3 * This file is part of LyX, the document processor.
4 * License details can be found in the file COPYING.
6 * \author Lars Gullik Bjønnes
9 * \author Alfredo Braunstein
10 * \author Tommaso Cucinotta
12 * Full author contact details are available in file CREDITS.
20 #include "BufferList.h"
21 #include "BufferParams.h"
22 #include "BufferView.h"
25 #include "CutAndPaste.h"
26 #include "FuncRequest.h"
28 #include "output_latex.h"
29 #include "OutputParams.h"
30 #include "Paragraph.h"
34 #include "frontends/Application.h"
35 #include "frontends/alert.h"
37 #include "mathed/InsetMath.h"
38 #include "mathed/InsetMathHull.h"
39 #include "mathed/MathData.h"
40 #include "mathed/MathStream.h"
41 #include "mathed/MathSupport.h"
43 #include "support/debug.h"
44 #include "support/docstream.h"
45 #include "support/FileName.h"
46 #include "support/gettext.h"
47 #include "support/lassert.h"
48 #include "support/lstrings.h"
50 #include "support/regex.h"
51 #include "support/textutils.h"
55 using namespace lyx::support;
60 // Helper class for deciding what should be ignored
64 IgnoreFormats() = default;
66 bool getFamily() const { return ignoreFamily_; }
68 bool getSeries() const { return ignoreSeries_; }
70 bool getShape() const { return ignoreShape_; }
72 bool getUnderline() const { return ignoreUnderline_; }
74 bool getMarkUp() const { return ignoreMarkUp_; }
76 bool getStrikeOut() const { return ignoreStrikeOut_; }
78 bool getSectioning() const { return ignoreSectioning_; }
80 bool getFrontMatter() const { return ignoreFrontMatter_; }
82 bool getColor() const { return ignoreColor_; }
84 bool getLanguage() const { return ignoreLanguage_; }
86 void setIgnoreFormat(string const & type, bool value);
90 bool ignoreFamily_ = false;
92 bool ignoreSeries_ = false;
94 bool ignoreShape_ = false;
96 bool ignoreUnderline_ = false;
98 bool ignoreMarkUp_ = false;
100 bool ignoreStrikeOut_ = false;
102 bool ignoreSectioning_ = false;
104 bool ignoreFrontMatter_ = false;
106 bool ignoreColor_ = false;
108 bool ignoreLanguage_ = false;
112 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
114 if (type == "color") {
115 ignoreColor_ = value;
117 else if (type == "language") {
118 ignoreLanguage_ = value;
120 else if (type == "sectioning") {
121 ignoreSectioning_ = value;
122 ignoreFrontMatter_ = value;
124 else if (type == "font") {
125 ignoreSeries_ = value;
126 ignoreShape_ = value;
127 ignoreFamily_ = value;
129 else if (type == "series") {
130 ignoreSeries_ = value;
132 else if (type == "shape") {
133 ignoreShape_ = value;
135 else if (type == "family") {
136 ignoreFamily_ = value;
138 else if (type == "markup") {
139 ignoreMarkUp_ = value;
141 else if (type == "underline") {
142 ignoreUnderline_ = value;
144 else if (type == "strike") {
145 ignoreStrikeOut_ = value;
149 // The global variable that can be changed from outside
150 IgnoreFormats ignoreFormats;
153 void setIgnoreFormat(string const & type, bool value)
155 ignoreFormats.setIgnoreFormat(type, value);
161 bool parse_bool(docstring & howto)
166 howto = split(howto, var, ' ');
174 MatchString(docstring const & s, bool cs, bool mw)
175 : str(s), case_sens(cs), whole_words(mw)
178 // returns true if the specified string is at the specified position
179 // del specifies whether deleted strings in ct mode will be considered
180 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
182 return par.find(str, case_sens, whole_words, pos, del);
190 // match whole words only
195 int findForward(DocIterator & cur, MatchString const & match,
196 bool find_del = true)
198 for (; cur; cur.forwardChar())
199 if (cur.inTexted()) {
200 int len = match(cur.paragraph(), cur.pos(), find_del);
208 int findBackwards(DocIterator & cur, MatchString const & match,
209 bool find_del = true)
213 if (cur.inTexted()) {
214 int len = match(cur.paragraph(), cur.pos(), find_del);
223 bool searchAllowed(docstring const & str)
226 frontend::Alert::error(_("Search error"), _("Search string is empty"));
233 bool findOne(BufferView * bv, docstring const & searchstr,
234 bool case_sens, bool whole, bool forward,
235 bool find_del = true, bool check_wrap = false)
237 if (!searchAllowed(searchstr))
240 DocIterator cur = forward
241 ? bv->cursor().selectionEnd()
242 : bv->cursor().selectionBegin();
244 MatchString const match(searchstr, case_sens, whole);
246 int match_len = forward
247 ? findForward(cur, match, find_del)
248 : findBackwards(cur, match, find_del);
251 bv->putSelectionAt(cur, match_len, !forward);
252 else if (check_wrap) {
253 DocIterator cur_orig(bv->cursor());
256 q = _("End of file reached while searching forward.\n"
257 "Continue searching from the beginning?");
259 q = _("Beginning of file reached while searching backward.\n"
260 "Continue searching from the end?");
261 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
262 q, 0, 1, _("&Yes"), _("&No"));
263 if (wrap_answer == 0) {
265 bv->cursor().clear();
266 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
268 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
269 bv->cursor().backwardPos();
271 bv->clearSelection();
272 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
275 bv->cursor().setCursor(cur_orig);
279 return match_len > 0;
283 int replaceAll(BufferView * bv,
284 docstring const & searchstr, docstring const & replacestr,
285 bool case_sens, bool whole)
287 Buffer & buf = bv->buffer();
289 if (!searchAllowed(searchstr) || buf.isReadonly())
292 DocIterator cur_orig(bv->cursor());
294 MatchString const match(searchstr, case_sens, whole);
297 int const rsize = replacestr.size();
298 int const ssize = searchstr.size();
301 cur.setCursor(doc_iterator_begin(&buf));
302 int match_len = findForward(cur, match, false);
303 while (match_len > 0) {
304 // Backup current cursor position and font.
305 pos_type const pos = cur.pos();
306 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
308 int striked = ssize -
309 cur.paragraph().eraseChars(pos, pos + match_len,
310 buf.params().track_changes);
311 cur.paragraph().insert(pos, replacestr, font,
312 Change(buf.params().track_changes
314 : Change::UNCHANGED));
315 for (int i = 0; i < rsize + striked; ++i)
318 match_len = findForward(cur, match, false);
321 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
323 cur_orig.fixIfBroken();
324 bv->setCursor(cur_orig);
330 // the idea here is that we are going to replace the string that
331 // is selected IF it is the search string.
332 // if there is a selection, but it is not the search string, then
333 // we basically ignore it. (FIXME We ought to replace only within
335 // if there is no selection, then:
336 // (i) if some search string has been provided, then we find it.
337 // (think of how the dialog works when you hit "replace" the
339 // (ii) if no search string has been provided, then we treat the
340 // word the cursor is in as the search string. (why? i have no
341 // idea.) but this only works in text?
343 // returns the number of replacements made (one, if any) and
344 // whether anything at all was done.
345 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
346 docstring const & replacestr, bool case_sens,
347 bool whole, bool forward, bool findnext)
349 Cursor & cur = bv->cursor();
350 if (!cur.selection()) {
351 // no selection, non-empty search string: find it
352 if (!searchstr.empty()) {
353 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
354 return make_pair(found, 0);
356 // empty search string
359 return make_pair(false, 0);
360 // select current word and treat it as the search string.
361 // This causes a minor bug as undo will restore this selection,
362 // which the user did not create (#8986).
363 cur.innerText()->selectWord(cur, WHOLE_WORD);
364 searchstr = cur.selectionAsString(false);
367 // if we still don't have a search string, report the error
369 if (!searchAllowed(searchstr))
370 return make_pair(false, 0);
372 bool have_selection = cur.selection();
373 docstring const selected = cur.selectionAsString(false);
376 ? searchstr == selected
377 : compare_no_case(searchstr, selected) == 0;
379 // no selection or current selection is not search word:
380 // just find the search word
381 if (!have_selection || !match) {
382 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
383 return make_pair(found, 0);
386 // we're now actually ready to replace. if the buffer is
387 // read-only, we can't, though.
388 if (bv->buffer().isReadonly())
389 return make_pair(false, 0);
391 cap::replaceSelectionWithString(cur, replacestr);
393 cur.pos() += replacestr.length();
394 LASSERT(cur.pos() <= cur.lastpos(),
395 cur.pos() = cur.lastpos());
398 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
400 return make_pair(true, 1);
406 docstring const find2string(docstring const & search,
407 bool casesensitive, bool matchword, bool forward)
411 << int(casesensitive) << ' '
412 << int(matchword) << ' '
418 docstring const replace2string(docstring const & replace,
419 docstring const & search,
420 bool casesensitive, bool matchword,
421 bool all, bool forward, bool findnext)
424 ss << replace << '\n'
426 << int(casesensitive) << ' '
427 << int(matchword) << ' '
429 << int(forward) << ' '
435 bool lyxfind(BufferView * bv, FuncRequest const & ev)
437 if (!bv || ev.action() != LFUN_WORD_FIND)
440 //lyxerr << "find called, cmd: " << ev << endl;
442 // data is of the form
444 // <casesensitive> <matchword> <forward>"
446 docstring howto = split(ev.argument(), search, '\n');
448 bool casesensitive = parse_bool(howto);
449 bool matchword = parse_bool(howto);
450 bool forward = parse_bool(howto);
452 return findOne(bv, search, casesensitive, matchword, forward, true, true);
456 bool lyxreplace(BufferView * bv,
457 FuncRequest const & ev, bool has_deleted)
459 if (!bv || ev.action() != LFUN_WORD_REPLACE)
462 // data is of the form
465 // <casesensitive> <matchword> <all> <forward> <findnext>"
468 docstring howto = split(ev.argument(), rplc, '\n');
469 howto = split(howto, search, '\n');
471 bool casesensitive = parse_bool(howto);
472 bool matchword = parse_bool(howto);
473 bool all = parse_bool(howto);
474 bool forward = parse_bool(howto);
475 bool findnext = howto.empty() ? true : parse_bool(howto);
480 int replace_count = 0;
482 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
483 update = replace_count > 0;
486 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
488 replace_count = rv.second;
491 Buffer const & buf = bv->buffer();
493 // emit message signal.
