3 * This file is part of LyX, the document processor.
4 * License details can be found in the file COPYING.
6 * \author Lars Gullik Bjønnes
9 * \author Alfredo Braunstein
10 * \author Tommaso Cucinotta
12 * Full author contact details are available in file CREDITS.
20 #include "buffer_funcs.h"
21 #include "BufferList.h"
22 #include "BufferParams.h"
23 #include "BufferView.h"
26 #include "CutAndPaste.h"
27 #include "FuncRequest.h"
29 #include "output_latex.h"
30 #include "OutputParams.h"
31 #include "Paragraph.h"
32 #include "ParIterator.h"
36 #include "frontends/Application.h"
37 #include "frontends/alert.h"
39 #include "mathed/InsetMath.h"
40 #include "mathed/InsetMathGrid.h"
41 #include "mathed/InsetMathHull.h"
42 #include "mathed/MathData.h"
43 #include "mathed/MathStream.h"
44 #include "mathed/MathSupport.h"
46 #include "support/convert.h"
47 #include "support/debug.h"
48 #include "support/docstream.h"
49 #include "support/FileName.h"
50 #include "support/gettext.h"
51 #include "support/lassert.h"
52 #include "support/lstrings.h"
54 #include "support/regex.h"
58 using namespace lyx::support;
63 // Helper class for deciding what should be ignored
68 : ignoreFamily_(false), ignoreSeries_(false),
69 ignoreShape_(false), ignoreUnderline_(false),
70 ignoreMarkUp_(false), ignoreStrikeOut_(false),
71 ignoreSectioning_(false), ignoreFrontMatter_(false),
72 ignoreColor_(false), ignoreLanguage_(false) {}
74 bool getFamily() { return ignoreFamily_; };
76 bool getSeries() { return ignoreSeries_; };
78 bool getShape() { return ignoreShape_; };
80 bool getUnderline() { return ignoreUnderline_; };
82 bool getMarkUp() { return ignoreMarkUp_; };
84 bool getStrikeOut() { return ignoreStrikeOut_; };
86 bool getSectioning() { return ignoreSectioning_; };
88 bool getFrontMatter() { return ignoreFrontMatter_; };
90 bool getColor() { return ignoreColor_; };
92 bool getLanguage() { return ignoreLanguage_; };
94 void setIgnoreFormat(string type, bool value);
104 bool ignoreUnderline_;
108 bool ignoreStrikeOut_;
110 bool ignoreSectioning_;
112 bool ignoreFrontMatter_;
116 bool ignoreLanguage_;
120 void IgnoreFormats::setIgnoreFormat(string type, bool value)
122 if (type == "color") {
123 ignoreColor_ = value;
125 else if (type == "language") {
126 ignoreLanguage_ = value;
128 else if (type == "sectioning") {
129 ignoreSectioning_ = value;
130 ignoreFrontMatter_ = value;
132 else if (type == "font") {
133 ignoreSeries_ = value;
134 ignoreShape_ = value;
135 ignoreFamily_ = value;
137 else if (type == "series") {
138 ignoreSeries_ = value;
140 else if (type == "shape") {
141 ignoreShape_ = value;
143 else if (type == "family") {
144 ignoreFamily_ = value;
146 else if (type == "markup") {
147 ignoreMarkUp_ = value;
149 else if (type == "underline") {
150 ignoreUnderline_ = value;
152 else if (type == "strike") {
153 ignoreStrikeOut_ = value;
157 // The global variable that can be changed from outside
158 IgnoreFormats ignoreFormats;
161 void setIgnoreFormat(string type, bool value)
163 ignoreFormats.setIgnoreFormat(type, value);
169 bool parse_bool(docstring & howto)
174 howto = split(howto, var, ' ');
179 class MatchString : public binary_function<Paragraph, pos_type, int>
182 MatchString(docstring const & str, bool cs, bool mw)
183 : str(str), case_sens(cs), whole_words(mw)
186 // returns true if the specified string is at the specified position
187 // del specifies whether deleted strings in ct mode will be considered
188 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
190 return par.find(str, case_sens, whole_words, pos, del);
198 // match whole words only
203 int findForward(DocIterator & cur, MatchString const & match,
204 bool find_del = true)
206 for (; cur; cur.forwardChar())
207 if (cur.inTexted()) {
208 int len = match(cur.paragraph(), cur.pos(), find_del);
216 int findBackwards(DocIterator & cur, MatchString const & match,
