 ignoreidx -= count - 1;
1276 typedef map<string, string> AccentsMap;
1277 static AccentsMap accents = map<string, string>();
1279 static void buildaccent(string n, string param, string values)
1283 const char delim = '|';
1284 while (getline(s, name, delim)) {
1286 for (size_t i = 0; i < param.size(); i++) {
1287 string key = name + "{" + param[i] + "}";
1288 // get the corresponding utf8-value
1289 if ((values[start] & 0xc0) != 0xc0) {
1290 // should not happen, utf8 encoding starts at least with 11xxxxxx
1291 // but value for '\dot{i}' is 'i', which is ascii
1292 if ((values[start] & 0x80) == 0) {
1294 accents[key] = values.substr(start, 1);
1295 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1300 for (int j = 1; ;j++) {
1301 if (start + j >= values.size()) {
1302 accents[key] = values.substr(start, j);
1303 start = values.size() - 1;
1306 else if ((values[start+j] & 0xc0) != 0x80) {
1307 // This is the first byte of following utf8 char
1308 accents[key] = values.substr(start, j);
1310 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1318 static void buildAccentsMap()
1320 accents["imath"] = "ı";
1322 accents["jmath"] = "ȷ";
1323 accents["lyxmathsym{ß}"] = "ß";
1324 accents["text{ß}"] = "ß";
1325 accents["ddot{\\imath}"] = "ï";
1326 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1327 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1328 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1329 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1330 accents["acute{\\imath}"] = "í";
1331 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1332 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1333 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1334 buildaccent("mathring|r", "aAuUwy",
1336 accents["check{\\imath}"] = "ǐ";
1337 accents["check{\\jmath}"] = "ǰ";
1338 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1339 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1340 accents["hat{\\imath}"] = "î";
1341 accents["hat{\\jmath}"] = "ĵ";
1342 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1343 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1344 accents["bar{\\imath}"] = "ī";
1345 buildaccent("bar|=", "aAeEiIoOuUyY",
1346 "āĀēĒīĪōŌūŪȳȲ"); // macron
1347 accents["tilde{\\imath}"] = "ĩ";
1348 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1349 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1350 accents["breve{\\imath}"] = "ĭ";
1351 buildaccent("breve|u", "aAeEgGiIoOuU",
1352 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1353 accents["grave{\\imath}"] = "ì";
1354 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1355 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1356 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1357 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1358 buildaccent("ogonek|k", "AaEeIiUuOo",
1359 "ĄąĘęĮįŲųǪǫ"); // ogonek
1360 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1361 "ÇçĢĢĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1362 buildaccent("subring|textsubring", "Aa",
1364 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1365 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1366 buildaccent("subtilde|textsubtilde", "EeIiUu",
1367 "ḚḛḬḭṴṵ"); // subtilde
1368 accents["dgrave{\\imath}"] = "ȉ";
1369 accents["textdoublegrave{\\i}"] = "ȉ";
1370 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1371 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1372 accents["rcap{\\imath}"] = "ȉ";
1373 accents["textroundcap{\\i}"] = "ȉ";
1374 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1375 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1376 buildaccent("slashed", "oO",
1381 * Created accents in math or regexp environment
1382 * are macros, but we need the utf8 equivalent
1384 void Intervall::removeAccents()
1386 if (accents.empty())
1388 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]))");
1390 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1392 string key = sub.str(1);
1393 if (accents.find(key) != accents.end()) {
1394 string val = accents[key];
1395 size_t pos = sub.position(size_t(0));
1396 for (size_t i = 0; i < val.size(); i++) {
1397 par[pos+i] = val[i];
1399 addIntervall(pos+val.size(), pos + sub.str(0).size());
1400 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1401 // remove traces of any remaining chars
1406 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1411 void Intervall::handleOpenP(int i)
1414 depts[actualdeptindex] = i+1;
1415 closes[actualdeptindex] = -1;
1416 checkDepthIndex(actualdeptindex);
1419 void Intervall::handleCloseP(int i, bool closingAllowed)
1421 if (actualdeptindex <= 0) {
1422 if (! closingAllowed)
1423 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1424 // if we are at the very end
1425 addIntervall(i, i+1);
1428 closes[actualdeptindex] = i+1;
1433 void Intervall::resetOpenedP(int openPos)
1435 // Used as initializer for foreignlanguage entry
1436 actualdeptindex = 1;
1437 depts[1] = openPos+1;
1441 int Intervall::previousNotIgnored(int start)
1443 int idx = 0; /* int intervalls */
1444 for (idx = ignoreidx; idx >= 0; --idx) {
1445 if (start > borders[idx].upper)
1447 if (start >= borders[idx].low)
1448 start = borders[idx].low-1;
1453 int Intervall::nextNotIgnored(int start)
1455 int idx = 0; /* int intervalls */
1456 for (idx = 0; idx <= ignoreidx; idx++) {
1457 if (start < borders[idx].low)
1459 if (start < borders[idx].upper)
1460 start = borders[idx].upper;
1465 typedef map<string, KeyInfo> KeysMap;
1466 typedef vector< KeyInfo> Entries;
1467 static KeysMap keys = map<string, KeyInfo>();
1473 Intervall interval_;
1474 void buildKeys(bool);
1475 void buildEntries(bool);
1476 void makeKey(const string &, KeyInfo, bool isPatternString);
1477 void processRegion(int start, int region_end); /* remove {} parts */
1478 void removeHead(KeyInfo&, int count=0);
1481 LatexInfo(string const & par, bool isPatternString)
1482 : entidx_(-1), interval_(isPatternString, par)
1484 buildKeys(isPatternString);
1485 entries_ = vector<KeyInfo>();
1486 buildEntries(isPatternString);
1490 if (entries_.empty()) {
1493 if (entries_[0].keytype == KeyInfo::isTitle) {
1494 if (! entries_[0].disabled) {
1495 interval_.hasTitle = true;
1496 interval_.titleValue = entries_[0].head;
1499 interval_.hasTitle = false;
1500 interval_.titleValue = "";
1502 removeHead(entries_[0]);
1503 if (entries_.size() > 1)
1512 if (int(entries_.size()) > entidx_) {
1519 bool setNextKey(int idx) {
1520 if ((idx == entidx_) && (entidx_ >= 0)) {
1527 int find(int start, KeyInfo::KeyType keytype) {
1531 while (tmpIdx < int(entries_.size())) {
1532 if (entries_[tmpIdx].keytype == keytype)
1538 int process(ostringstream &os, KeyInfo &actual);
1539 int dispatch(ostringstream &os, int previousStart, KeyInfo &actual);
1540 // string show(int lastpos) { return interval.show(lastpos);};
1541 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1542 KeyInfo &getKeyInfo(int keyinfo) {
1543 static KeyInfo invalidInfo = KeyInfo();
1544 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1547 return entries_[keyinfo];
1549 void setForDefaultLang(KeyInfo &defLang) {interval_.setForDefaultLang(defLang);};
1550 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1554 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1558 for (int i = start; i < end; i += 1 + skip) {
1562 if (c == '\\') skip = 1;
1566 else if (c == down) {
1569 if ((repeat <= 0) || (par[i+1] != up))
1587 vector<MathEntry> entries_;
1592 void insert(string const & wait, size_t start, size_t end) {
1593 MathEntry m = MathEntry();
1595 m.mathStart = start;
1597 m.mathSize = end - start;
1598 entries_.push_back(m);
1600 bool empty() { return entries_.empty(); };
1601 size_t getEndPos() {
1602 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1605 return entries_[actualIdx_].mathEnd;
1607 size_t getStartPos() {
1608 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1609 return 100000; /* definitely enough? */
1611 return entries_[actualIdx_].mathStart;
1613 size_t getFirstPos() {
1615 return getStartPos();
1618 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1621 return entries_[actualIdx_].mathSize;
1623 void incrEntry() { actualIdx_++; };