494 buf.message(_("String not found."));
496 if (replace_count == 0) {
497 buf.message(_("String found."));
498 } else if (replace_count == 1) {
499 buf.message(_("String has been replaced."));
501 docstring const str =
502 bformat(_("%1$d strings have been replaced."), replace_count);
506 } else if (findnext) {
507 // if we have deleted characters, we do not replace at all, but
508 // rather search for the next occurence
509 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
512 bv->message(_("String not found."));
518 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
520 for (; cur; cur.forwardPos())
521 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
525 DocIterator cur_orig(bv->cursor());
526 docstring q = _("End of file reached while searching forward.\n"
527 "Continue searching from the beginning?");
528 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
529 q, 0, 1, _("&Yes"), _("&No"));
530 if (wrap_answer == 0) {
531 bv->cursor().clear();
532 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
533 bv->clearSelection();
534 cur.setCursor(bv->cursor().selectionBegin());
535 if (findNextChange(bv, cur, false))
538 bv->cursor().setCursor(cur_orig);
545 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
547 for (cur.backwardPos(); cur; cur.backwardPos()) {
548 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
553 DocIterator cur_orig(bv->cursor());
554 docstring q = _("Beginning of file reached while searching backward.\n"
555 "Continue searching from the end?");
556 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
557 q, 0, 1, _("&Yes"), _("&No"));
558 if (wrap_answer == 0) {
559 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
560 bv->cursor().backwardPos();
561 bv->clearSelection();
562 cur.setCursor(bv->cursor().selectionBegin());
563 if (findPreviousChange(bv, cur, false))
566 bv->cursor().setCursor(cur_orig);
573 bool selectChange(Cursor & cur, bool forward)
575 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
577 Change ch = cur.paragraph().lookupChange(cur.pos());
579 CursorSlice tip1 = cur.top();
580 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
581 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
582 if (!ch2.isSimilarTo(ch))
585 CursorSlice tip2 = cur.top();
586 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
588 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
589 if (!ch2.isSimilarTo(ch)) {
590 // take a step forward to correctly set the selection
598 cur.bv().mouseSetCursor(cur, false);
600 cur.bv().mouseSetCursor(cur, true);
608 bool findChange(BufferView * bv, bool forward)
611 cur.setCursor(forward ? bv->cursor().selectionEnd()
612 : bv->cursor().selectionBegin());
613 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
614 return selectChange(cur, forward);
619 bool findNextChange(BufferView * bv)
621 return findChange(bv, true);
625 bool findPreviousChange(BufferView * bv)
627 return findChange(bv, false);
634 typedef vector<pair<string, string> > Escapes;
636 /// A map of symbols and their escaped equivalent needed within a regex.
637 /// @note Beware of order
638 Escapes const & get_regexp_escapes()
640 typedef std::pair<std::string, std::string> P;
642 static Escapes escape_map;
643 if (escape_map.empty()) {
644 escape_map.push_back(P("$", "_x_$"));
645 escape_map.push_back(P("{", "_x_{"));
646 escape_map.push_back(P("}", "_x_}"));
647 escape_map.push_back(P("[", "_x_["));
648 escape_map.push_back(P("]", "_x_]"));
649 escape_map.push_back(P("(", "_x_("));
650 escape_map.push_back(P(")", "_x_)"));
651 escape_map.push_back(P("+", "_x_+"));
652 escape_map.push_back(P("*", "_x_*"));
653 escape_map.push_back(P(".", "_x_."));
654 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
655 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
656 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
657 escape_map.push_back(P("_x_", "\\"));
662 /// A map of lyx escaped strings and their unescaped equivalent.
663 Escapes const & get_lyx_unescapes()
665 typedef std::pair<std::string, std::string> P;
667 static Escapes escape_map;
668 if (escape_map.empty()) {
669 escape_map.push_back(P("\\%", "%"));
670 escape_map.push_back(P("\\mathcircumflex ", "^"));
671 escape_map.push_back(P("\\mathcircumflex", "^"));
672 escape_map.push_back(P("\\backslash ", "\\"));
673 escape_map.push_back(P("\\backslash", "\\"));
674 escape_map.push_back(P("\\\\{", "_x_<"));
675 escape_map.push_back(P("\\\\}", "_x_>"));
676 escape_map.push_back(P("\\sim ", "~"));
677 escape_map.push_back(P("\\sim", "~"));
682 /// A map of escapes turning a regexp matching text to one matching latex.
683 Escapes const & get_regexp_latex_escapes()
685 typedef std::pair<std::string, std::string> P;
687 static Escapes escape_map;
688 if (escape_map.empty()) {
689 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
690 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
691 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
692 escape_map.push_back(P("\\[", "\\{\\[\\}"));
693 escape_map.push_back(P("\\]", "\\{\\]\\}"));
694 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
695 escape_map.push_back(P("%", "\\\\\\%"));
696 escape_map.push_back(P("#", "\\\\#"));
701 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
702 ** the found occurrence were escaped.
704 string apply_escapes(string s, Escapes const & escape_map)
706 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
707 Escapes::const_iterator it;
708 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
709 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
710 unsigned int pos = 0;
711 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
712 s.replace(pos, it->first.length(), it->second);
713 LYXERR(Debug::FIND, "After escape: " << s);
714 pos += it->second.length();
715 // LYXERR(Debug::FIND, "pos: " << pos);
718 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
723 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
724 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
725 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
726 string escape_for_regex(string s, bool match_latex)
729 while (pos < s.size()) {
730 size_t new_pos = s.find("\\regexp{", pos);
731 if (new_pos == string::npos)
735 LYXERR(Debug::FIND, "new_pos: " << new_pos);
736 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
737 LYXERR(Debug::FIND, "t [lyx]: " << t);
738 t = apply_escapes(t, get_regexp_escapes());
739 LYXERR(Debug::FIND, "t [rxp]: " << t);
740 s.replace(pos, new_pos - pos, t);
741 new_pos = pos + t.size();
742 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
743 LYXERR(Debug::FIND, "new_pos: " << new_pos);
744 if (new_pos == s.size())
747 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
748 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
749 LYXERR(Debug::FIND, "end_pos: " << end_pos);
750 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
751 LYXERR(Debug::FIND, "t in regexp : " << t);
752 t = apply_escapes(t, get_lyx_unescapes());
753 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
755 t = apply_escapes(t, get_regexp_latex_escapes());
756 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
758 if (end_pos == s.size()) {
759 s.replace(new_pos, end_pos - new_pos, t);
760 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
763 s.replace(new_pos, end_pos + 13 - new_pos, t);
764 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
765 pos = new_pos + t.size();
766 LYXERR(Debug::FIND, "pos: " << pos);
772 /// Wrapper for lyx::regex_replace with simpler interface
773 bool regex_replace(string const & s, string & t, string const & searchstr,
774 string const & replacestr)
776 lyx::regex e(searchstr, regex_constants::ECMAScript);
778 ostream_iterator<char, char> it(oss);
779 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
780 // tolerate t and s be references to the same variable
781 bool rv = (s != oss.str());
787 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
789 ** Verify that closed braces exactly match open braces. This avoids that, for example,
790 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
793 ** Number of open braces that must remain open at the end for the verification to succeed.
795 bool braces_match(string::const_iterator const & beg,
796 string::const_iterator const & end,
800 string::const_iterator it = beg;
801 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
802 for (; it != end; ++it) {
803 // Skip escaped braces in the count
808 } else if (*it == '{') {
810 } else if (*it == '}') {
811 if (open_pars == 0) {
812 LYXERR(Debug::FIND, "Found unmatched closed brace");
818 if (open_pars != unmatched) {
819 LYXERR(Debug::FIND, "Found " << open_pars
820 << " instead of " << unmatched
821 << " unmatched open braces at the end of count");
824 LYXERR(Debug::FIND, "Braces match as expected");
834 MatchResult(): match_len(0),match2end(0), pos(0) {};
837 /** The class performing a match between a position in the document and the FindAdvOptions.
839 class MatchStringAdv {
841 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
843 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
844 ** constructor as opt.search, under the opt.* options settings.
847 ** If set, then match is searched only against beginning of text starting at cur.
848 ** If unset, then match is searched anywhere in text starting at cur.
851 ** The length of the matching text, or zero if no match was found.
853 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
858 /// first buffer on which search was started
859 lyx::Buffer * const p_first_buf;
861 FindAndReplaceOptions const & opt;
864 /// Auxiliary find method (does not account for opt.matchword)
865 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
867 /** Normalize a stringified or latexified LyX paragraph.
871 ** <li>if search is not casesensitive, then lowercase the string;
872 ** <li>remove any newline at begin or end of the string;
873 ** <li>replace any newline in the middle of the string with a simple space;
874 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
877 ** @todo Normalization should also expand macros, if the corresponding
878 ** search option was checked.
880 string normalize(docstring const & s, bool hack_braces) const;
881 // normalized string to search
882 string par_as_string;
883 // regular expression to use for searching
885 // same as regexp, but prefixed with a ".*?"
887 // leading format material as string
888 string lead_as_string;
889 // par_as_string after removal of lead_as_string
890 string par_as_string_nolead;
891 // unmatched open braces in the search string/regexp
893 // number of (.*?) subexpressions added at end of search regexp for closing
894 // environments, math mode, styles, etc...
896 // Are we searching with regular expressions ?
901 static docstring buffer_to_latex(Buffer & buffer)
903 //OutputParams runparams(&buffer.params().encoding());
904 OutputParams runparams(encodings.fromLyXName("utf8"));
905 odocstringstream ods;
907 runparams.nice = true;
908 runparams.flavor = OutputParams::XETEX;
909 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
910 // No side effect of file copying and image conversion
911 runparams.dryrun = true;
912 runparams.for_search = true;
913 pit_type const endpit = buffer.paragraphs().size();
914 for (pit_type pit = 0; pit != endpit; ++pit) {
915 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
916 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
922 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
925 if (!opt.ignoreformat) {
926 str = buffer_to_latex(buffer);
928 // OutputParams runparams(&buffer.params().encoding());
929 OutputParams runparams(encodings.fromLyXName("utf8"));
930 runparams.nice = true;
931 runparams.flavor = OutputParams::XETEX;
932 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
933 runparams.dryrun = true;
934 runparams.for_search = true;
935 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
936 Paragraph const & par = buffer.paragraphs().at(pit);
937 LYXERR(Debug::FIND, "Adding to search string: '"
938 << par.asString(pos_type(0), par.size(),
939 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
942 str += par.asString(pos_type(0), par.size(),
943 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
951 /// Return separation pos between the leading material and the rest
952 static size_t identifyLeading(string const & s)
955 // @TODO Support \item[text]
956 // Kornel: Added textsl, textsf, textit, texttt and noun
957 // + allow to search for colored text too
958 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)\\*?)\\{", "")
959 || regex_replace(t, t, REGEX_BOS "\\$", "")
960 || regex_replace(t, t, REGEX_BOS "\\\\\\[", "")
961 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
962 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
964 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
969 * Given a latexified string, retrieve some handled features
970 * The features of the regex will later be compared with the features
971 * of the searched text. If the regex features are not a
972 * subset of the analized, then, in not format ignoring search
973 * we can early stop the search in the relevant inset.