217 bool find_del = true)
221 if (cur.inTexted()) {
222 int len = match(cur.paragraph(), cur.pos(), find_del);
231 bool searchAllowed(docstring const & str)
234 frontend::Alert::error(_("Search error"), _("Search string is empty"));
241 bool findOne(BufferView * bv, docstring const & searchstr,
242 bool case_sens, bool whole, bool forward,
243 bool find_del = true, bool check_wrap = false)
245 if (!searchAllowed(searchstr))
248 DocIterator cur = forward
249 ? bv->cursor().selectionEnd()
250 : bv->cursor().selectionBegin();
252 MatchString const match(searchstr, case_sens, whole);
254 int match_len = forward
255 ? findForward(cur, match, find_del)
256 : findBackwards(cur, match, find_del);
259 bv->putSelectionAt(cur, match_len, !forward);
260 else if (check_wrap) {
261 DocIterator cur_orig(bv->cursor());
264 q = _("End of file reached while searching forward.\n"
265 "Continue searching from the beginning?");
267 q = _("Beginning of file reached while searching backward.\n"
268 "Continue searching from the end?");
269 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
270 q, 0, 1, _("&Yes"), _("&No"));
271 if (wrap_answer == 0) {
273 bv->cursor().clear();
274 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
276 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
277 bv->cursor().backwardPos();
279 bv->clearSelection();
280 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
283 bv->cursor().setCursor(cur_orig);
287 return match_len > 0;
291 int replaceAll(BufferView * bv,
292 docstring const & searchstr, docstring const & replacestr,
293 bool case_sens, bool whole)
295 Buffer & buf = bv->buffer();
297 if (!searchAllowed(searchstr) || buf.isReadonly())
300 DocIterator cur_orig(bv->cursor());
302 MatchString const match(searchstr, case_sens, whole);
305 int const rsize = replacestr.size();
306 int const ssize = searchstr.size();
309 cur.setCursor(doc_iterator_begin(&buf));
310 int match_len = findForward(cur, match, false);
311 while (match_len > 0) {
312 // Backup current cursor position and font.
313 pos_type const pos = cur.pos();
314 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
316 int striked = ssize -
317 cur.paragraph().eraseChars(pos, pos + match_len,
318 buf.params().track_changes);
319 cur.paragraph().insert(pos, replacestr, font,
320 Change(buf.params().track_changes
322 : Change::UNCHANGED));
323 for (int i = 0; i < rsize + striked; ++i)
326 match_len = findForward(cur, match, false);
329 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
331 cur_orig.fixIfBroken();
332 bv->setCursor(cur_orig);
338 // the idea here is that we are going to replace the string that
339 // is selected IF it is the search string.
340 // if there is a selection, but it is not the search string, then
341 // we basically ignore it. (FIXME We ought to replace only within
343 // if there is no selection, then:
344 // (i) if some search string has been provided, then we find it.
345 // (think of how the dialog works when you hit "replace" the
347 // (ii) if no search string has been provided, then we treat the
348 // word the cursor is in as the search string. (why? i have no
349 // idea.) but this only works in text?
351 // returns the number of replacements made (one, if any) and
352 // whether anything at all was done.
353 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
354 docstring const & replacestr, bool case_sens,
355 bool whole, bool forward, bool findnext)
357 Cursor & cur = bv->cursor();
359 if (!cur.selection()) {
360 // no selection, non-empty search string: find it
361 if (!searchstr.empty()) {
362 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
363 return make_pair(found, 0);
365 // empty search string
368 return make_pair(false, 0);
369 // select current word and treat it as the search string.
370 // This causes a minor bug as undo will restore this selection,
371 // which the user did not create (#8986).