1626 void LatexInfo::buildEntries(bool isPatternString)
1628 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1629 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1630 static bool disableLanguageOverride = false;
1631 smatch sub, submath;
1632 bool evaluatingRegexp = false;
1634 bool evaluatingMath = false;
1635 bool evaluatingCode = false;
1637 bool evaluatingOptional = false;
1638 size_t optionalEnd = 0;
1641 bool math_end_waiting = false;
1642 size_t math_pos = 10000;
1645 interval_.removeAccents();
1647 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1649 if (math_end_waiting) {
1650 size_t pos = submath.position(size_t(0));
1651 if ((math_end == "$") &&
1652 (submath.str(0) == "$") &&
1653 (interval_.par[pos-1] != '\\')) {
1654 mi.insert("$", math_pos, pos + 1);
1655 math_end_waiting = false;
1657 else if ((math_end == "\\]") &&
1658 (submath.str(0) == "\\]")) {
1659 mi.insert("\\]", math_pos, pos + 2);
1660 math_end_waiting = false;
1662 else if ((submath.str(1).compare("end") == 0) &&
1663 (submath.str(2).compare(math_end) == 0)) {
1664 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1665 math_end_waiting = false;
1671 if (submath.str(1).compare("begin") == 0) {
1672 math_end_waiting = true;
1673 math_end = submath.str(2);
1674 math_pos = submath.position(size_t(0));
1676 else if (submath.str(0).compare("\\[") == 0) {
1677 math_end_waiting = true;
1679 math_pos = submath.position(size_t(0));
1681 else if (submath.str(0) == "$") {
1682 size_t pos = submath.position(size_t(0));
1683 if ((pos == 0) || (interval_.par[pos-1] != '\\')) {
1684 math_end_waiting = true;
1691 // Ignore language if there is math somewhere in pattern-string
1692 if (isPatternString) {
1695 keys["foreignlanguage"].disabled = true;
1696 disableLanguageOverride = true;
1699 disableLanguageOverride = false;
1702 if (disableLanguageOverride) {
1703 keys["foreignlanguage"].disabled = true;
1706 math_pos = mi.getFirstPos();
1707 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1709 string key = sub.str(3);
1711 if (sub.str(0)[0] == '\\')
1712 key = sub.str(0)[1];
1716 size_t k_pos = sub.position(size_t(0));
1717 if ((k_pos > 0) && (interval_.par[k_pos - 1] == '\\')) {
1718 // Escaped '$', ignoring
1724 if (evaluatingRegexp) {
1725 if (sub.str(1).compare("endregexp") == 0) {
1726 evaluatingRegexp = false;
1727 // found._tokenstart already set
1728 found._dataEnd = sub.position(size_t(0)) + 13;
1729 found._dataStart = found._dataEnd;
1730 found._tokensize = found._dataEnd - found._tokenstart;
1731 found.parenthesiscount = 0;
1732 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1739 if (evaluatingMath) {
1740 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1742 evaluatingMath = false;
1744 math_pos = mi.getStartPos();
1746 if (keys.find(key) == keys.end()) {
1747 found = KeyInfo(KeyInfo::isStandard, 0, true);
1748 if (isPatternString) {
1749 found.keytype = KeyInfo::isChar;
1750 found.disabled = false;
1757 if (key.compare("regexp") == 0) {
1758 evaluatingRegexp = true;
1759 found._tokenstart = sub.position(size_t(0));
1760 found._tokensize = 0;
1764 // Handle the other params of key
1765 if (found.keytype == KeyInfo::isIgnored)
1767 else if (found.keytype == KeyInfo::isMath) {
1768 if (size_t(sub.position(size_t(0))) == math_pos) {
1770 found._tokenstart = sub.position(size_t(0));
1771 found._tokensize = mi.getSize();
1772 found._dataEnd = found._tokenstart + found._tokensize;
1773 found._dataStart = found._dataEnd;
1774 found.parenthesiscount = 0;
1775 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1776 evaluatingMath = true;
1779 // begin|end of unknown env, discard
1780 // First handle tables
1781 // longtable|tabular
1782 bool discardComment;
1784 found.keytype = KeyInfo::doRemove;
1785 if ((sub.str(5).compare("longtable") == 0) ||
1786 (sub.str(5).compare("tabular") == 0)) {
1787 discardComment = true; /* '%' */
1790 discardComment = false;
1791 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1793 string token = sub.str(5);
1794 if (regex_match(token, sub2, removeArgs)) {
1795 found.keytype = KeyInfo::removeWithArg;
1798 // discard spaces before pos(0)
1799 int pos = sub.position(size_t(0));
1801 for (count = 0; pos - count > 0; count++) {
1802 char c = interval_.par[pos-count-1];
1803 if (discardComment) {
1804 if ((c != ' ') && (c != '%'))
1810 found._tokenstart = pos - count;
1811 if (sub.str(1).compare(0, 5, "begin") == 0) {
1812 size_t pos1 = pos + sub.str(0).length();
1813 if (sub.str(5).compare("cjk") == 0) {
1814 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1815 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1817 found.keytype = KeyInfo::isMain;
1818 found._dataStart = pos1;
1819 found._dataEnd = interval_.par.length();
1820 found.disabled = keys["foreignlanguage"].disabled;
1821 found.used = keys["foreignlanguage"].used;
1822 found._tokensize = pos1 - found._tokenstart;
1823 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1826 // Swallow possible optional params
1827 while (interval_.par[pos1] == '[') {
1828 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1830 // Swallow also the eventual parameter
1831 if (interval_.par[pos1] == '{') {
1832 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1835 found._dataEnd = pos1;
1837 found._dataStart = found._dataEnd;
1838 found._tokensize = count + found._dataEnd - pos;
1839 found.parenthesiscount = 0;
1840 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1841 found.disabled = true;
1845 // Handle "\end{...}"
1846 found._dataStart = pos + sub.str(0).length();
1847 found._dataEnd = found._dataStart;
1848 found._tokensize = count + found._dataEnd - pos;
1849 found.parenthesiscount = 0;
1850 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1851 found.disabled = true;
1855 else if (found.keytype != KeyInfo::isRegex) {
1856 found._tokenstart = sub.position(size_t(0));
1857 if (found.parenthesiscount == 0) {
1858 // Probably to be discarded
1859 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1860 char following = interval_.par[following_pos];
1861 if (following == ' ')
1862 found.head = "\\" + sub.str(3) + " ";
1863 else if (following == '=') {
1864 // like \uldepth=1000pt
1865 found.head = sub.str(0);
1868 found.head = "\\" + key;
1869 found._tokensize = found.head.length();
1870 found._dataEnd = found._tokenstart + found._tokensize;
1871 found._dataStart = found._dataEnd;
1874 int params = found._tokenstart + key.length() + 1;
1875 if (evaluatingOptional) {
1876 if (size_t(found._tokenstart) > optionalEnd) {
1877 evaluatingOptional = false;
1880 found.disabled = true;
1883 int optend = params;
1884 while (interval_.par[optend] == '[') {
1885 // discard optional parameters
1886 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
1888 if (optend > params) {
1889 key += interval_.par.substr(params, optend-params);
1890 evaluatingOptional = true;
1891 optionalEnd = optend;
1893 string token = sub.str(5);
1894 int closings = found.parenthesiscount;
1895 if (found.parenthesiscount == 1) {
1896 found.head = "\\" + key + "{";
1898 else if (found.parenthesiscount > 1) {
1900 found.head = sub.str(0) + "{";
1901 closings = found.parenthesiscount - 1;
1904 found.head = "\\" + key + "{";
1907 found._tokensize = found.head.length();
1908 found._dataStart = found._tokenstart + found.head.length();
1909 if (found.keytype == KeyInfo::doRemove) {
1910 int endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1911 found._dataStart = endpar;
1912 found._tokensize = found._dataStart - found._tokenstart;
1915 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1916 found._dataStart += 15;
1920 endpos = found._dataStart - 1;
1922 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1923 if (found.keytype == KeyInfo::isList) {
1924 // Check if it really is list env
1925 static regex const listre("^([a-z]+)$");
1927 if (!regex_match(token, sub2, listre)) {
1928 // Change the key of this entry. It is not in a list/item environment
1929 found.keytype = KeyInfo::endArguments;
1932 if (found.keytype == KeyInfo::noMain) {
1933 evaluatingCode = true;
1935 codeStart = found._dataStart;
1937 else if (evaluatingCode) {
1938 if (size_t(found._dataStart) > codeEnd)