975 typedef map<string, bool> Features;
977 static Features identifyFeatures(string const & s)
979 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
980 static regex const valid("^("
982 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
983 "emph|noun|text(bf|md|sl|sf|it|tt)|"
984 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
985 "(u|uu)line|(s|x)out|uwave|"
986 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
987 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
992 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
995 if (sub.str(1).compare("regexp") == 0) {
999 string token = sub.str(1);
1001 if (regex_match(token, sub2, valid)) {
1009 if (sub.str(1).compare("endregexp") == 0) {
1019 * defines values features of a key "\\[a-z]+{"
1024 /* Char type with content discarded
1025 * like \hspace{1cm} */
1027 /* Char, like \backslash */
1029 /* \part, \section*, ... */
1031 /* title, author etc */
1033 /* \foreignlanguage{ngerman}, ... */
1036 * to discard language in content */
1039 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1041 /* fonts, colors, markups, ... */
1043 /* footnotesize, ... large, ...
1044 * Ignore all of them */
1047 /* inputencoding, ...
1048 * Discard also content, because they do not help in search */
1051 * like remove, but also all arguments */
1053 /* item, listitem */
1055 /* tex, latex, ... like isChar */
1057 /* like \lettrine[lines=5]{}{} */
1059 /* End of arguments marker for lettrine,
1060 * so that they can be ignored */
1063 KeyInfo() = default;
1064 KeyInfo(KeyType type, int parcount, bool disable)
1066 parenthesiscount(parcount),
1067 disabled(disable) {}
1068 KeyType keytype = invalid;
1070 int _tokensize = -1;
1071 int _tokenstart = -1;
1072 int _dataStart = -1;
1074 int parenthesiscount = 1;
1075 bool disabled = false;
1076 bool used = false; /* by pattern */
1081 Border(int l=0, int u=0) : low(l), upper(u) {};
1086 #define MAXOPENED 30
1088 bool isPatternString_;
1090 explicit Intervall(bool isPattern, string const & p) :
1091 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1092 hasTitle(false), langcount(0)
1100 static vector<Border> borders;
1101 int depts[MAXOPENED];
1102 int closes[MAXOPENED];
1103 int actualdeptindex;
1104 int previousNotIgnored(int);
1105 int nextNotIgnored(int);
1106 void handleOpenP(int i);
1107 void handleCloseP(int i, bool closingAllowed);
1108 void resetOpenedP(int openPos);
1109 void addIntervall(int upper);
1110 void addIntervall(int low, int upper); /* if explicit */
1111 void removeAccents();
1112 void setForDefaultLang(KeyInfo const & defLang);
1113 int findclosing(int start, int end, char up, char down, int repeat);
1114 void handleParentheses(int lastpos, bool closingAllowed);
1116 int langcount; // Number of disabled language specs up to current position in actual interval
1117 int isOpeningPar(int pos);
1119 void output(ostringstream &os, int lastpos);
1120 // string show(int lastpos);
1123 vector<Border> Intervall::borders = vector<Border>(30);
1125 int Intervall::isOpeningPar(int pos)
1127 if ((pos < 0) || (size_t(pos) >= par.size()))
1129 if (par[pos] != '{')
1131 if (size_t(pos) + 2 >= par.size())
1133 if (par[pos+2] != '}')
1135 if (par[pos+1] == '[' || par[pos+1] == ']')
1140 void Intervall::setForDefaultLang(KeyInfo const & defLang)
1142 // Enable the use of first token again
1143 if (ignoreidx >= 0) {
1144 int value = defLang._tokenstart + defLang._tokensize;
1146 if (borders[0].low < value)
1147 borders[0].low = value;
1148 if (borders[0].upper < value)
1149 borders[0].upper = value;
1154 static void checkDepthIndex(int val)
1156 static int maxdepthidx = MAXOPENED-2;
1157 static int lastmaxdepth = 0;
1158 if (val > lastmaxdepth) {
1159 LYXERR(Debug::INFO, "Depth reached " << val);
1162 if (val > maxdepthidx) {
1164 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1169 // Not needed, because borders are now dynamically expanded
1170 static void checkIgnoreIdx(int val)
1172 static int lastmaxignore = -1;
1173 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1174 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1175 lastmaxignore = val;
1181 * Expand the region of ignored parts of the input latex string
1182 * The region is only relevant in output()
1184 void Intervall::addIntervall(int low, int upper)
1187 if (low == upper) return;
1188 for (idx = ignoreidx+1; idx > 0; --idx) {
1189 if (low > borders[idx-1].upper) {
1193 Border br(low, upper);
1194 if (idx > ignoreidx) {
1195 if (borders.size() <= size_t(idx)) {
1196 borders.push_back(br);
1202 // checkIgnoreIdx(ignoreidx);
1206 // Expand only if one of the new bound is inside the interwall
1207 // We know here that br.low > borders[idx-1].upper
1208 if (br.upper < borders[idx].low) {
1209 // We have to insert at this pos
1210 if (size_t(ignoreidx+1) >= borders.size()) {
1211 borders.push_back(borders[ignoreidx]);
1214 borders[ignoreidx+1] = borders[ignoreidx];
1216 for (int i = ignoreidx; i > idx; --i) {
1217 borders[i] = borders[i-1];
1221 // checkIgnoreIdx(ignoreidx);
1224 // Here we know, that we are overlapping
1225 if (br.low > borders[idx].low)
1226 br.low = borders[idx].low;
1227 // check what has to be concatenated
1229 for (int i = idx; i <= ignoreidx; i++) {
1230 if (br.upper >= borders[i].low) {
1232 if (br.upper < borders[i].upper)
1233 br.upper = borders[i].upper;
1239 // count should be >= 1 here
1242 for (int i = idx + count; i <= ignoreidx; i++) {
1243 borders[i-count+1] = borders[i];
1245 ignoreidx -= count - 1;
1251 typedef map<string, string> AccentsMap;
1252 static AccentsMap accents = map<string, string>();
1254 static void buildaccent(string n, string param, string values)
1258 const char delim = '|';
1259 while (getline(s, name, delim)) {
1261 for (char c : param) {
1262 string key = name + "{" + c + "}";
1263 // get the corresponding utf8-value
1264 if ((values[start] & 0xc0) != 0xc0) {
1265 // should not happen, utf8 encoding starts at least with 11xxxxxx
1266 // but value for '\dot{i}' is 'i', which is ascii
1267 if ((values[start] & 0x80) == 0) {
1269 accents[key] = values.substr(start, 1);
1270 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1275 for (int j = 1; ;j++) {
1276 if (start + j >= values.size()) {
1277 accents[key] = values.substr(start, j);
1278 start = values.size() - 1;
1281 else if ((values[start+j] & 0xc0) != 0x80) {
1282 // This is the first byte of following utf8 char
1283 accents[key] = values.substr(start, j);
1285 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1293 static void buildAccentsMap()
1295 accents["imath"] = "ı";
1297 accents["jmath"] = "ȷ";
1298 accents["lyxmathsym{ß}"] = "ß";
1299 accents["text{ß}"] = "ß";
1300 accents["ddot{\\imath}"] = "ï";
1301 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1302 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1303 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1304 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1305 accents["acute{\\imath}"] = "í";
1306 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1307 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1308 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1309 buildaccent("mathring|r", "aAuUwy",
1311 accents["check{\\imath}"] = "ǐ";
1312 accents["check{\\jmath}"] = "ǰ";
1313 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1314 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1315 accents["hat{\\imath}"] = "î";
1316 accents["hat{\\jmath}"] = "ĵ";
1317 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1318 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1319 accents["bar{\\imath}"] = "ī";
1320 buildaccent("bar|=", "aAeEiIoOuUyY",
1321 "āĀēĒīĪōŌūŪȳȲ"); // macron
1322 accents["tilde{\\imath}"] = "ĩ";
1323 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1324 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1325 accents["breve{\\imath}"] = "ĭ";
1326 buildaccent("breve|u", "aAeEgGiIoOuU",
1327 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1328 accents["grave{\\imath}"] = "ì";
1329 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1330 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1331 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1332 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1333 buildaccent("ogonek|k", "AaEeIiUuOo",
1334 "ĄąĘęĮįŲųǪǫ"); // ogonek
1335 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1336 "ÇçĢĢĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1337 buildaccent("subring|textsubring", "Aa",
1339 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1340 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1341 buildaccent("subtilde|textsubtilde", "EeIiUu",
1342 "ḚḛḬḭṴṵ"); // subtilde
1343 accents["dgrave{\\imath}"] = "ȉ";
1344 accents["textdoublegrave{\\i}"] = "ȉ";
1345 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1346 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1347 accents["rcap{\\imath}"] = "ȉ";
1348 accents["textroundcap{\\i}"] = "ȉ";
1349 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1350 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1351 buildaccent("slashed", "oO",
1356 * Created accents in math or regexp environment
1357 * are macros, but we need the utf8 equivalent
1359 void Intervall::removeAccents()
1361 if (accents.empty())
1363 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]))");
1365 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1367 string key = sub.str(1);
1368 if (accents.find(key) != accents.end()) {
1369 string val = accents[key];
1370 size_t pos = sub.position(size_t(0));
1371 for (size_t i = 0; i < val.size(); i++) {
1372 par[pos+i] = val[i];
1374 addIntervall(pos+val.size(), pos + sub.str(0).size());
1375 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1376 // remove traces of any remaining chars
1381 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1386 void Intervall::handleOpenP(int i)
1389 depts[actualdeptindex] = i+1;
1390 closes[actualdeptindex] = -1;
1391 checkDepthIndex(actualdeptindex);
1394 void Intervall::handleCloseP(int i, bool closingAllowed)
1396 if (actualdeptindex <= 0) {
1397 if (! closingAllowed)
1398 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1399 // if we are at the very end
1400 addIntervall(i, i+1);
1403 closes[actualdeptindex] = i+1;
1408 void Intervall::resetOpenedP(int openPos)
1410 // Used as initializer for foreignlanguage entry
1411 actualdeptindex = 1;
1412 depts[1] = openPos+1;
1416 int Intervall::previousNotIgnored(int start)
1418 int idx = 0; /* int intervalls */
1419 for (idx = ignoreidx; idx >= 0; --idx) {
1420 if (start > borders[idx].upper)
1422 if (start >= borders[idx].low)
1423 start = borders[idx].low-1;
1428 int Intervall::nextNotIgnored(int start)
1430 int idx = 0; /* int intervalls */
1431 for (idx = 0; idx <= ignoreidx; idx++) {
1432 if (start < borders[idx].low)
1434 if (start < borders[idx].upper)
1435 start = borders[idx].upper;
1440 typedef map<string, KeyInfo> KeysMap;
1441 typedef vector< KeyInfo> Entries;
1442 static KeysMap keys = map<string, KeyInfo>();
1448 Intervall interval_;
1449 void buildKeys(bool);