372 cur.innerText()->selectWord(cur, WHOLE_WORD);
373 searchstr = cur.selectionAsString(false);
376 // if we still don't have a search string, report the error
378 if (!searchAllowed(searchstr))
379 return make_pair(false, 0);
381 bool have_selection = cur.selection();
382 docstring const selected = cur.selectionAsString(false);
385 ? searchstr == selected
386 : compare_no_case(searchstr, selected) == 0;
388 // no selection or current selection is not search word:
389 // just find the search word
390 if (!have_selection || !match) {
391 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
392 return make_pair(found, 0);
395 // we're now actually ready to replace. if the buffer is
396 // read-only, we can't, though.
397 if (bv->buffer().isReadonly())
398 return make_pair(false, 0);
400 cap::replaceSelectionWithString(cur, replacestr);
402 cur.pos() += replacestr.length();
403 LASSERT(cur.pos() <= cur.lastpos(),
404 cur.pos() = cur.lastpos());
407 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
409 return make_pair(true, 1);
415 docstring const find2string(docstring const & search,
416 bool casesensitive, bool matchword, bool forward)
420 << int(casesensitive) << ' '
421 << int(matchword) << ' '
427 docstring const replace2string(docstring const & replace,
428 docstring const & search,
429 bool casesensitive, bool matchword,
430 bool all, bool forward, bool findnext)
433 ss << replace << '\n'
435 << int(casesensitive) << ' '
436 << int(matchword) << ' '
438 << int(forward) << ' '
444 bool lyxfind(BufferView * bv, FuncRequest const & ev)
446 if (!bv || ev.action() != LFUN_WORD_FIND)
449 //lyxerr << "find called, cmd: " << ev << endl;
451 // data is of the form
453 // <casesensitive> <matchword> <forward>"
455 docstring howto = split(ev.argument(), search, '\n');
457 bool casesensitive = parse_bool(howto);
458 bool matchword = parse_bool(howto);
459 bool forward = parse_bool(howto);
461 return findOne(bv, search, casesensitive, matchword, forward, true, true);
465 bool lyxreplace(BufferView * bv,
466 FuncRequest const & ev, bool has_deleted)
468 if (!bv || ev.action() != LFUN_WORD_REPLACE)
471 // data is of the form
474 // <casesensitive> <matchword> <all> <forward> <findnext>"
477 docstring howto = split(ev.argument(), rplc, '\n');
478 howto = split(howto, search, '\n');
480 bool casesensitive = parse_bool(howto);
481 bool matchword = parse_bool(howto);
482 bool all = parse_bool(howto);
483 bool forward = parse_bool(howto);
484 bool findnext = howto.empty() ? true : parse_bool(howto);
489 int replace_count = 0;
491 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
492 update = replace_count > 0;
495 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
497 replace_count = rv.second;
500 Buffer const & buf = bv->buffer();
502 // emit message signal.
503 buf.message(_("String not found."));
505 if (replace_count == 0) {
506 buf.message(_("String found."));
507 } else if (replace_count == 1) {
508 buf.message(_("String has been replaced."));
510 docstring const str =
511 bformat(_("%1$d strings have been replaced."), replace_count);
515 } else if (findnext) {
516 // if we have deleted characters, we do not replace at all, but
517 // rather search for the next occurence
518 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
521 bv->message(_("String not found."));
527 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
529 for (; cur; cur.forwardPos())
530 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
534 DocIterator cur_orig(bv->cursor());
535 docstring q = _("End of file reached while searching forward.\n"
536 "Continue searching from the beginning?");
537 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
538 q, 0, 1, _("&Yes"), _("&No"));
539 if (wrap_answer == 0) {
540 bv->cursor().clear();
541 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
542 bv->clearSelection();
543 cur.setCursor(bv->cursor().selectionBegin());
544 if (findNextChange(bv, cur, false))
547 bv->cursor().setCursor(cur_orig);
554 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
556 for (cur.backwardPos(); cur; cur.backwardPos()) {
557 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
562 DocIterator cur_orig(bv->cursor());
563 docstring q = _("Beginning of file reached while searching backward.\n"
564 "Continue searching from the end?");
565 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
566 q, 0, 1, _("&Yes"), _("&No"));
567 if (wrap_answer == 0) {
568 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
569 bv->cursor().backwardPos();
570 bv->clearSelection();
571 cur.setCursor(bv->cursor().selectionBegin());
572 if (findPreviousChange(bv, cur, false))
575 bv->cursor().setCursor(cur_orig);
582 bool selectChange(Cursor & cur, bool forward)
584 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
586 Change ch = cur.paragraph().lookupChange(cur.pos());
588 CursorSlice tip1 = cur.top();
589 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
590 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
591 if (!ch2.isSimilarTo(ch))
594 CursorSlice tip2 = cur.top();
595 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
597 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
598 if (!ch2.isSimilarTo(ch)) {
599 // take a step forward to correctly set the selection
607 cur.bv().mouseSetCursor(cur, false);
609 cur.bv().mouseSetCursor(cur, true);
617 bool findChange(BufferView * bv, bool forward)
620 cur.setCursor(forward ? bv->cursor().selectionEnd()
621 : bv->cursor().selectionBegin());
622 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
623 return selectChange(cur, forward);
628 bool findNextChange(BufferView * bv)
630 return findChange(bv, true);
634 bool findPreviousChange(BufferView * bv)
636 return findChange(bv, false);
643 typedef vector<pair<string, string> > Escapes;
645 /// A map of symbols and their escaped equivalent needed within a regex.