1939 evaluatingCode = false;
1940 else if (found.keytype == KeyInfo::isMain) {
1941 // Disable this key, treate it as standard
1942 found.keytype = KeyInfo::isStandard;
1943 found.disabled = true;
1944 if ((codeEnd == interval_.par.length()) &&
1945 (found._tokenstart == codeStart)) {
1946 // trickery, because the code inset starts
1947 // with \selectlanguage ...
1949 if (entries_.size() > 1) {
1950 entries_[entries_.size()-1]._dataEnd = codeEnd;
1955 if ((endpos == interval_.par.length()) &&
1956 (found.keytype == KeyInfo::doRemove)) {
1957 // Missing closing => error in latex-input?
1958 // therefore do not delete remaining data
1959 found._dataStart -= 1;
1960 found._dataEnd = found._dataStart;
1963 found._dataEnd = endpos;
1965 if (isPatternString) {
1966 keys[key].used = true;
1969 entries_.push_back(found);
1973 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1975 stringstream s(keysstring);
1977 const char delim = '|';
1978 while (getline(s, key, delim)) {
1979 KeyInfo keyII(keyI);
1980 if (isPatternString) {
1983 else if ( !keys[key].used)
1984 keyII.disabled = true;
1989 void LatexInfo::buildKeys(bool isPatternString)
1992 static bool keysBuilt = false;
1993 if (keysBuilt && !isPatternString) return;
1995 // Known standard keys with 1 parameter.
1996 // Split is done, if not at start of region
1997 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1998 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1999 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2000 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2001 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2002 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2004 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2005 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2006 makeKey("section*|subsection*|subsubsection*|paragraph*",
2007 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2008 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2009 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2011 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2013 // Split is done, if not at start of region
2014 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2015 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2017 // Split is done always.
2018 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2022 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2023 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2024 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2025 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2027 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2028 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2029 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2031 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2032 // Custom space/skip, remove the content (== length value)
2033 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2034 // Found in fr/UserGuide.lyx
2035 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2037 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2038 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2039 // Known macros to remove (including their parameter)
2041 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2042 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2043 // handle like standard keys with 1 parameter.
2044 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2046 // Ignore deleted text
2047 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2048 // but preserve added text
2049 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2051 // Macros to remove, but let the parameter survive
2053 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2055 // Remove language spec from content of these insets
2056 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2058 // Same effect as previous, parameter will survive (because there is no one anyway)
2060 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2061 // Remove table decorations
2062 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2063 // Discard shape-header.
2064 // For footnote or shortcut too, because of lang settings
2065 // and wrong handling if used 'KeyInfo::noMain'
2066 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2067 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2068 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2069 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2070 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2071 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2072 // like ('tiny{}' or '\tiny ' ... )
2073 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2075 // Survives, like known character
2076 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2077 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2079 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2080 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2081 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2083 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2084 // Remove RTL/LTR marker
2085 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2086 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2087 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2088 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2089 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2090 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2091 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2092 if (isPatternString) {
2093 // Allow the first searched string to rebuild the keys too
2097 // no need to rebuild again
2103 * Keep the list of actual opened parentheses actual
2104 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2106 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2109 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2113 if (c == '\\') skip = 1;
2114 else if (c == '{') {
2117 else if (c == '}') {
2118 handleCloseP(i, closingAllowed);
2124 string Intervall::show(int lastpos)
2126 int idx = 0; /* int intervalls */
2129 for (idx = 0; idx <= ignoreidx; idx++) {
2130 while (i < lastpos) {
2132 if (i <= borders[idx].low) {
2133 if (borders[idx].low > lastpos)
2134 printsize = lastpos - i;
2136 printsize = borders[idx].low - i;
2137 s += par.substr(i, printsize);
2139 if (i >= borders[idx].low)
2140 i = borders[idx].upper;
2143 i = borders[idx].upper;
2149 s += par.substr(i, lastpos-i);
2155 void Intervall::output(ostringstream &os, int lastpos)
2157 // get number of chars to output
2158 int idx = 0; /* int intervalls */
2161 string startTitle = titleValue;
2162 for (idx = 0; idx <= ignoreidx; idx++) {
2164 if (i <= borders[idx].low) {
2166 if (borders[idx].low > lastpos)
2167 printsize = lastpos - i;
2169 printsize = borders[idx].low - i;
2170 if (printsize > 0) {
2171 os << startTitle << par.substr(i, printsize);
2173 printed += printsize;
2176 handleParentheses(i, false);
2177 if (i >= borders[idx].low)
2178 i = borders[idx].upper;
2181 i = borders[idx].upper;
2188 os << startTitle << par.substr(i, lastpos-i);
2189 printed += lastpos-i;
2191 handleParentheses(lastpos, false);
2193 if (keys["foreignlanguage"].disabled)
2194 startindex = actualdeptindex-langcount;
2196 startindex = actualdeptindex;
2197 for (int i = startindex; i > 0; --i) {
2200 if (hasTitle && (printed > 0))
2202 if (! isPatternString_)
2204 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2207 void LatexInfo::processRegion(int start, int region_end)
2209 while (start < region_end) { /* Let {[} and {]} survive */
2210 int cnt = interval_.isOpeningPar(start);
2212 // Closing is allowed past the region
2213 int closing = interval_.findclosing(start+1, interval_.par.length());
2214 interval_.addIntervall(start, start+1);
2215 interval_.addIntervall(closing, closing+1);
2219 start = interval_.nextNotIgnored(start+1);
2223 void LatexInfo::removeHead(KeyInfo &actual, int count)
2225 if (actual.parenthesiscount == 0) {
2226 // "{\tiny{} ...}" ==> "{{} ...}"
2227 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2230 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2231 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2232 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2236 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2239 switch (actual.keytype)