1450 void buildEntries(bool);
1451 void makeKey(const string &, KeyInfo, bool isPatternString);
1452 void processRegion(int start, int region_end); /* remove {} parts */
1453 void removeHead(KeyInfo const &, int count=0);
1456 LatexInfo(string const & par, bool isPatternString)
1457 : entidx_(-1), interval_(isPatternString, par)
1459 buildKeys(isPatternString);
1460 entries_ = vector<KeyInfo>();
1461 buildEntries(isPatternString);
1465 if (entries_.empty()) {
1468 if (entries_[0].keytype == KeyInfo::isTitle) {
1469 if (! entries_[0].disabled) {
1470 interval_.hasTitle = true;
1471 interval_.titleValue = entries_[0].head;
1474 interval_.hasTitle = false;
1475 interval_.titleValue = "";
1477 removeHead(entries_[0]);
1478 if (entries_.size() > 1)
1487 if (int(entries_.size()) > entidx_) {
1494 bool setNextKey(int idx) {
1495 if ((idx == entidx_) && (entidx_ >= 0)) {
1502 int find(int start, KeyInfo::KeyType keytype) {
1506 while (tmpIdx < int(entries_.size())) {
1507 if (entries_[tmpIdx].keytype == keytype)
1513 int process(ostringstream & os, KeyInfo const & actual);
1514 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1515 // string show(int lastpos) { return interval.show(lastpos);};
1516 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1517 KeyInfo &getKeyInfo(int keyinfo) {
1518 static KeyInfo invalidInfo = KeyInfo();
1519 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1522 return entries_[keyinfo];
1524 void setForDefaultLang(KeyInfo &defLang) {interval_.setForDefaultLang(defLang);};
1525 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1529 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1533 for (int i = start; i < end; i += 1 + skip) {
1537 if (c == '\\') skip = 1;
1541 else if (c == down) {
1544 if ((repeat <= 0) || (par[i+1] != up))
1562 vector<MathEntry> entries_;
1567 void insert(string const & wait, size_t start, size_t end) {
1568 MathEntry m = MathEntry();
1570 m.mathStart = start;
1572 m.mathSize = end - start;
1573 entries_.push_back(m);
1575 bool empty() { return entries_.empty(); };
1576 size_t getEndPos() {
1577 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1580 return entries_[actualIdx_].mathEnd;
1582 size_t getStartPos() {
1583 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1584 return 100000; /* definitely enough? */
1586 return entries_[actualIdx_].mathStart;
1588 size_t getFirstPos() {
1590 return getStartPos();
1593 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1596 return entries_[actualIdx_].mathSize;
1598 void incrEntry() { actualIdx_++; };
1601 void LatexInfo::buildEntries(bool isPatternString)
1603 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1604 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1605 static bool disableLanguageOverride = false;
1606 smatch sub, submath;
1607 bool evaluatingRegexp = false;
1609 bool evaluatingMath = false;
1610 bool evaluatingCode = false;
1612 bool evaluatingOptional = false;
1613 size_t optionalEnd = 0;
1616 bool math_end_waiting = false;
1617 size_t math_pos = 10000;
1620 interval_.removeAccents();
1622 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1624 if (math_end_waiting) {
1625 size_t pos = submath.position(size_t(0));
1626 if ((math_end == "$") &&
1627 (submath.str(0) == "$") &&
1628 (interval_.par[pos-1] != '\\')) {
1629 mi.insert("$", math_pos, pos + 1);
1630 math_end_waiting = false;
1632 else if ((math_end == "\\]") &&
1633 (submath.str(0) == "\\]")) {
1634 mi.insert("\\]", math_pos, pos + 2);
1635 math_end_waiting = false;
1637 else if ((submath.str(1).compare("end") == 0) &&
1638 (submath.str(2).compare(math_end) == 0)) {
1639 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1640 math_end_waiting = false;
1646 if (submath.str(1).compare("begin") == 0) {
1647 math_end_waiting = true;
1648 math_end = submath.str(2);
1649 math_pos = submath.position(size_t(0));
1651 else if (submath.str(0).compare("\\[") == 0) {
1652 math_end_waiting = true;
1654 math_pos = submath.position(size_t(0));
1656 else if (submath.str(0) == "$") {
1657 size_t pos = submath.position(size_t(0));
1658 if ((pos == 0) || (interval_.par[pos-1] != '\\')) {
1659 math_end_waiting = true;
1666 // Ignore language if there is math somewhere in pattern-string
1667 if (isPatternString) {
1670 keys["foreignlanguage"].disabled = true;
1671 disableLanguageOverride = true;
1674 disableLanguageOverride = false;
1677 if (disableLanguageOverride) {
1678 keys["foreignlanguage"].disabled = true;
1681 math_pos = mi.getFirstPos();
1682 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1684 string key = sub.str(3);
1686 if (sub.str(0)[0] == '\\')
1687 key = sub.str(0)[1];
1691 size_t k_pos = sub.position(size_t(0));
1692 if ((k_pos > 0) && (interval_.par[k_pos - 1] == '\\')) {
1693 // Escaped '$', ignoring
1699 if (evaluatingRegexp) {
1700 if (sub.str(1).compare("endregexp") == 0) {
1701 evaluatingRegexp = false;
1702 // found._tokenstart already set
1703 found._dataEnd = sub.position(size_t(0)) + 13;
1704 found._dataStart = found._dataEnd;
1705 found._tokensize = found._dataEnd - found._tokenstart;
1706 found.parenthesiscount = 0;
1707 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1714 if (evaluatingMath) {
1715 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1717 evaluatingMath = false;
1719 math_pos = mi.getStartPos();
1721 if (keys.find(key) == keys.end()) {
1722 found = KeyInfo(KeyInfo::isStandard, 0, true);
1723 if (isPatternString) {
1724 found.keytype = KeyInfo::isChar;
1725 found.disabled = false;
1732 if (key.compare("regexp") == 0) {
1733 evaluatingRegexp = true;
1734 found._tokenstart = sub.position(size_t(0));
1735 found._tokensize = 0;
1739 // Handle the other params of key
1740 if (found.keytype == KeyInfo::isIgnored)
1742 else if (found.keytype == KeyInfo::isMath) {
1743 if (size_t(sub.position(size_t(0))) == math_pos) {
1745 found._tokenstart = sub.position(size_t(0));
1746 found._tokensize = mi.getSize();
1747 found._dataEnd = found._tokenstart + found._tokensize;
1748 found._dataStart = found._dataEnd;
1749 found.parenthesiscount = 0;
1750 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1751 evaluatingMath = true;
1754 // begin|end of unknown env, discard
1755 // First handle tables
1756 // longtable|tabular
1757 bool discardComment;
1759 found.keytype = KeyInfo::doRemove;
1760 if ((sub.str(5).compare("longtable") == 0) ||
1761 (sub.str(5).compare("tabular") == 0)) {
1762 discardComment = true; /* '%' */
1765 discardComment = false;
1766 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1768 string token = sub.str(5);
1769 if (regex_match(token, sub2, removeArgs)) {
1770 found.keytype = KeyInfo::removeWithArg;
1773 // discard spaces before pos(0)
1774 int pos = sub.position(size_t(0));
1776 for (count = 0; pos - count > 0; count++) {
1777 char c = interval_.par[pos-count-1];
1778 if (discardComment) {
1779 if ((c != ' ') && (c != '%'))
1785 found._tokenstart = pos - count;
1786 if (sub.str(1).compare(0, 5, "begin") == 0) {
1787 size_t pos1 = pos + sub.str(0).length();
1788 if (sub.str(5).compare("cjk") == 0) {
1789 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1790 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1792 found.keytype = KeyInfo::isMain;
1793 found._dataStart = pos1;
1794 found._dataEnd = interval_.par.length();
1795 found.disabled = keys["foreignlanguage"].disabled;
1796 found.used = keys["foreignlanguage"].used;
1797 found._tokensize = pos1 - found._tokenstart;
1798 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1801 // Swallow possible optional params
1802 while (interval_.par[pos1] == '[') {
1803 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1805 // Swallow also the eventual parameter
1806 if (interval_.par[pos1] == '{') {
1807 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1810 found._dataEnd = pos1;
1812 found._dataStart = found._dataEnd;
1813 found._tokensize = count + found._dataEnd - pos;
1814 found.parenthesiscount = 0;
1815 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1816 found.disabled = true;
1820 // Handle "\end{...}"
1821 found._dataStart = pos + sub.str(0).length();
1822 found._dataEnd = found._dataStart;
1823 found._tokensize = count + found._dataEnd - pos;
1824 found.parenthesiscount = 0;
1825 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1826 found.disabled = true;
1830 else if (found.keytype != KeyInfo::isRegex) {
1831 found._tokenstart = sub.position(size_t(0));
1832 if (found.parenthesiscount == 0) {
1833 // Probably to be discarded
1834 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1835 char following = interval_.par[following_pos];
1836 if (following == ' ')
1837 found.head = "\\" + sub.str(3) + " ";
1838 else if (following == '=') {
1839 // like \uldepth=1000pt
1840 found.head = sub.str(0);
1843 found.head = "\\" + key;
1844 found._tokensize = found.head.length();
1845 found._dataEnd = found._tokenstart + found._tokensize;
1846 found._dataStart = found._dataEnd;
1849 int params = found._tokenstart + key.length() + 1;
1850 if (evaluatingOptional) {
1851 if (size_t(found._tokenstart) > optionalEnd) {
1852 evaluatingOptional = false;
1855 found.disabled = true;
1858 int optend = params;
1859 while (interval_.par[optend] == '[') {
1860 // discard optional parameters
1861 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
1863 if (optend > params) {
1864 key += interval_.par.substr(params, optend-params);
1865 evaluatingOptional = true;
1866 optionalEnd = optend;
1868 string token = sub.str(5);
1869 int closings = found.parenthesiscount;
1870 if (found.parenthesiscount == 1) {
1871 found.head = "\\" + key + "{";
1873 else if (found.parenthesiscount > 1) {
1875 found.head = sub.str(0) + "{";
1876 closings = found.parenthesiscount - 1;
1879 found.head = "\\" + key + "{";
1882 found._tokensize = found.head.length();
1883 found._dataStart = found._tokenstart + found.head.length();
1884 if (found.keytype == KeyInfo::doRemove) {
1885 int endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1886 found._dataStart = endpar;
1887 found._tokensize = found._dataStart - found._tokenstart;
1890 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1891 found._dataStart += 15;
1895 endpos = found._dataStart - 1;
1897 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1898 if (found.keytype == KeyInfo::isList) {
1899 // Check if it really is list env
1900 static regex const listre("^([a-z]+)$");
1902 if (!regex_match(token, sub2, listre)) {
1903 // Change the key of this entry. It is not in a list/item environment
1904 found.keytype = KeyInfo::endArguments;
1907 if (found.keytype == KeyInfo::noMain) {
1908 evaluatingCode = true;
1910 codeStart = found._dataStart;
1912 else if (evaluatingCode) {
1913 if (size_t(found._dataStart) > codeEnd)
1914 evaluatingCode = false;
1915 else if (found.keytype == KeyInfo::isMain) {
1916 // Disable this key, treate it as standard
1917 found.keytype = KeyInfo::isStandard;
1918 found.disabled = true;
1919 if ((codeEnd == interval_.par.length()) &&
1920 (found._tokenstart == codeStart)) {
1921 // trickery, because the code inset starts
1922 // with \selectlanguage ...