646 /// @note Beware of order
647 Escapes const & get_regexp_escapes()
649 typedef std::pair<std::string, std::string> P;
651 static Escapes escape_map;
652 if (escape_map.empty()) {
653 escape_map.push_back(P("$", "_x_$"));
654 escape_map.push_back(P("{", "_x_{"));
655 escape_map.push_back(P("}", "_x_}"));
656 escape_map.push_back(P("[", "_x_["));
657 escape_map.push_back(P("]", "_x_]"));
658 escape_map.push_back(P("(", "_x_("));
659 escape_map.push_back(P(")", "_x_)"));
660 escape_map.push_back(P("+", "_x_+"));
661 escape_map.push_back(P("*", "_x_*"));
662 escape_map.push_back(P(".", "_x_."));
663 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
664 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
665 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
666 escape_map.push_back(P("_x_", "\\"));
671 /// A map of lyx escaped strings and their unescaped equivalent.
672 Escapes const & get_lyx_unescapes()
674 typedef std::pair<std::string, std::string> P;
676 static Escapes escape_map;
677 if (escape_map.empty()) {
678 escape_map.push_back(P("\\%", "%"));
679 escape_map.push_back(P("\\mathcircumflex ", "^"));
680 escape_map.push_back(P("\\mathcircumflex", "^"));
681 escape_map.push_back(P("\\backslash ", "\\"));
682 escape_map.push_back(P("\\backslash", "\\"));
683 escape_map.push_back(P("\\\\{", "_x_<"));
684 escape_map.push_back(P("\\\\}", "_x_>"));
685 escape_map.push_back(P("\\sim ", "~"));
686 escape_map.push_back(P("\\sim", "~"));
691 /// A map of escapes turning a regexp matching text to one matching latex.
692 Escapes const & get_regexp_latex_escapes()
694 typedef std::pair<std::string, std::string> P;
696 static Escapes escape_map;
697 if (escape_map.empty()) {
698 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
699 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
700 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
701 escape_map.push_back(P("\\[", "\\{\\[\\}"));
702 escape_map.push_back(P("\\]", "\\{\\]\\}"));
703 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
704 escape_map.push_back(P("%", "\\\\\\%"));
705 escape_map.push_back(P("#", "\\\\#"));
710 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
711 ** the found occurrence were escaped.