2241 case KeyInfo::isTitle: {
2243 nextKeyIdx = getNextKey();
2246 case KeyInfo::cleanToStart: {
2247 actual._dataEnd = actual._dataStart;
2248 nextKeyIdx = getNextKey();
2249 // Search for end of arguments
2250 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2252 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2253 entries_[i].disabled = true;
2255 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2257 while (interval_.par[actual._dataEnd] == ' ')
2259 interval_.addIntervall(0, actual._dataEnd+1);
2260 interval_.actualdeptindex = 0;
2261 interval_.depts[0] = actual._dataEnd+1;
2262 interval_.closes[0] = -1;
2265 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2266 if (actual.disabled)
2267 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2269 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2272 case KeyInfo::isChar: {
2273 nextKeyIdx = getNextKey();
2276 case KeyInfo::isSize: {
2277 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2278 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2279 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2280 nextKeyIdx = getNextKey();
2282 // Here _dataStart points to '{', so correct it
2283 actual._dataStart += 1;
2284 actual._tokensize += 1;
2285 actual.parenthesiscount = 1;
2286 if (interval_.par[actual._dataStart] == '}') {
2287 // Determine the end if used like '{\tiny{}...}'
2288 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2289 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2292 // Determine the end if used like '\tiny{...}'
2293 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2295 // Split on this key if not at start
2296 int start = interval_.nextNotIgnored(previousStart);
2297 if (start < actual._tokenstart) {
2298 interval_.output(os, actual._tokenstart);
2299 interval_.addIntervall(start, actual._tokenstart);
2301 // discard entry if at end of actual
2302 nextKeyIdx = process(os, actual);
2306 case KeyInfo::endArguments:
2307 // Remove trailing '{}' too
2308 actual._dataStart += 1;
2309 actual._dataEnd += 1;
2310 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2311 nextKeyIdx = getNextKey();
2313 case KeyInfo::noMain:
2315 case KeyInfo::isStandard: {
2316 if (actual.disabled) {
2318 processRegion(actual._dataStart, actual._dataStart+1);
2319 nextKeyIdx = getNextKey();
2321 // Split on this key if not at datastart of calling entry
2322 int start = interval_.nextNotIgnored(previousStart);
2323 if (start < actual._tokenstart) {
2324 interval_.output(os, actual._tokenstart);
2325 interval_.addIntervall(start, actual._tokenstart);
2327 // discard entry if at end of actual
2328 nextKeyIdx = process(os, actual);
2332 case KeyInfo::removeWithArg: {
2333 nextKeyIdx = getNextKey();
2334 // Search for end of arguments
2335 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2337 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2338 entries_[i].disabled = true;
2340 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2342 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2345 case KeyInfo::doRemove: {
2346 // Remove the key with all parameters and following spaces
2348 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2349 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2352 // Remove also enclosing parentheses [] and {}
2355 while (actual._tokenstart > numpars) {
2356 if (pos+numpars >= interval_.par.size())
2358 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2360 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2366 if (interval_.par[pos+numpars] == ' ')
2370 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2371 nextKeyIdx = getNextKey();
2374 case KeyInfo::isList: {
2375 // Discard space before _tokenstart
2377 for (count = 0; count < actual._tokenstart; count++) {
2378 if (interval_.par[actual._tokenstart-count-1] != ' ')
2381 nextKeyIdx = getNextKey();
2382 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2384 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2386 // How else can we catch this one?
2387 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2388 entries_[i].disabled = true;
2390 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2392 else if (nextKeyIdx > 0) {
2393 // Ignore any lang entries inside data region
2394 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2395 if (entries_[i].keytype == KeyInfo::isMain)
2396 entries_[i].disabled = true;
2399 if (actual.disabled) {
2400 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2403 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2405 if (interval_.par[actual._dataEnd+1] == '[') {
2406 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2407 if ((interval_.par[actual._dataEnd+2] == '{') &&
2408 (interval_.par[posdown-1] == '}')) {
2409 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2410 interval_.addIntervall(posdown-1, posdown+1);
2413 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2414 interval_.addIntervall(posdown, posdown+1);
2416 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2417 if (blk > posdown) {
2418 // Discard at most 1 space after empty item
2420 for (count = 0; count < 1; count++) {
2421 if (interval_.par[blk+count] != ' ')
2425 interval_.addIntervall(blk, blk+count);
2430 case KeyInfo::isSectioning: {
2431 // Discard spaces before _tokenstart
2433 int val = actual._tokenstart;
2434 for (count = 0; count < actual._tokenstart;) {
2435 val = interval_.previousNotIgnored(val-1);
2436 if (val < 0 || interval_.par[val] != ' ')
2439 count = actual._tokenstart - val;
2442 if (actual.disabled) {
2443 removeHead(actual, count);
2444 nextKeyIdx = getNextKey();
2446 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2447 nextKeyIdx = process(os, actual);
2451 case KeyInfo::isMath: {
2452 // Same as regex, use the content unchanged
2453 nextKeyIdx = getNextKey();
2456 case KeyInfo::isRegex: {
2457 // DO NOT SPLIT ON REGEX
2459 nextKeyIdx = getNextKey();
2462 case KeyInfo::isIgnored: {
2463 // Treat like a character for now
2464 nextKeyIdx = getNextKey();
2467 case KeyInfo::isMain: {
2468 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2469 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2470 if (actual._tokenstart > 0) {
2471 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2472 if ((prev >= 0) && interval_.par[prev] == '%')
2473 interval_.addIntervall(prev, prev+1);
2475 if (actual.disabled) {
2477 interval_.langcount++;
2478 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2479 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2480 // Discard also the space before math-equation
2481 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2483 nextKeyIdx = getNextKey();
2484 // interval.resetOpenedP(actual._dataStart-1);
2487 if (actual._tokenstart < 26) {
2488 // for the first (and maybe dummy) language
2489 interval_.setForDefaultLang(actual);
2491 interval_.resetOpenedP(actual._dataStart-1);
2495 case KeyInfo::invalid:
2496 // This cannot happen, already handled
2499 // LYXERR(Debug::INFO, "Unhandled keytype");
2500 nextKeyIdx = getNextKey();
2507 int LatexInfo::process(ostringstream &os, KeyInfo &actual )
2509 int end = interval_.nextNotIgnored(actual._dataEnd);
2510 int oldStart = actual._dataStart;
2511 int nextKeyIdx = getNextKey();
2513 if ((nextKeyIdx < 0) ||
2514 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2515 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2516 if (oldStart <= end) {
2517 processRegion(oldStart, end);
2522 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2524 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2525 (void) dispatch(os, actual._dataStart, nextKey);
2526 end = nextKey._tokenstart;
2529 processRegion(oldStart, nextKey._tokenstart);
2530 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2532 oldStart = nextKey._dataEnd+1;
2534 // now nextKey is either invalid or is outside of actual._dataEnd
2535 // output the remaining and discard myself
2536 if (oldStart <= end) {
2537 processRegion(oldStart, end);
2539 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2541 // This is the normal case.