1924 if (entries_.size() > 1) {
1925 entries_[entries_.size()-1]._dataEnd = codeEnd;
1930 if ((endpos == interval_.par.length()) &&
1931 (found.keytype == KeyInfo::doRemove)) {
1932 // Missing closing => error in latex-input?
1933 // therefore do not delete remaining data
1934 found._dataStart -= 1;
1935 found._dataEnd = found._dataStart;
1938 found._dataEnd = endpos;
1940 if (isPatternString) {
1941 keys[key].used = true;
1944 entries_.push_back(found);
1948 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1950 stringstream s(keysstring);
1952 const char delim = '|';
1953 while (getline(s, key, delim)) {
1954 KeyInfo keyII(keyI);
1955 if (isPatternString) {
1958 else if ( !keys[key].used)
1959 keyII.disabled = true;
1964 void LatexInfo::buildKeys(bool isPatternString)
1967 static bool keysBuilt = false;
1968 if (keysBuilt && !isPatternString) return;
1970 // Known standard keys with 1 parameter.
1971 // Split is done, if not at start of region
1972 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1973 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1974 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1975 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1976 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1977 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
1979 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
1980 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1981 makeKey("section*|subsection*|subsubsection*|paragraph*",
1982 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1983 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1984 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
1986 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
1988 // Split is done, if not at start of region
1989 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
1990 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
1992 // Split is done always.
1993 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
1997 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1998 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1999 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2000 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2002 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2003 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2004 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2006 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2007 // Custom space/skip, remove the content (== length value)
2008 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2009 // Found in fr/UserGuide.lyx
2010 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2012 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2013 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2014 // Known macros to remove (including their parameter)
2016 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2017 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2018 // handle like standard keys with 1 parameter.
2019 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2021 // Ignore deleted text
2022 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2023 // but preserve added text
2024 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2026 // Macros to remove, but let the parameter survive
2028 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2030 // Remove language spec from content of these insets
2031 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2033 // Same effect as previous, parameter will survive (because there is no one anyway)
2035 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2036 // Remove table decorations
2037 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2038 // Discard shape-header.
2039 // For footnote or shortcut too, because of lang settings
2040 // and wrong handling if used 'KeyInfo::noMain'
2041 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2042 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2043 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2044 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2045 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2046 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2047 // like ('tiny{}' or '\tiny ' ... )
2048 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2050 // Survives, like known character
2051 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2052 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2054 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2055 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2056 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2058 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2059 // Remove RTL/LTR marker
2060 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2061 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2062 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2063 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2064 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2065 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2066 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2067 if (isPatternString) {
2068 // Allow the first searched string to rebuild the keys too
2072 // no need to rebuild again
2078 * Keep the list of actual opened parentheses actual
2079 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2081 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2084 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2088 if (c == '\\') skip = 1;
2089 else if (c == '{') {
2092 else if (c == '}') {
2093 handleCloseP(i, closingAllowed);
2099 string Intervall::show(int lastpos)
2101 int idx = 0; /* int intervalls */
2104 for (idx = 0; idx <= ignoreidx; idx++) {
2105 while (i < lastpos) {
2107 if (i <= borders[idx].low) {
2108 if (borders[idx].low > lastpos)
2109 printsize = lastpos - i;
2111 printsize = borders[idx].low - i;
2112 s += par.substr(i, printsize);
2114 if (i >= borders[idx].low)
2115 i = borders[idx].upper;
2118 i = borders[idx].upper;
2124 s += par.substr(i, lastpos-i);
2130 void Intervall::output(ostringstream &os, int lastpos)
2132 // get number of chars to output
2133 int idx = 0; /* int intervalls */
2136 string startTitle = titleValue;
2137 for (idx = 0; idx <= ignoreidx; idx++) {
2139 if (i <= borders[idx].low) {
2141 if (borders[idx].low > lastpos)
2142 printsize = lastpos - i;
2144 printsize = borders[idx].low - i;
2145 if (printsize > 0) {
2146 os << startTitle << par.substr(i, printsize);
2148 printed += printsize;
2151 handleParentheses(i, false);
2152 if (i >= borders[idx].low)
2153 i = borders[idx].upper;
2156 i = borders[idx].upper;
2163 os << startTitle << par.substr(i, lastpos-i);
2164 printed += lastpos-i;
2166 handleParentheses(lastpos, false);
2168 if (keys["foreignlanguage"].disabled)
2169 startindex = actualdeptindex-langcount;
2171 startindex = actualdeptindex;
2172 for (int i = startindex; i > 0; --i) {
2175 if (hasTitle && (printed > 0))
2177 if (! isPatternString_)
2179 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2182 void LatexInfo::processRegion(int start, int region_end)
2184 while (start < region_end) { /* Let {[} and {]} survive */
2185 int cnt = interval_.isOpeningPar(start);
2187 // Closing is allowed past the region
2188 int closing = interval_.findclosing(start+1, interval_.par.length());
2189 interval_.addIntervall(start, start+1);
2190 interval_.addIntervall(closing, closing+1);
2194 start = interval_.nextNotIgnored(start+1);
2198 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2200 if (actual.parenthesiscount == 0) {
2201 // "{\tiny{} ...}" ==> "{{} ...}"
2202 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2205 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2206 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2207 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2211 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2214 switch (actual.keytype)
2216 case KeyInfo::isTitle: {
2218 nextKeyIdx = getNextKey();
2221 case KeyInfo::cleanToStart: {
2222 actual._dataEnd = actual._dataStart;
2223 nextKeyIdx = getNextKey();
2224 // Search for end of arguments
2225 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2227 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2228 entries_[i].disabled = true;
2230 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2232 while (interval_.par[actual._dataEnd] == ' ')
2234 interval_.addIntervall(0, actual._dataEnd+1);
2235 interval_.actualdeptindex = 0;
2236 interval_.depts[0] = actual._dataEnd+1;
2237 interval_.closes[0] = -1;
2240 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2241 if (actual.disabled)
2242 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2244 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2247 case KeyInfo::isChar: {
2248 nextKeyIdx = getNextKey();
2251 case KeyInfo::isSize: {
2252 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2253 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2254 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2255 nextKeyIdx = getNextKey();
2257 // Here _dataStart points to '{', so correct it
2258 actual._dataStart += 1;
2259 actual._tokensize += 1;
2260 actual.parenthesiscount = 1;
2261 if (interval_.par[actual._dataStart] == '}') {
2262 // Determine the end if used like '{\tiny{}...}'
2263 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2264 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2267 // Determine the end if used like '\tiny{...}'
2268 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2270 // Split on this key if not at start
2271 int start = interval_.nextNotIgnored(previousStart);
2272 if (start < actual._tokenstart) {
2273 interval_.output(os, actual._tokenstart);
2274 interval_.addIntervall(start, actual._tokenstart);
2276 // discard entry if at end of actual
2277 nextKeyIdx = process(os, actual);
2281 case KeyInfo::endArguments:
2282 // Remove trailing '{}' too
2283 actual._dataStart += 1;
2284 actual._dataEnd += 1;
2285 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2286 nextKeyIdx = getNextKey();
2288 case KeyInfo::noMain:
2290 case KeyInfo::isStandard: {
2291 if (actual.disabled) {
2293 processRegion(actual._dataStart, actual._dataStart+1);
2294 nextKeyIdx = getNextKey();
2296 // Split on this key if not at datastart of calling entry
2297 int start = interval_.nextNotIgnored(previousStart);
2298 if (start < actual._tokenstart) {
2299 interval_.output(os, actual._tokenstart);
2300 interval_.addIntervall(start, actual._tokenstart);
2302 // discard entry if at end of actual
2303 nextKeyIdx = process(os, actual);
2307 case KeyInfo::removeWithArg: {
2308 nextKeyIdx = getNextKey();
2309 // Search for end of arguments
2310 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2312 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2313 entries_[i].disabled = true;
2315 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2317 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2320 case KeyInfo::doRemove: {
2321 // Remove the key with all parameters and following spaces
2323 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2324 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2327 // Remove also enclosing parentheses [] and {}
2330 while (actual._tokenstart > numpars) {
2331 if (pos+numpars >= interval_.par.size())
2333 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2335 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2341 if (interval_.par[pos+numpars] == ' ')
2345 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2346 nextKeyIdx = getNextKey();
2349 case KeyInfo::isList: {
2350 // Discard space before _tokenstart
2352 for (count = 0; count < actual._tokenstart; count++) {
2353 if (interval_.par[actual._tokenstart-count-1] != ' ')
2356 nextKeyIdx = getNextKey();
2357 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2359 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2361 // How else can we catch this one?