713 string apply_escapes(string s, Escapes const & escape_map)
715 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
716 Escapes::const_iterator it;
717 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
718 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
719 unsigned int pos = 0;
720 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
721 s.replace(pos, it->first.length(), it->second);
722 LYXERR(Debug::FIND, "After escape: " << s);
723 pos += it->second.length();
724 // LYXERR(Debug::FIND, "pos: " << pos);
727 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
732 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
733 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
734 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
735 string escape_for_regex(string s, bool match_latex)
738 while (pos < s.size()) {
739 size_t new_pos = s.find("\\regexp{", pos);
740 if (new_pos == string::npos)
742 LYXERR(Debug::FIND, "new_pos: " << new_pos);
743 string t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
744 LYXERR(Debug::FIND, "t [lyx]: " << t);
745 t = apply_escapes(t, get_regexp_escapes());
746 LYXERR(Debug::FIND, "t [rxp]: " << t);
747 s.replace(pos, new_pos - pos, t);
748 new_pos = pos + t.size();
749 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
750 LYXERR(Debug::FIND, "new_pos: " << new_pos);
751 if (new_pos == s.size())
753 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
754 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
755 LYXERR(Debug::FIND, "end_pos: " << end_pos);
756 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
757 LYXERR(Debug::FIND, "t in regexp : " << t);
758 t = apply_escapes(t, get_lyx_unescapes());
759 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
761 t = apply_escapes(t, get_regexp_latex_escapes());
762 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
764 if (end_pos == s.size()) {
765 s.replace(new_pos, end_pos - new_pos, t);
766 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
769 s.replace(new_pos, end_pos + 13 - new_pos, t);
770 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
771 pos = new_pos + t.size();
772 LYXERR(Debug::FIND, "pos: " << pos);
778 /// Wrapper for lyx::regex_replace with simpler interface
779 bool regex_replace(string const & s, string & t, string const & searchstr,
780 string const & replacestr)
782 lyx::regex e(searchstr, regex_constants::ECMAScript);
784 ostream_iterator<char, char> it(oss);
785 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
786 // tolerate t and s be references to the same variable
787 bool rv = (s != oss.str());
793 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
795 ** Verify that closed braces exactly match open braces. This avoids that, for example,
796 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
799 ** Number of open braces that must remain open at the end for the verification to succeed.
801 bool braces_match(string::const_iterator const & beg,
802 string::const_iterator const & end,
806 string::const_iterator it = beg;
807 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
808 for (; it != end; ++it) {
809 // Skip escaped braces in the count
814 } else if (*it == '{') {
816 } else if (*it == '}') {
817 if (open_pars == 0) {
818 LYXERR(Debug::FIND, "Found unmatched closed brace");
824 if (open_pars != unmatched) {
825 LYXERR(Debug::FIND, "Found " << open_pars
826 << " instead of " << unmatched
827 << " unmatched open braces at the end of count");
830 LYXERR(Debug::FIND, "Braces match as expected");
835 /** The class performing a match between a position in the document and the FindAdvOptions.
837 class MatchStringAdv {
839 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
841 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
842 ** constructor as opt.search, under the opt.* options settings.
845 ** If set, then match is searched only against beginning of text starting at cur.
846 ** If unset, then match is searched anywhere in text starting at cur.
849 ** The length of the matching text, or zero if no match was found.
851 int operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
856 /// first buffer on which search was started
857 lyx::Buffer * const p_first_buf;
859 FindAndReplaceOptions const & opt;
862 /// Auxiliary find method (does not account for opt.matchword)
863 int findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
865 /** Normalize a stringified or latexified LyX paragraph.
869 ** <li>if search is not casesensitive, then lowercase the string;
870 ** <li>remove any newline at begin or end of the string;
871 ** <li>replace any newline in the middle of the string with a simple space;
872 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
875 ** @todo Normalization should also expand macros, if the corresponding
876 ** search option was checked.
878 string normalize(docstring const & s, bool hack_braces) const;
879 // normalized string to search
880 string par_as_string;
881 // regular expression to use for searching
883 // same as regexp, but prefixed with a ".*?"
885 // leading format material as string
886 string lead_as_string;
887 // par_as_string after removal of lead_as_string
888 string par_as_string_nolead;
889 // unmatched open braces in the search string/regexp
891 // number of (.*?) subexpressions added at end of search regexp for closing
892 // environments, math mode, styles, etc...
894 // Are we searching with regular expressions ?