2542 // But if using the firstlanguage, the closing may be missing
2544 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2546 if (actual._dataEnd < end)
2547 output_end = interval_.nextNotIgnored(actual._dataEnd);
2549 output_end = interval_.nextNotIgnored(end);
2550 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2551 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2553 // Remove possible empty data
2554 int dstart = interval_.nextNotIgnored(actual._dataStart);
2555 while (interval_.isOpeningPar(dstart) == 1) {
2556 interval_.addIntervall(dstart, dstart+1);
2557 int dend = interval_.findclosing(dstart+1, output_end);
2558 interval_.addIntervall(dend, dend+1);
2559 dstart = interval_.nextNotIgnored(dstart+1);
2561 if (dstart < output_end)
2562 interval_.output(os, output_end);
2563 interval_.addIntervall(actual._tokenstart, end);
2567 string splitOnKnownMacros(string par, bool isPatternString)
2570 LatexInfo li(par, isPatternString);
2571 // LYXERR(Debug::INFO, "Berfore split: " << par);
2572 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2574 DummyKey._tokensize = 0;
2575 DummyKey._dataStart = 0;
2576 DummyKey._dataEnd = par.length();
2577 DummyKey.disabled = true;
2578 int firstkeyIdx = li.getFirstKey();
2580 if (firstkeyIdx >= 0) {
2581 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2582 DummyKey._tokenstart = firstKey._tokenstart;
2584 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2585 // Use dummy firstKey
2586 firstKey = DummyKey;
2587 (void) li.setNextKey(firstkeyIdx);
2590 if (par.substr(firstKey._dataStart, 2) == "% ")
2591 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2593 nextkeyIdx = li.process(os, firstKey);
2594 while (nextkeyIdx >= 0) {
2595 // Check for a possible gap between the last
2596 // entry and this one
2597 int datastart = li.nextNotIgnored(firstKey._dataStart);
2598 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2599 if ((nextKey._tokenstart > datastart)) {
2601 firstKey._dataStart = datastart;
2602 firstKey._dataEnd = par.length();
2603 (void) li.setNextKey(nextkeyIdx);
2604 // Fake the last opened parenthesis
2605 li.setForDefaultLang(firstKey);
2606 nextkeyIdx = li.process(os, firstKey);
2609 if (nextKey.keytype != KeyInfo::isMain) {
2610 firstKey._dataStart = datastart;
2611 firstKey._dataEnd = nextKey._dataEnd+1;
2612 (void) li.setNextKey(nextkeyIdx);
2613 li.setForDefaultLang(firstKey);
2614 nextkeyIdx = li.process(os, firstKey);
2617 nextkeyIdx = li.process(os, nextKey);
2621 // Handle the remaining
2622 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2623 firstKey._dataEnd = par.length();
2625 if ((firstKey._dataStart < firstKey._dataEnd) &&
2626 (par[firstKey._dataStart] != '}')) {
2627 li.setForDefaultLang(firstKey);
2628 (void) li.process(os, firstKey);
2632 // return string definitelly impossible to match
2633 s = "\\foreignlanguage{ignore}{ }";
2637 s = par; /* no known macros found */
2638 // LYXERR(Debug::INFO, "After split: " << s);
2643 * Try to unify the language specs in the latexified text.
2644 * Resulting modified string is set to "", if
2645 * the searched tex does not contain all the features in the search pattern
2647 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2649 static Features regex_f;
2650 static int missed = 0;
2651 static bool regex_with_format = false;
2653 int parlen = par.length();
2655 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2658 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2659 // Happens to be there in case of description or labeling environment
2664 // Split the latex input into pieces which
2665 // can be digested by our search engine
2666 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2667 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2668 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2671 result = par.substr(0, parlen);
2672 if (isPatternString) {
2675 regex_f = identifyFeatures(result);
2676 string features = "";
2677 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2678 string a = it->first;
2679 regex_with_format = true;
2680 features += " " + a;
2681 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2683 LYXERR(Debug::FIND, "Identified Features" << features);
2686 } else if (regex_with_format) {
2687 Features info = identifyFeatures(result);
2688 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2689 string a = it->first;
2690 bool b = it->second;
2691 if (b && ! info[a]) {
2693 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2699 // LYXERR(Debug::INFO, "No regex formats");
2705 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2706 static int identifyClosing(string & t)
2708 int open_braces = 0;
2710 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2711 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2713 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]" REGEX_EOS, "$1"))
2715 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2717 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2726 static int num_replaced = 0;
2727 static bool previous_single_replace = true;
2729 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2730 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2732 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2733 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2734 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2735 if (opt.replace_all && previous_single_replace) {
2736 previous_single_replace = false;
2739 else if (!opt.replace_all) {
2740 num_replaced = 0; // count number of replaced strings
2741 previous_single_replace = true;
2743 // When using regexp, braces are hacked already by escape_for_regex()
2744 par_as_string = normalize(ds, !use_regexp);
2746 close_wildcards = 0;
2748 size_t lead_size = 0;
2749 // correct the language settings
2750 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2751 if (opt.ignoreformat) {
2753 // if par_as_string_nolead were emty,
2754 // the following call to findAux will always *find* the string
2755 // in the checked data, and thus always using the slow
2756 // examining of the current text part.