2362 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2363 entries_[i].disabled = true;
2365 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2367 else if (nextKeyIdx > 0) {
2368 // Ignore any lang entries inside data region
2369 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2370 if (entries_[i].keytype == KeyInfo::isMain)
2371 entries_[i].disabled = true;
2374 if (actual.disabled) {
2375 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2378 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2380 if (interval_.par[actual._dataEnd+1] == '[') {
2381 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2382 if ((interval_.par[actual._dataEnd+2] == '{') &&
2383 (interval_.par[posdown-1] == '}')) {
2384 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2385 interval_.addIntervall(posdown-1, posdown+1);
2388 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2389 interval_.addIntervall(posdown, posdown+1);
2391 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2392 if (blk > posdown) {
2393 // Discard at most 1 space after empty item
2395 for (count = 0; count < 1; count++) {
2396 if (interval_.par[blk+count] != ' ')
2400 interval_.addIntervall(blk, blk+count);
2405 case KeyInfo::isSectioning: {
2406 // Discard spaces before _tokenstart
2408 int val = actual._tokenstart;
2409 for (count = 0; count < actual._tokenstart;) {
2410 val = interval_.previousNotIgnored(val-1);
2411 if (val < 0 || interval_.par[val] != ' ')
2414 count = actual._tokenstart - val;
2417 if (actual.disabled) {
2418 removeHead(actual, count);
2419 nextKeyIdx = getNextKey();
2421 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2422 nextKeyIdx = process(os, actual);
2426 case KeyInfo::isMath: {
2427 // Same as regex, use the content unchanged
2428 nextKeyIdx = getNextKey();
2431 case KeyInfo::isRegex: {
2432 // DO NOT SPLIT ON REGEX
2434 nextKeyIdx = getNextKey();
2437 case KeyInfo::isIgnored: {
2438 // Treat like a character for now
2439 nextKeyIdx = getNextKey();
2442 case KeyInfo::isMain: {
2443 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2444 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2445 if (actual._tokenstart > 0) {
2446 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2447 if ((prev >= 0) && interval_.par[prev] == '%')
2448 interval_.addIntervall(prev, prev+1);
2450 if (actual.disabled) {
2452 interval_.langcount++;
2453 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2454 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2455 // Discard also the space before math-equation
2456 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2458 nextKeyIdx = getNextKey();
2459 // interval.resetOpenedP(actual._dataStart-1);
2462 if (actual._tokenstart < 26) {
2463 // for the first (and maybe dummy) language
2464 interval_.setForDefaultLang(actual);
2466 interval_.resetOpenedP(actual._dataStart-1);
2470 case KeyInfo::invalid:
2471 // This cannot happen, already handled
2474 // LYXERR(Debug::INFO, "Unhandled keytype");
2475 nextKeyIdx = getNextKey();
2482 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2484 int end = interval_.nextNotIgnored(actual._dataEnd);
2485 int oldStart = actual._dataStart;
2486 int nextKeyIdx = getNextKey();
2488 if ((nextKeyIdx < 0) ||
2489 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2490 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2491 if (oldStart <= end) {
2492 processRegion(oldStart, end);
2497 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2499 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2500 (void) dispatch(os, actual._dataStart, nextKey);
2501 end = nextKey._tokenstart;
2504 processRegion(oldStart, nextKey._tokenstart);
2505 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2507 oldStart = nextKey._dataEnd+1;
2509 // now nextKey is either invalid or is outside of actual._dataEnd
2510 // output the remaining and discard myself
2511 if (oldStart <= end) {
2512 processRegion(oldStart, end);
2514 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2516 // This is the normal case.
2517 // But if using the firstlanguage, the closing may be missing
2519 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2521 if (actual._dataEnd < end)
2522 output_end = interval_.nextNotIgnored(actual._dataEnd);
2524 output_end = interval_.nextNotIgnored(end);
2525 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2526 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2528 // Remove possible empty data
2529 int dstart = interval_.nextNotIgnored(actual._dataStart);
2530 while (interval_.isOpeningPar(dstart) == 1) {
2531 interval_.addIntervall(dstart, dstart+1);
2532 int dend = interval_.findclosing(dstart+1, output_end);
2533 interval_.addIntervall(dend, dend+1);
2534 dstart = interval_.nextNotIgnored(dstart+1);
2536 if (dstart < output_end)
2537 interval_.output(os, output_end);
2538 interval_.addIntervall(actual._tokenstart, end);
2542 string splitOnKnownMacros(string par, bool isPatternString)
2545 LatexInfo li(par, isPatternString);
2546 // LYXERR(Debug::INFO, "Berfore split: " << par);
2547 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2549 DummyKey._tokensize = 0;
2550 DummyKey._dataStart = 0;
2551 DummyKey._dataEnd = par.length();
2552 DummyKey.disabled = true;
2553 int firstkeyIdx = li.getFirstKey();
2555 if (firstkeyIdx >= 0) {
2556 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2557 DummyKey._tokenstart = firstKey._tokenstart;
2559 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2560 // Use dummy firstKey
2561 firstKey = DummyKey;
2562 (void) li.setNextKey(firstkeyIdx);
2565 if (par.substr(firstKey._dataStart, 2) == "% ")
2566 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2568 nextkeyIdx = li.process(os, firstKey);
2569 while (nextkeyIdx >= 0) {
2570 // Check for a possible gap between the last
2571 // entry and this one
2572 int datastart = li.nextNotIgnored(firstKey._dataStart);
2573 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2574 if ((nextKey._tokenstart > datastart)) {
2576 firstKey._dataStart = datastart;
2577 firstKey._dataEnd = par.length();
2578 (void) li.setNextKey(nextkeyIdx);
2579 // Fake the last opened parenthesis
2580 li.setForDefaultLang(firstKey);
2581 nextkeyIdx = li.process(os, firstKey);
2584 if (nextKey.keytype != KeyInfo::isMain) {
2585 firstKey._dataStart = datastart;
2586 firstKey._dataEnd = nextKey._dataEnd+1;
2587 (void) li.setNextKey(nextkeyIdx);
2588 li.setForDefaultLang(firstKey);
2589 nextkeyIdx = li.process(os, firstKey);
2592 nextkeyIdx = li.process(os, nextKey);
2596 // Handle the remaining
2597 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2598 firstKey._dataEnd = par.length();
2600 if ((firstKey._dataStart < firstKey._dataEnd) &&
2601 (par[firstKey._dataStart] != '}')) {
2602 li.setForDefaultLang(firstKey);
2603 (void) li.process(os, firstKey);
2607 // return string definitelly impossible to match
2608 s = "\\foreignlanguage{ignore}{ }";
2612 s = par; /* no known macros found */
2613 // LYXERR(Debug::INFO, "After split: " << s);
2618 * Try to unify the language specs in the latexified text.
2619 * Resulting modified string is set to "", if
2620 * the searched tex does not contain all the features in the search pattern
2622 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2624 static Features regex_f;
2625 static int missed = 0;
2626 static bool regex_with_format = false;
2628 int parlen = par.length();
2630 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2633 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2634 // Happens to be there in case of description or labeling environment
2639 // Split the latex input into pieces which
2640 // can be digested by our search engine
2641 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2642 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2643 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2646 result = par.substr(0, parlen);
2647 if (isPatternString) {
2650 regex_f = identifyFeatures(result);
2651 string features = "";
2652 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2653 string a = it->first;
2654 regex_with_format = true;
2655 features += " " + a;
2656 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2658 LYXERR(Debug::FIND, "Identified Features" << features);
2661 } else if (regex_with_format) {
2662 Features info = identifyFeatures(result);
2663 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2664 string a = it->first;
2665 bool b = it->second;
2666 if (b && ! info[a]) {
2668 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2674 // LYXERR(Debug::INFO, "No regex formats");
2680 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2681 static int identifyClosing(string & t)
2683 int open_braces = 0;
2685 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2686 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2688 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]" REGEX_EOS, "$1"))
2690 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2692 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2701 static int num_replaced = 0;
2702 static bool previous_single_replace = true;
2704 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2705 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2707 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2708 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2709 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2710 if (opt.replace_all && previous_single_replace) {
2711 previous_single_replace = false;
2714 else if (!opt.replace_all) {
2715 num_replaced = 0; // count number of replaced strings
2716 previous_single_replace = true;
2718 // When using regexp, braces are hacked already by escape_for_regex()
2719 par_as_string = normalize(ds, !use_regexp);
2721 close_wildcards = 0;
2723 size_t lead_size = 0;
2724 // correct the language settings
2725 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2726 if (opt.ignoreformat) {
2728 // if par_as_string_nolead were emty,
2729 // the following call to findAux will always *find* the string
2730 // in the checked data, and thus always using the slow
2731 // examining of the current text part.