899 static docstring buffer_to_latex(Buffer & buffer)
901 OutputParams runparams(&buffer.params().encoding());
902 odocstringstream ods;
904 runparams.nice = true;
905 runparams.flavor = OutputParams::LATEX;
906 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
907 // No side effect of file copying and image conversion
908 runparams.dryrun = true;
909 runparams.for_search = true;
910 pit_type const endpit = buffer.paragraphs().size();
911 for (pit_type pit = 0; pit != endpit; ++pit) {
912 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
913 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
919 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
922 if (!opt.ignoreformat) {
923 str = buffer_to_latex(buffer);
925 OutputParams runparams(&buffer.params().encoding());
926 runparams.nice = true;
927 runparams.flavor = OutputParams::LATEX;
928 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
929 runparams.dryrun = true;
930 runparams.for_search = true;
931 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
932 Paragraph const & par = buffer.paragraphs().at(pit);
933 LYXERR(Debug::FIND, "Adding to search string: '"
934 << par.asString(pos_type(0), par.size(),
935 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
938 str += par.asString(pos_type(0), par.size(),
939 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
947 /// Return separation pos between the leading material and the rest
948 static size_t identifyLeading(string const & s)
951 // @TODO Support \item[text]
952 // Kornel: Added textsl, textsf, textit, texttt and noun
953 // + allow to search for colored text too
954 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)\\*?)\\{", "")
955 || regex_replace(t, t, REGEX_BOS "\\$", "")
956 || regex_replace(t, t, REGEX_BOS "\\\\\\[ ", "")
957 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
958 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
960 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
965 * Given a latexified string, retrieve some handled features
966 * The features of the regex will later be compared with the features
967 * of the searched text. If the regex features are not a
968 * subset of the analized, then, in not format ignoring search
969 * we can early stop the search in the relevant inset.
971 typedef map<string, bool> Features;
973 static Features identifyFeatures(string const & s)
975 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
976 static regex const valid("^("
978 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
979 "emph|noun|text(bf|md|sl|sf|it|tt)|"
980 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
981 "(u|uu)line|(s|x)out|uwave|"
982 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
983 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
988 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
991 if (sub.str(1).compare("regexp") == 0) {
995 string token = sub.str(1);
997 if (regex_match(token, sub2, valid)) {
1005 if (sub.str(1).compare("endregexp") == 0) {
1015 * defines values features of a key "\\[a-z]+{"
1020 /* Char type with content discarded
1021 * like \hspace{1cm} */
1023 /* Char, like \backslash */
1025 /* \part, \section*, ... */
1027 /* title, author etc */
1029 /* \foreignlanguage{ngerman}, ... */
1032 * to discard language in content */
1035 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1037 /* fonts, colors, markups, ... */
1039 /* footnotesize, ... large, ...
1040 * Ignore all of them */
1043 /* inputencoding, shortcut, ...
1044 * Discard also content, because they do not help in search */
1047 * like remove, but also all arguments */
1049 /* item, listitem */
1051 /* tex, latex, ... like isChar */
1053 /* like \lettrine[lines=5]{}{} */
1055 /* End of arguments marker for lettrine,
1056 * so that they can be ignored */
1066 parenthesiscount(1),
1070 KeyInfo(KeyType type, int parcount, bool disable)
1076 parenthesiscount(parcount),
1085 int parenthesiscount;
1087 bool used; /* by pattern */