2757 par_as_string_nolead = par_as_string;
2760 lead_size = identifyLeading(par_as_string);
2761 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2762 lead_as_string = par_as_string.substr(0, lead_size);
2763 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2767 open_braces = identifyClosing(par_as_string);
2768 identifyClosing(par_as_string_nolead);
2769 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2770 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2772 string lead_as_regexp;
2773 if (lead_size > 0) {
2774 // @todo No need to search for \regexp{} insets in leading material
2775 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2776 par_as_string = par_as_string_nolead;
2777 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2778 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2780 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2781 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2782 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2784 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2785 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2786 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2787 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2788 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2789 || regex_replace(par_as_string, par_as_string,
2790 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2791 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2792 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2796 if (!opt.ignoreformat) {
2797 // Remove extra '\}' at end if not part of \{\.\}
2798 size_t lng = par_as_string.size();
2800 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2802 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2811 if (lng < par_as_string.size())
2812 par_as_string = par_as_string.substr(0,lng);
2815 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2816 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2817 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2818 // replace '[^...]' with '[^...\}\{\\]'
2819 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2820 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2821 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2823 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2826 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2827 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2828 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2829 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2831 // If entered regexp must match at begin of searched string buffer
2832 // Kornel: Added parentheses to use $1 for size of the leading string
2836 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2837 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2838 // so the convert has no effect in that case
2839 for (int i = 8; i > 0; --i) {
2840 string orig = "\\\\" + std::to_string(i);
2841 string dest = "\\" + std::to_string(i+1);
2842 while (regex_replace(par_as_string, par_as_string, orig, dest));
2844 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2845 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2847 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2848 regexp = lyx::regex(regexp_str);
2850 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2851 regexp2 = lyx::regex(regexp2_str);
2856 // Count number of characters in string
2860 // \\[a-zA-Z]+ ==> 1
2861 static int computeSize(string s, int len)
2867 for (int i = 0; i < len; i += skip, count++) {
2870 if (isalpha(s[i+1])) {
2871 for (int j = 2; i+j < len; j++) {
2872 if (! isalpha(s[i+j])) {
2875 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2877 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2885 else if (s[i] == '{') {
2891 else if (s[i] == '-') {
2892 if (s[i+1] == '-') {
2908 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2913 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2916 docstring docstr = stringifyFromForSearch(opt, cur, len);
2917 string str = normalize(docstr, true);
2918 if (!opt.ignoreformat) {
2919 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2922 mres.match_len = -1;
2925 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2926 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2929 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2930 regex const *p_regexp;
2931 regex_constants::match_flag_type flags;
2933 flags = regex_constants::match_continuous;
2936 flags = regex_constants::match_default;
2937 p_regexp = ®exp2;
2939 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2940 if (re_it == sregex_iterator())
2942 match_results<string::const_iterator> const & m = *re_it;
2944 if (0) { // Kornel Benko: DO NOT CHECKK
2945 // Check braces on the segment that matched the entire regexp expression,
2946 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2947 if (!braces_match(m[0].first, m[0].second, open_braces))
2951 // Check braces on segments that matched all (.*?) subexpressions,
2952 // except the last "padding" one inserted by lyx.
2953 for (size_t i = 1; i < m.size() - 1; ++i)
2954 if (!braces_match(m[i].first, m[i].second, open_braces))
2957 // Exclude from the returned match length any length
2958 // due to close wildcards added at end of regexp
2959 // and also the length of the leading (e.g. '\emph{}')
2961 // Whole found string, including the leading: m[0].second - m[0].first
2962 // Size of the leading string: m[1].second - m[1].first
2963 int leadingsize = 0;
2965 leadingsize = m[1].second - m[1].first;
2967 for (size_t i = 0; i < m.size(); i++) {
2968 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2970 if (close_wildcards == 0)
2971 result = m[0].second - m[0].first;
2974 result = m[m.size() - close_wildcards].first - m[0].first;
2976 size_t pos = m.position(size_t(0));
2977 // Ignore last closing characters
2978 while (result > 0) {
2979 if (str[pos+result-1] == '}')
2984 if (result > leadingsize)
2985 result -= leadingsize;
2988 mres.match_len = computeSize(str.substr(pos+leadingsize,result), result);
2989 mres.match2end = str.size() - pos - leadingsize;
2990 mres.pos = pos+leadingsize;
2994 // else !use_regexp: but all code paths above return
2995 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2996 << par_as_string << "', str='" << str << "'");
2997 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2998 << lead_as_string << "', par_as_string_nolead='"
2999 << par_as_string_nolead << "'");
3002 LYXERR(Debug::FIND, "size=" << par_as_string.size()
3003 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
3004 if (str.substr(0, par_as_string.size()) == par_as_string) {
3005 mres.match_len = par_as_string.size();
3006 mres.match2end = str.size();
3011 // Start the search _after_ the leading part
3012 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3013 if (pos != string::npos) {
3014 mres.match_len = par_as_string.size();
3015 mres.match2end = str.size() - pos;
3024 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3026 MatchResult mres = findAux(cur, len, at_begin);
3027 int res = mres.match_len;
3029 "res=" << res << ", at_begin=" << at_begin
3030 << ", matchword=" << opt.matchword
3031 << ", inTexted=" << cur.inTexted());
3032 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3034 if ((len > 0) && (res < len)) {
3038 Paragraph const & par = cur.paragraph();
3039 bool ws_left = (cur.pos() > 0)
3040 ? par.isWordSeparator(cur.pos() - 1)
3046 ws_right = (cur.pos() + len < par.size())
3047 ? par.isWordSeparator(cur.pos() + len)
3051 "cur.pos()=" << cur.pos() << ", res=" << res
3052 << ", separ: " << ws_left << ", " << ws_right
3055 if (ws_left && ws_right) {
3056 // Check for word separators inside the found 'word'
3057 for (int i = 0; i < len; i++) {
3058 if (par.isWordSeparator(cur.pos() + i)) {
3070 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
3073 if (! opt.casesensitive)
3074 t = lyx::to_utf8(lowercase(s));
3076 t = lyx::to_utf8(s);
3077 // Remove \n at begin
3078 while (!t.empty() && t[0] == '\n')
3081 while (!t.empty() && t[t.size() - 1] == '\n')
3082 t = t.substr(0, t.size() - 1);
3084 // Handle all other '\n'
3085 while ((pos = t.find("\n")) != string::npos) {
3086 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3088 if (isAlnumASCII(t[pos+1])) {
3089 t.replace(pos-2, 3, " ");
3092 t.replace(pos-2, 3, "");
3095 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3096 // '\n' adjacent to non-alpha-numerics, discard
3097 t.replace(pos, 1, "");
3100 // Replace all other \n with spaces
3101 t.replace(pos, 1, " ");
3104 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3105 // Kornel: Added textsl, textsf, textit, texttt and noun
3106 // + allow to seach for colored text too
3107 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3108 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3109 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3110 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3111 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3113 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3114 // FIXME - check what preceeds the brace
3116 if (opt.ignoreformat)
3117 while (regex_replace(t, t, "\\{", "_x_<")
3118 || regex_replace(t, t, "\\}", "_x_>"))
3119 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3121 while (regex_replace(t, t, "\\\\\\{", "_x_<")
3122 || regex_replace(t, t, "\\\\\\}", "_x_>"))
3123 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3130 docstring stringifyFromCursor(DocIterator const & cur, int len)
3132 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3133 if (cur.inTexted()) {
3134 Paragraph const & par = cur.paragraph();
3135 // TODO what about searching beyond/across paragraph breaks ?