2732 par_as_string_nolead = par_as_string;
2735 lead_size = identifyLeading(par_as_string);
2736 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2737 lead_as_string = par_as_string.substr(0, lead_size);
2738 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2742 open_braces = identifyClosing(par_as_string);
2743 identifyClosing(par_as_string_nolead);
2744 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2745 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2747 string lead_as_regexp;
2748 if (lead_size > 0) {
2749 // @todo No need to search for \regexp{} insets in leading material
2750 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2751 par_as_string = par_as_string_nolead;
2752 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2753 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2755 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2756 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2757 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2759 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2760 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2761 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2762 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2763 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2764 || regex_replace(par_as_string, par_as_string,
2765 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2766 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2767 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2771 if (!opt.ignoreformat) {
2772 // Remove extra '\}' at end if not part of \{\.\}
2773 size_t lng = par_as_string.size();
2775 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2777 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2786 if (lng < par_as_string.size())
2787 par_as_string = par_as_string.substr(0,lng);
2790 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2791 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2792 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2793 // replace '[^...]' with '[^...\}\{\\]'
2794 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2795 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2796 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2798 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2801 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2802 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2803 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2804 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2806 // If entered regexp must match at begin of searched string buffer
2807 // Kornel: Added parentheses to use $1 for size of the leading string
2811 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2812 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2813 // so the convert has no effect in that case
2814 for (int i = 8; i > 0; --i) {
2815 string orig = "\\\\" + std::to_string(i);
2816 string dest = "\\" + std::to_string(i+1);
2817 while (regex_replace(par_as_string, par_as_string, orig, dest));
2819 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2820 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2822 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2823 regexp = lyx::regex(regexp_str);
2825 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2826 regexp2 = lyx::regex(regexp2_str);
2831 // Count number of characters in string
2835 // \\[a-zA-Z]+ ==> 1
2836 static int computeSize(string s, int len)
2842 for (int i = 0; i < len; i += skip, count++) {
2845 if (isalpha(s[i+1])) {
2846 for (int j = 2; i+j < len; j++) {
2847 if (! isalpha(s[i+j])) {
2850 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2852 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2860 else if (s[i] == '{') {
2866 else if (s[i] == '-') {
2867 if (s[i+1] == '-') {
2883 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2888 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2891 docstring docstr = stringifyFromForSearch(opt, cur, len);
2892 string str = normalize(docstr, true);
2893 if (!opt.ignoreformat) {
2894 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2897 mres.match_len = -1;
2900 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2901 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2904 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2905 regex const *p_regexp;
2906 regex_constants::match_flag_type flags;
2908 flags = regex_constants::match_continuous;
2911 flags = regex_constants::match_default;
2912 p_regexp = ®exp2;
2914 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2915 if (re_it == sregex_iterator())
2917 match_results<string::const_iterator> const & m = *re_it;
2919 if (0) { // Kornel Benko: DO NOT CHECKK
2920 // Check braces on the segment that matched the entire regexp expression,
2921 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2922 if (!braces_match(m[0].first, m[0].second, open_braces))
2926 // Check braces on segments that matched all (.*?) subexpressions,
2927 // except the last "padding" one inserted by lyx.
2928 for (size_t i = 1; i < m.size() - 1; ++i)
2929 if (!braces_match(m[i].first, m[i].second, open_braces))
2932 // Exclude from the returned match length any length
2933 // due to close wildcards added at end of regexp
2934 // and also the length of the leading (e.g. '\emph{}')
2936 // Whole found string, including the leading: m[0].second - m[0].first
2937 // Size of the leading string: m[1].second - m[1].first
2938 int leadingsize = 0;
2940 leadingsize = m[1].second - m[1].first;
2942 for (size_t i = 0; i < m.size(); i++) {
2943 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2945 if (close_wildcards == 0)
2946 result = m[0].second - m[0].first;
2949 result = m[m.size() - close_wildcards].first - m[0].first;
2951 size_t pos = m.position(size_t(0));
2952 // Ignore last closing characters
2953 while (result > 0) {
2954 if (str[pos+result-1] == '}')
2959 if (result > leadingsize)
2960 result -= leadingsize;
2963 mres.match_len = computeSize(str.substr(pos+leadingsize,result), result);
2964 mres.match2end = str.size() - pos - leadingsize;
2965 mres.pos = pos+leadingsize;
2969 // else !use_regexp: but all code paths above return
2970 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2971 << par_as_string << "', str='" << str << "'");
2972 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2973 << lead_as_string << "', par_as_string_nolead='"
2974 << par_as_string_nolead << "'");
2977 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2978 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2979 if (str.substr(0, par_as_string.size()) == par_as_string) {
2980 mres.match_len = par_as_string.size();
2981 mres.match2end = str.size();
2986 // Start the search _after_ the leading part
2987 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
2988 if (pos != string::npos) {
2989 mres.match_len = par_as_string.size();
2990 mres.match2end = str.size() - pos;
2999 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3001 MatchResult mres = findAux(cur, len, at_begin);
3002 int res = mres.match_len;
3004 "res=" << res << ", at_begin=" << at_begin
3005 << ", matchword=" << opt.matchword
3006 << ", inTexted=" << cur.inTexted());
3007 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3009 if ((len > 0) && (res < len)) {
3013 Paragraph const & par = cur.paragraph();
3014 bool ws_left = (cur.pos() > 0)
3015 ? par.isWordSeparator(cur.pos() - 1)
3021 ws_right = (cur.pos() + len < par.size())
3022 ? par.isWordSeparator(cur.pos() + len)
3026 "cur.pos()=" << cur.pos() << ", res=" << res
3027 << ", separ: " << ws_left << ", " << ws_right
3030 if (ws_left && ws_right) {
3031 // Check for word separators inside the found 'word'
3032 for (int i = 0; i < len; i++) {
3033 if (par.isWordSeparator(cur.pos() + i)) {
3045 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
3048 if (! opt.casesensitive)
3049 t = lyx::to_utf8(lowercase(s));
3051 t = lyx::to_utf8(s);
3052 // Remove \n at begin
3053 while (!t.empty() && t[0] == '\n')
3056 while (!t.empty() && t[t.size() - 1] == '\n')
3057 t = t.substr(0, t.size() - 1);
3059 // Handle all other '\n'
3060 while ((pos = t.find("\n")) != string::npos) {
3061 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3063 if (isAlnumASCII(t[pos+1])) {
3064 t.replace(pos-2, 3, " ");
3067 t.replace(pos-2, 3, "");
3070 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3071 // '\n' adjacent to non-alpha-numerics, discard
3072 t.replace(pos, 1, "");
3075 // Replace all other \n with spaces
3076 t.replace(pos, 1, " ");
3079 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3080 // Kornel: Added textsl, textsf, textit, texttt and noun
3081 // + allow to seach for colored text too
3082 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3083 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3084 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3085 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3086 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3088 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3089 // FIXME - check what preceeds the brace
3091 if (opt.ignoreformat)
3092 while (regex_replace(t, t, "\\{", "_x_<")
3093 || regex_replace(t, t, "\\}", "_x_>"))
3094 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3096 while (regex_replace(t, t, "\\\\\\{", "_x_<")
3097 || regex_replace(t, t, "\\\\\\}", "_x_>"))
3098 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3105 docstring stringifyFromCursor(DocIterator const & cur, int len)
3107 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3108 if (cur.inTexted()) {
3109 Paragraph const & par = cur.paragraph();
3110 // TODO what about searching beyond/across paragraph breaks ?
3111 // TODO Try adding a AS_STR_INSERTS as last arg
3112 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3113 int(par.size()) : cur.pos() + len;
3114 // OutputParams runparams(&cur.buffer()->params().encoding());
3115 OutputParams runparams(encodings.fromLyXName("utf8"));
3116 runparams.nice = true;
3117 runparams.flavor = OutputParams::XETEX;
3118 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3119 // No side effect of file copying and image conversion
3120 runparams.dryrun = true;
3121 runparams.for_search = true;
3122 LYXERR(Debug::FIND, "Stringifying with cur: "
3123 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3124 return par.asString(cur.pos(), end,
3125 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3127 } else if (cur.inMathed()) {
3128 CursorSlice cs = cur.top();
3129 MathData md = cs.cell();
3130 MathData::const_iterator it_end =
3131 (( len == -1 || cs.pos() + len > int(md.size()))
3133 : md.begin() + cs.pos() + len );
3135 for (MathData::const_iterator it = md.begin() + cs.pos();
3138 docstring s = asString(md2);
3139 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3142 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3147 /** Computes the LaTeX export of buf starting from cur and ending len positions
3148 * after cur, if len is positive, or at the paragraph or innermost inset end
3151 docstring latexifyFromCursor(DocIterator const & cur, int len)
3153 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3154 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3155 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3156 Buffer const & buf = *cur.buffer();
3158 odocstringstream ods;
3160 //OutputParams runparams(&buf.params().encoding());
3161 OutputParams runparams(encodings.fromLyXName("utf8"));
3162 runparams.nice = false;
3163 runparams.flavor = OutputParams::XETEX;
3164 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3165 // No side effect of file copying and image conversion
3166 runparams.dryrun = true;
3167 runparams.for_search = true;
3169 if (cur.inTexted()) {
3170 // @TODO what about searching beyond/across paragraph breaks ?
3171 pos_type endpos = cur.paragraph().size();
3172 if (len != -1 && endpos > cur.pos() + len)
3173 endpos = cur.pos() + len;
3174 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3175 string(), cur.pos(), endpos);
3176 string s = lyx::to_utf8(ods.str());
3177 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3178 return(lyx::from_utf8(s));
3179 } else if (cur.inMathed()) {
3180 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3181 for (int s = cur.depth() - 1; s >= 0; --s) {
3182 CursorSlice const & cs = cur[s];
3183 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3185 cs.asInsetMath()->asHullInset()->header_write(ws);
3190 CursorSlice const & cs = cur.top();
3191 MathData md = cs.cell();
3192 MathData::const_iterator it_end =
3193 ((len == -1 || cs.pos() + len > int(md.size()))
3195 : md.begin() + cs.pos() + len);
3197 for (MathData::const_iterator it = md.begin() + cs.pos();
3201 ods << asString(md2);
3202 // Retrieve the math environment type, and add '$' or '$]'
3203 // or others (\end{equation}) accordingly
3204 for (int s = cur.depth() - 1; s >= 0; --s) {
3205 CursorSlice const & cs2 = cur[s];
3206 InsetMath * inset = cs2.asInsetMath();
3207 if (inset && inset->asHullInset()) {
3209 inset->asHullInset()->footer_write(ws);
3213 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3215 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3221 /** Finalize an advanced find operation, advancing the cursor to the innermost
3222 ** position that matches, plus computing the length of the matching text to
3225 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
3227 // Search the foremost position that matches (avoids find of entire math
3228 // inset when match at start of it)
3230 DocIterator old_cur(cur.buffer());
3232 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3236 } while (cur && cur.depth() > d && match(cur).match_len > 0);
3238 int max_match = match(cur).match_len; /* match valid only if not searching whole words */
3239 if (max_match <= 0) return 0;
3240 LYXERR(Debug::FIND, "Ok");
3242 // Compute the match length
3244 if (cur.pos() + len > cur.lastpos())
3246 if (match.opt.matchword) {
3247 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3248 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3250 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3252 // Length of matched text (different from len param)
3253 int old_match = match(cur, len).match_len;
3257 // Greedy behaviour while matching regexps
3258 while ((new_match = match(cur, len + 1).match_len) > old_match) {
3260 old_match = new_match;
3261 LYXERR(Debug::FIND, "verifying match with len = " << len);
3268 int maxl = cur.lastpos() - cur.pos();
3269 // Greedy behaviour while matching regexps
3270 while (maxl > minl) {
3271 int actual_match = match(cur, len).match_len;
3272 if (actual_match >= max_match) {
3273 // actual_match > max_match _can_ happen,
3274 // if the search area splits
3275 // some following word so that the regex
3276 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3278 // This means, the len value is too big
3280 len = (int)((maxl + minl)/2);
3283 // (actual_match < max_match)
3285 len = (int)((maxl + minl)/2);
3289 // Search for real start of matched characters
3294 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3295 if (cur.depth() < old_cur.depth()) {
3297 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3300 if (cur.pos() != old_cur.pos()) {
3301 // OK, forwarded 1 pos in actual inset
3302 actual_match = match(cur, len-1).match_len;
3303 if (actual_match == max_match) {
3304 // Ha, got it! The shorter selection has the same match length
3309 // OK, the shorter selection matches less chars, revert to previous value
3315 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3316 actual_match = match(cur, len).match_len;
3317 if (actual_match == max_match)
3327 int findForwardAdv(DocIterator & cur, MatchStringAdv const & match)
3331 while (!theApp()->longOperationCancelled() && cur) {
3332 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3333 MatchResult mres = match(cur, -1, false);
3334 int match_len = mres.match_len;
3335 LYXERR(Debug::FIND, "match_len: " << match_len);
3336 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3337 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3340 if (match_len > 0) {
3341 // Try to find the begin of searched string
3342 int increment = mres.pos/2;
3343 while (mres.pos > 5 && (increment > 5)) {
3344 DocIterator old_cur = cur;
3345 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3347 if (! cur || (cur.pit() > old_cur.pit())) {
3348 // Are we outside of the paragraph?