1092 Border(int l=0, int u=0) : low(l), upper(u) {};
1097 #define MAXOPENED 30
1099 bool isPatternString;
1101 explicit Intervall(bool isPattern) :
1102 isPatternString(isPattern),
1104 actualdeptindex(0) { depts[0] = 0; closes[0] = 0;};
1107 int depts[MAXOPENED];
1108 int closes[MAXOPENED];
1109 int actualdeptindex;
1110 Border borders[2*MAXOPENED];
1111 int previousNotIgnored(int);
1112 int nextNotIgnored(int);
1113 void handleOpenP(int i);
1114 void handleCloseP(int i, bool closingAllowed);
1115 void resetOpenedP(int openPos);
1116 void addIntervall(int upper);
1117 void addIntervall(int low, int upper); /* if explicit */
1118 void setForDefaultLang(int upTo);
1119 int findclosing(int start, int end, char up, char down, int repeat);
1120 void handleParentheses(int lastpos, bool closingAllowed);
1123 void output(ostringstream &os, int lastpos);
1124 // string show(int lastpos);
1127 void Intervall::setForDefaultLang(int upTo)
1129 // Enable the use of first token again
1130 if (ignoreidx >= 0) {
1131 if (borders[0].low < upTo)
1132 borders[0].low = upTo;
1133 if (borders[0].upper < upTo)
1134 borders[0].upper = upTo;
1138 static void checkDepthIndex(int val)
1140 static int maxdepthidx = MAXOPENED-2;
1141 if (val > maxdepthidx) {
1143 LYXERR0("maxdepthidx now " << val);
1147 static void checkIgnoreIdx(int val)
1149 static int maxignoreidx = 2*MAXOPENED - 4;
1150 if (val > maxignoreidx) {
1152 LYXERR0("maxignoreidx now " << val);
1157 * Expand the region of ignored parts of the input latex string
1158 * The region is only relevant in output()
1160 void Intervall::addIntervall(int low, int upper)
1163 if (low == upper) return;
1164 for (idx = ignoreidx+1; idx > 0; --idx) {
1165 if (low > borders[idx-1].upper) {
1169 Border br(low, upper);
1170 if (idx > ignoreidx) {
1173 checkIgnoreIdx(ignoreidx);
1177 // Expand only if one of the new bound is inside the interwall
1178 // We know here that br.low > borders[idx-1].upper
1179 if (br.upper < borders[idx].low) {
1180 // We have to insert at this pos
1181 for (int i = ignoreidx+1; i > idx; --i) {
1182 borders[i] = borders[i-1];
1186 checkIgnoreIdx(ignoreidx);
1189 // Here we know, that we are overlapping
1190 if (br.low > borders[idx].low)
1191 br.low = borders[idx].low;
1192 // check what has to be concatenated
1194 for (int i = idx; i <= ignoreidx; i++) {
1195 if (br.upper >= borders[i].low) {
1197 if (br.upper < borders[i].upper)
1198 br.upper = borders[i].upper;
1204 // count should be >= 1 here
1207 for (int i = idx + count; i <= ignoreidx; i++) {
1208 borders[i-count+1] = borders[i];
1210 ignoreidx -= count - 1;
1216 void Intervall::handleOpenP(int i)
1219 depts[actualdeptindex] = i+1;
1220 closes[actualdeptindex] = -1;
1221 checkDepthIndex(actualdeptindex);
1224 void Intervall::handleCloseP(int i, bool closingAllowed)
1226 if (actualdeptindex <= 0) {
1227 if (! closingAllowed)
1228 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1229 // if we are at the very end
1230 addIntervall(i, i+1);
1233 closes[actualdeptindex] = i+1;
1238 void Intervall::resetOpenedP(int openPos)
1240 // Used as initializer for foreignlanguage entry
1241 actualdeptindex = 1;
1242 depts[1] = openPos+1;
1246 int Intervall::previousNotIgnored(int start)
1248 int idx = 0; /* int intervalls */
1249 for (idx = ignoreidx; idx >= 0; --idx) {
1250 if (start > borders[idx].upper)
1252 if (start >= borders[idx].low)
1253 start = borders[idx].low-1;
1258 int Intervall::nextNotIgnored(int start)
1260 int idx = 0; /* int intervalls */
1261 for (idx = 0; idx <= ignoreidx; idx++) {
1262 if (start < borders[idx].low)
1264 if (start < borders[idx].upper)
1265 start = borders[idx].upper;
1270 typedef map<string, KeyInfo> KeysMap;
1271 typedef vector< KeyInfo> Entries;
1272 static KeysMap keys = map<string, KeyInfo>();
1279 void buildKeys(bool);
1280 void buildEntries(bool);
1281 void makeKey(const string &, KeyInfo, bool isPatternString);
1282 void processRegion(int start, int region_end); /* remove {} parts */