3136 // TODO Try adding a AS_STR_INSERTS as last arg
3137 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3138 int(par.size()) : cur.pos() + len;
3139 // OutputParams runparams(&cur.buffer()->params().encoding());
3140 OutputParams runparams(encodings.fromLyXName("utf8"));
3141 runparams.nice = true;
3142 runparams.flavor = OutputParams::XETEX;
3143 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3144 // No side effect of file copying and image conversion
3145 runparams.dryrun = true;
3146 runparams.for_search = true;
3147 LYXERR(Debug::FIND, "Stringifying with cur: "
3148 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3149 return par.asString(cur.pos(), end,
3150 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3152 } else if (cur.inMathed()) {
3153 CursorSlice cs = cur.top();
3154 MathData md = cs.cell();
3155 MathData::const_iterator it_end =
3156 (( len == -1 || cs.pos() + len > int(md.size()))
3158 : md.begin() + cs.pos() + len );
3160 for (MathData::const_iterator it = md.begin() + cs.pos();
3163 docstring s = asString(md2);
3164 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3167 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3172 /** Computes the LaTeX export of buf starting from cur and ending len positions
3173 * after cur, if len is positive, or at the paragraph or innermost inset end
3176 docstring latexifyFromCursor(DocIterator const & cur, int len)
3178 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3179 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3180 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3181 Buffer const & buf = *cur.buffer();
3183 odocstringstream ods;
3185 //OutputParams runparams(&buf.params().encoding());
3186 OutputParams runparams(encodings.fromLyXName("utf8"));
3187 runparams.nice = false;
3188 runparams.flavor = OutputParams::XETEX;
3189 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3190 // No side effect of file copying and image conversion
3191 runparams.dryrun = true;
3192 runparams.for_search = true;
3194 if (cur.inTexted()) {
3195 // @TODO what about searching beyond/across paragraph breaks ?
3196 pos_type endpos = cur.paragraph().size();
3197 if (len != -1 && endpos > cur.pos() + len)
3198 endpos = cur.pos() + len;
3199 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3200 string(), cur.pos(), endpos);
3201 string s = lyx::to_utf8(ods.str());
3202 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3203 return(lyx::from_utf8(s));
3204 } else if (cur.inMathed()) {
3205 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3206 for (int s = cur.depth() - 1; s >= 0; --s) {
3207 CursorSlice const & cs = cur[s];
3208 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3210 cs.asInsetMath()->asHullInset()->header_write(ws);
3215 CursorSlice const & cs = cur.top();
3216 MathData md = cs.cell();
3217 MathData::const_iterator it_end =
3218 ((len == -1 || cs.pos() + len > int(md.size()))
3220 : md.begin() + cs.pos() + len);
3222 for (MathData::const_iterator it = md.begin() + cs.pos();
3226 ods << asString(md2);
3227 // Retrieve the math environment type, and add '$' or '$]'
3228 // or others (\end{equation}) accordingly
3229 for (int s = cur.depth() - 1; s >= 0; --s) {
3230 CursorSlice const & cs2 = cur[s];
3231 InsetMath * inset = cs2.asInsetMath();
3232 if (inset && inset->asHullInset()) {
3234 inset->asHullInset()->footer_write(ws);
3238 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3240 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3246 /** Finalize an advanced find operation, advancing the cursor to the innermost
3247 ** position that matches, plus computing the length of the matching text to
3250 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
3252 // Search the foremost position that matches (avoids find of entire math
3253 // inset when match at start of it)
3255 DocIterator old_cur(cur.buffer());
3257 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3261 } while (cur && cur.depth() > d && match(cur).match_len > 0);
3263 int max_match = match(cur).match_len; /* match valid only if not searching whole words */
3264 if (max_match <= 0) return 0;
3265 LYXERR(Debug::FIND, "Ok");
3267 // Compute the match length
3269 if (cur.pos() + len > cur.lastpos())
3271 if (match.opt.matchword) {
3272 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3273 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3275 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3277 // Length of matched text (different from len param)
3278 int old_match = match(cur, len).match_len;
3282 // Greedy behaviour while matching regexps
3283 while ((new_match = match(cur, len + 1).match_len) > old_match) {
3285 old_match = new_match;
3286 LYXERR(Debug::FIND, "verifying match with len = " << len);
3293 int maxl = cur.lastpos() - cur.pos();
3294 // Greedy behaviour while matching regexps
3295 while (maxl > minl) {
3296 int actual_match = match(cur, len).match_len;
3297 if (actual_match >= max_match) {
3298 // actual_match > max_match _can_ happen,
3299 // if the search area splits
3300 // some following word so that the regex
3301 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3303 // This means, the len value is too big
3305 len = (int)((maxl + minl)/2);
3308 // (actual_match < max_match)
3310 len = (int)((maxl + minl)/2);
3314 // Search for real start of matched characters
3319 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3320 if (cur.depth() < old_cur.depth()) {
3322 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3325 if (cur.pos() != old_cur.pos()) {
3326 // OK, forwarded 1 pos in actual inset
3327 actual_match = match(cur, len-1).match_len;
3328 if (actual_match == max_match) {
3329 // Ha, got it! The shorter selection has the same match length
3334 // OK, the shorter selection matches less chars, revert to previous value
3340 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3341 actual_match = match(cur, len).match_len;
3342 if (actual_match == max_match)
3352 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3356 while (!theApp()->longOperationCancelled() && cur) {
3357 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3358 MatchResult mres = match(cur, -1, false);
3359 int match_len = mres.match_len;
3360 LYXERR(Debug::FIND, "match_len: " << match_len);
3361 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3362 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3365 if (match_len > 0) {
3366 // Try to find the begin of searched string
3367 int increment = mres.pos/2;
3368 while (mres.pos > 5 && (increment > 5)) {
3369 DocIterator old_cur = cur;
3370 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3372 if (! cur || (cur.pit() > old_cur.pit())) {
3373 // Are we outside of the paragraph?
3374 // This can happen if moving past some UTF8-encoded chars
3379 MatchResult mres2 = match(cur, -1, false);
3380 if ((mres2.match2end < mres.match2end) ||
3381 (mres2.match_len < mres.match_len)) {
3388 if (increment > mres.pos/2)
3389 increment = mres.pos/2;
3393 int match_len_zero_count = 0;
3394 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3396 int remaining_len = match(cur, -1, false).match_len;
3397 if (remaining_len <= 0) {
3398 // Apparently the searched string is not in the remaining part
3405 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3406 int match_len3 = match(cur, 1).match_len;
3409 int match_len2 = match(cur).match_len;
3410 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3411 if (match_len2 > 0) {
3412 // Sometimes in finalize we understand it wasn't a match
3413 // and we need to continue the outest loop
3414 int len = findAdvFinalize(cur, match);
3419 if (match_len2 >= 0) {
3420 if (match_len2 == 0)
3421 match_len_zero_count++;
3423 match_len_zero_count = 0;
3426 if (++match_len_zero_count > 3) {
3427 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3435 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3436 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3439 // This should exit nested insets, if any, or otherwise undefine the currsor.
3440 cur.pos() = cur.lastpos();
3441 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3449 /// Find the most backward consecutive match within same paragraph while searching backwards.