3349 // This can happen if moving past some UTF8-encoded chars
3354 MatchResult mres2 = match(cur, -1, false);
3355 if ((mres2.match2end < mres.match2end) ||
3356 (mres2.match_len < mres.match_len)) {
3363 if (increment > mres.pos/2)
3364 increment = mres.pos/2;
3368 int match_len_zero_count = 0;
3369 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3371 int remaining_len = match(cur, -1, false).match_len;
3372 if (remaining_len <= 0) {
3373 // Apparently the searched string is not in the remaining part
3380 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3381 int match_len3 = match(cur, 1).match_len;
3384 int match_len2 = match(cur).match_len;
3385 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3386 if (match_len2 > 0) {
3387 // Sometimes in finalize we understand it wasn't a match
3388 // and we need to continue the outest loop
3389 int len = findAdvFinalize(cur, match);
3394 if (match_len2 >= 0) {
3395 if (match_len2 == 0)
3396 match_len_zero_count++;
3398 match_len_zero_count = 0;
3401 if (++match_len_zero_count > 3) {
3402 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3410 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3411 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3414 // This should exit nested insets, if any, or otherwise undefine the currsor.
3415 cur.pos() = cur.lastpos();
3416 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3424 /// Find the most backward consecutive match within same paragraph while searching backwards.
3425 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3427 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3428 DocIterator tmp_cur = cur;
3429 int len = findAdvFinalize(tmp_cur, match);
3430 Inset & inset = cur.inset();
3431 for (; cur != cur_begin; cur.backwardPos()) {
3432 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3433 DocIterator new_cur = cur;
3434 new_cur.backwardPos();
3435 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3437 int new_len = findAdvFinalize(new_cur, match);
3442 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3448 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3452 // Backup of original position
3453 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3454 if (cur == cur_begin)
3457 DocIterator cur_orig(cur);
3458 bool pit_changed = false;
3461 bool found_match = (match(cur, -1, false).match_len > 0);
3465 cur.pos() = cur.lastpos();
3467 cur.pos() = cur_orig.pos();
3468 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3469 DocIterator cur_prev_iter;
3471 found_match = (match(cur).match_len > 0);
3472 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3473 << found_match << ", cur: " << cur);
3475 return findMostBackwards(cur, match);
3477 // Stop if begin of document reached
3478 if (cur == cur_begin)
3480 cur_prev_iter = cur;
3484 if (cur == cur_begin)
3491 } while (!theApp()->longOperationCancelled());
3499 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3500 DocIterator const & cur, int len)
3502 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3504 if (!opt.ignoreformat)
3505 return latexifyFromCursor(cur, len);
3507 return stringifyFromCursor(cur, len);
3511 FindAndReplaceOptions::FindAndReplaceOptions(
3512 docstring const & _find_buf_name, bool _casesensitive,
3513 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3514 docstring const & _repl_buf_name, bool _keep_case,
3515 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3516 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3517 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3518 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3526 /** Check if 'len' letters following cursor are all non-lowercase */
3527 static bool allNonLowercase(Cursor const & cur, int len)
3529 pos_type beg_pos = cur.selectionBegin().pos();
3530 pos_type end_pos = cur.selectionBegin().pos() + len;
3531 if (len > cur.lastpos() + 1 - beg_pos) {
3532 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3533 len = cur.lastpos() + 1 - beg_pos;
3534 end_pos = beg_pos + len;
3536 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3537 if (isLowerCase(cur.paragraph().getChar(pos)))
3543 /** Check if first letter is upper case and second one is lower case */
3544 static bool firstUppercase(Cursor const & cur)
3547 pos_type pos = cur.selectionBegin().pos();
3548 if (pos >= cur.lastpos() - 1) {
3549 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3552 ch1 = cur.paragraph().getChar(pos);
3553 ch2 = cur.paragraph().getChar(pos + 1);
3554 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3555 LYXERR(Debug::FIND, "firstUppercase(): "
3556 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3557 << ch2 << "(" << char(ch2) << ")"
3558 << ", result=" << result << ", cur=" << cur);
3563 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3565 ** \fixme What to do with possible further paragraphs in replace buffer ?
3567 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3569 ParagraphList::iterator pit = buffer.paragraphs().begin();
3570 LASSERT(!pit->empty(), /**/);
3571 pos_type right = pos_type(1);
3572 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3573 right = pit->size();
3574 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3580 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3582 Cursor & cur = bv->cursor();
3583 if (opt.repl_buf_name.empty()
3584 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3585 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3588 DocIterator sel_beg = cur.selectionBegin();
3589 DocIterator sel_end = cur.selectionEnd();
3590 if (&sel_beg.inset() != &sel_end.inset()
3591 || sel_beg.pit() != sel_end.pit()
3592 || sel_beg.idx() != sel_end.idx())
3594 int sel_len = sel_end.pos() - sel_beg.pos();
3595 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3596 << ", sel_len: " << sel_len << endl);
3599 LASSERT(sel_len > 0, return 0);
3601 if (!matchAdv(sel_beg, sel_len).match_len)
3604 // Build a copy of the replace buffer, adapted to the KeepCase option
3605 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3607 repl_buffer_orig.write(oss);
3608 string lyx = oss.str();
3609 Buffer repl_buffer("", false);
3610 repl_buffer.setUnnamed(true);
3611 LASSERT(repl_buffer.readString(lyx), return 0);
3612 if (opt.keep_case && sel_len >= 2) {
3613 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3614 if (cur.inTexted()) {
3615 if (firstUppercase(cur))
3616 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3617 else if (allNonLowercase(cur, sel_len))
3618 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3621 cap::cutSelection(cur, false);
3622 if (cur.inTexted()) {
3623 repl_buffer.changeLanguage(
3624 repl_buffer.language(),
3625 cur.getFont().language());
3626 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3627 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3628 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3629 repl_buffer.params().documentClassPtr(),
3630 bv->buffer().errorList("Paste"));
3631 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3632 sel_len = repl_buffer.paragraphs().begin()->size();
3633 } else if (cur.inMathed()) {
3634 odocstringstream ods;
3636 // OutputParams runparams(&repl_buffer.params().encoding());
3637 OutputParams runparams(encodings.fromLyXName("utf8"));
3638 runparams.nice = false;
3639 runparams.flavor = OutputParams::XETEX;
3640 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3641 runparams.dryrun = true;
3642 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3643 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3644 docstring repl_latex = ods.str();
3645 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3647 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3648 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3649 repl_latex = from_utf8(s);
3650 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3651 MathData ar(cur.buffer());
3652 asArray(repl_latex, ar, Parse::NORMAL);
3654 sel_len = ar.size();
3655 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3657 if (cur.pos() >= sel_len)
3658 cur.pos() -= sel_len;
3661 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3662 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3663 bv->processUpdateFlags(Update::Force);
3668 /// Perform a FindAdv operation.
3669 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3674 // e.g., when invoking word-findadv from mini-buffer wither with
3675 // wrong options syntax or before ever opening advanced F&R pane
3676 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3680 MatchStringAdv matchAdv(bv->buffer(), opt);
3681 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3683 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3684 num_replaced += findAdvReplace(bv, opt, matchAdv);
3687 match_len = findForwardAdv(cur, matchAdv);
3689 match_len = findBackwardsAdv(cur, matchAdv);
3691 // This may only be raised by lyx::regex()
3692 bv->message(_("Invalid regular expression!"));
3696 if (match_len == 0) {
3697 if (num_replaced > 0) {
3698 switch (num_replaced)
3701 bv->message(_("One match has been replaced."));
3704 bv->message(_("Two matches have been replaced."));
3707 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
3713 bv->message(_("Match not found."));
3718 if (num_replaced > 0)
3719 bv->message(_("Match has been replaced."));
3721 bv->message(_("Match found."));
3723 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3724 bv->putSelectionAt(cur, match_len, !opt.forward);
3730 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3732 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3733 << opt.casesensitive << ' '
3734 << opt.matchword << ' '
3735 << opt.forward << ' '
3736 << opt.expandmacros << ' '
3737 << opt.ignoreformat << ' '
3738 << opt.replace_all << ' '
3739 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3740 << opt.keep_case << ' '
3741 << int(opt.scope) << ' '
3744 LYXERR(Debug::FIND, "built: " << os.str());
3750 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3752 LYXERR(Debug::FIND, "parsing");
3756 while (line != "EOSS") {
3760 if (is.eof()) // Tolerate malformed request
3764 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3765 opt.find_buf_name = from_utf8(s);
3766 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
3767 is.get(); // Waste space before replace string
3770 while (line != "EOSS") {
3774 if (is.eof()) // Tolerate malformed request
3778 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3779 opt.repl_buf_name = from_utf8(s);
3780 is >> opt.keep_case;
3783 opt.scope = FindAndReplaceOptions::SearchScope(i);
3785 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3787 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3788 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3789 << opt.scope << ' ' << opt.restr);