3450 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3452 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3453 DocIterator tmp_cur = cur;
3454 int len = findAdvFinalize(tmp_cur, match);
3455 Inset & inset = cur.inset();
3456 for (; cur != cur_begin; cur.backwardPos()) {
3457 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3458 DocIterator new_cur = cur;
3459 new_cur.backwardPos();
3460 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3462 int new_len = findAdvFinalize(new_cur, match);
3467 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3473 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3477 // Backup of original position
3478 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3479 if (cur == cur_begin)
3482 DocIterator cur_orig(cur);
3483 bool pit_changed = false;
3486 bool found_match = (match(cur, -1, false).match_len > 0);
3490 cur.pos() = cur.lastpos();
3492 cur.pos() = cur_orig.pos();
3493 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3494 DocIterator cur_prev_iter;
3496 found_match = (match(cur).match_len > 0);
3497 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3498 << found_match << ", cur: " << cur);
3500 return findMostBackwards(cur, match);
3502 // Stop if begin of document reached
3503 if (cur == cur_begin)
3505 cur_prev_iter = cur;
3509 if (cur == cur_begin)
3516 } while (!theApp()->longOperationCancelled());
3524 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3525 DocIterator const & cur, int len)
3527 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3529 if (!opt.ignoreformat)
3530 return latexifyFromCursor(cur, len);
3532 return stringifyFromCursor(cur, len);
3536 FindAndReplaceOptions::FindAndReplaceOptions(
3537 docstring const & _find_buf_name, bool _casesensitive,
3538 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3539 docstring const & _repl_buf_name, bool _keep_case,
3540 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3541 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3542 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3543 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3551 /** Check if 'len' letters following cursor are all non-lowercase */
3552 static bool allNonLowercase(Cursor const & cur, int len)
3554 pos_type beg_pos = cur.selectionBegin().pos();
3555 pos_type end_pos = cur.selectionBegin().pos() + len;
3556 if (len > cur.lastpos() + 1 - beg_pos) {
3557 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3558 len = cur.lastpos() + 1 - beg_pos;
3559 end_pos = beg_pos + len;
3561 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3562 if (isLowerCase(cur.paragraph().getChar(pos)))
3568 /** Check if first letter is upper case and second one is lower case */
3569 static bool firstUppercase(Cursor const & cur)
3572 pos_type pos = cur.selectionBegin().pos();
3573 if (pos >= cur.lastpos() - 1) {
3574 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3577 ch1 = cur.paragraph().getChar(pos);
3578 ch2 = cur.paragraph().getChar(pos + 1);
3579 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3580 LYXERR(Debug::FIND, "firstUppercase(): "
3581 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3582 << ch2 << "(" << char(ch2) << ")"
3583 << ", result=" << result << ", cur=" << cur);
3588 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3590 ** \fixme What to do with possible further paragraphs in replace buffer ?
3592 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3594 ParagraphList::iterator pit = buffer.paragraphs().begin();
3595 LASSERT(!pit->empty(), /**/);
3596 pos_type right = pos_type(1);
3597 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3598 right = pit->size();
3599 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3605 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3607 Cursor & cur = bv->cursor();
3608 if (opt.repl_buf_name.empty()
3609 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3610 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3613 DocIterator sel_beg = cur.selectionBegin();
3614 DocIterator sel_end = cur.selectionEnd();
3615 if (&sel_beg.inset() != &sel_end.inset()
3616 || sel_beg.pit() != sel_end.pit()
3617 || sel_beg.idx() != sel_end.idx())
3619 int sel_len = sel_end.pos() - sel_beg.pos();
3620 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3621 << ", sel_len: " << sel_len << endl);
3624 LASSERT(sel_len > 0, return 0);
3626 if (!matchAdv(sel_beg, sel_len).match_len)
3629 // Build a copy of the replace buffer, adapted to the KeepCase option
3630 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3632 repl_buffer_orig.write(oss);
3633 string lyx = oss.str();
3634 Buffer repl_buffer("", false);
3635 repl_buffer.setUnnamed(true);
3636 LASSERT(repl_buffer.readString(lyx), return 0);
3637 if (opt.keep_case && sel_len >= 2) {
3638 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3639 if (cur.inTexted()) {
3640 if (firstUppercase(cur))
3641 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3642 else if (allNonLowercase(cur, sel_len))
3643 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3646 cap::cutSelection(cur, false);
3647 if (cur.inTexted()) {
3648 repl_buffer.changeLanguage(
3649 repl_buffer.language(),
3650 cur.getFont().language());
3651 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3652 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3653 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3654 repl_buffer.params().documentClassPtr(),
3655 bv->buffer().errorList("Paste"));
3656 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3657 sel_len = repl_buffer.paragraphs().begin()->size();
3658 } else if (cur.inMathed()) {
3659 odocstringstream ods;
3661 // OutputParams runparams(&repl_buffer.params().encoding());
3662 OutputParams runparams(encodings.fromLyXName("utf8"));
3663 runparams.nice = false;
3664 runparams.flavor = OutputParams::XETEX;
3665 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3666 runparams.dryrun = true;
3667 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3668 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3669 docstring repl_latex = ods.str();
3670 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3672 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3673 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3674 repl_latex = from_utf8(s);
3675 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3676 MathData ar(cur.buffer());
3677 asArray(repl_latex, ar, Parse::NORMAL);
3679 sel_len = ar.size();
3680 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3682 if (cur.pos() >= sel_len)
3683 cur.pos() -= sel_len;
3686 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3687 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3688 bv->processUpdateFlags(Update::Force);
3693 /// Perform a FindAdv operation.
3694 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3699 // e.g., when invoking word-findadv from mini-buffer wither with
3700 // wrong options syntax or before ever opening advanced F&R pane
3701 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3705 MatchStringAdv matchAdv(bv->buffer(), opt);
3706 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3708 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3709 num_replaced += findAdvReplace(bv, opt, matchAdv);
3712 match_len = findForwardAdv(cur, matchAdv);
3714 match_len = findBackwardsAdv(cur, matchAdv);
3716 // This may only be raised by lyx::regex()
3717 bv->message(_("Invalid regular expression!"));
3721 if (match_len == 0) {
3722 if (num_replaced > 0) {
3723 switch (num_replaced)
3726 bv->message(_("One match has been replaced."));
3729 bv->message(_("Two matches have been replaced."));
3732 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
3738 bv->message(_("Match not found."));
3743 if (num_replaced > 0)
3744 bv->message(_("Match has been replaced."));
3746 bv->message(_("Match found."));
3748 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3749 bv->putSelectionAt(cur, match_len, !opt.forward);
3755 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3757 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3758 << opt.casesensitive << ' '
3759 << opt.matchword << ' '
3760 << opt.forward << ' '
3761 << opt.expandmacros << ' '
3762 << opt.ignoreformat << ' '
3763 << opt.replace_all << ' '
3764 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3765 << opt.keep_case << ' '
3766 << int(opt.scope) << ' '
3769 LYXERR(Debug::FIND, "built: " << os.str());
3775 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3777 LYXERR(Debug::FIND, "parsing");
3781 while (line != "EOSS") {
3785 if (is.eof()) // Tolerate malformed request
3789 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3790 opt.find_buf_name = from_utf8(s);
3791 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
3792 is.get(); // Waste space before replace string
3795 while (line != "EOSS") {
3799 if (is.eof()) // Tolerate malformed request
3803 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3804 opt.repl_buf_name = from_utf8(s);
3805 is >> opt.keep_case;
3808 opt.scope = FindAndReplaceOptions::SearchScope(i);
3810 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3812 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3813 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3814 << opt.scope << ' ' << opt.restr);