1283 void removeHead(KeyInfo&, int count=0);
1286 LatexInfo(string par, bool isPatternString) : entidx(-1), interval(isPatternString) {
1288 interval.hasTitle = false;
1289 interval.titleValue = "";
1290 buildKeys(isPatternString);
1291 entries = vector<KeyInfo>();
1292 buildEntries(isPatternString);
1296 if (entries.empty()) {
1299 if (entries[0].keytype == KeyInfo::isTitle) {
1300 if (! entries[0].disabled) {
1301 interval.hasTitle = true;
1302 interval.titleValue = entries[0].head;
1305 interval.hasTitle = false;
1306 interval.titleValue = "";
1308 removeHead(entries[0]);
1309 if (entries.size() > 1)
1318 if (int(entries.size()) > entidx) {
1325 bool setNextKey(int idx) {
1326 if ((idx == entidx) && (entidx >= 0)) {
1333 int find(int start, KeyInfo::KeyType keytype) {
1337 while (tmpIdx < int(entries.size())) {
1338 if (entries[tmpIdx].keytype == keytype)
1344 int process(ostringstream &os, KeyInfo &actual);
1345 int dispatch(ostringstream &os, int previousStart, KeyInfo &actual);
1346 // string show(int lastpos) { return interval.show(lastpos);};
1347 int nextNotIgnored(int start) { return interval.nextNotIgnored(start);};
1348 KeyInfo &getKeyInfo(int keyinfo) {
1349 static KeyInfo invalidInfo = KeyInfo();
1350 if ((keyinfo < 0) || ( keyinfo >= int(entries.size())))
1353 return entries[keyinfo];
1355 void setForDefaultLang(int upTo) {interval.setForDefaultLang(upTo);};
1356 void addIntervall(int low, int up) { interval.addIntervall(low, up); };
1360 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1365 for (int i = start; i < end; i += 1 + skip) {
1369 if (c == '\\') skip = 1;
1373 else if (c == down) {
1375 if ((repeat <= 0) || (par[i+1] != up))
1393 vector<MathEntry> entries;
1398 void insert(string wait, size_t start, size_t end) {
1399 MathEntry m = MathEntry();
1401 m.mathStart = start;
1403 m.mathSize = end - start;
1404 entries.push_back(m);
1406 bool empty() { return entries.empty(); };
1407 size_t getEndPos() {
1408 if (entries.empty() || (actualIdx >= entries.size())) {
1411 return entries[actualIdx].mathEnd;
1413 size_t getStartPos() {
1414 if (entries.empty() || (actualIdx >= entries.size())) {
1415 return 100000; /* definitely enough? */
1417 return entries[actualIdx].mathStart;
1419 size_t getFirstPos() {
1421 return getStartPos();
1424 if (entries.empty() || (actualIdx >= entries.size())) {
1427 return entries[actualIdx].mathSize;
1429 void incrEntry() { actualIdx++; };
1432 void LatexInfo::buildEntries(bool isPatternString)
1434 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1435 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1436 static bool disableLanguageOverride = false;
1437 smatch sub, submath;
1438 bool evaluatingRegexp = false;
1440 bool evaluatingMath = false;
1441 bool evaluatingCode = false;
1443 bool evaluatingOptional = false;
1444 size_t optionalEnd = 0;
1447 bool math_end_waiting = false;
1448 size_t math_pos = 10000;
1451 for (sregex_iterator itmath(interval.par.begin(), interval.par.end(), rmath), end; itmath != end; ++itmath) {
1453 if (math_end_waiting) {
1454 size_t pos = submath.position(size_t(0));
1455 if ((math_end == "$") &&
1456 (submath.str(0) == "$") &&
1457 (interval.par[pos-1] != '\\')) {
1458 mi.insert("$", math_pos, pos + 1);
1459 math_end_waiting = false;
1461 else if ((math_end == "\\]") &&
1462 (submath.str(0) == "\\]")) {
1463 mi.insert("\\]", math_pos, pos + 2);
1464 math_end_waiting = false;
1466 else if ((submath.str(1).compare("end") == 0) &&
1467 (submath.str(2).compare(math_end) == 0)) {
1468 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1469 math_end_waiting = false;
1475 if (submath.str(1).compare("begin") == 0) {
1476 math_end_waiting = true;
1477 math_end = submath.str(2);
1478 math_pos = submath.position(size_t(0));
1480 else if (submath.str(0).compare("\\[") == 0) {
1481 math_end_waiting = true;
1483 math_pos = submath.position(size_t(0));
1485 else if (submath.str(0) == "$") {
1486 size_t pos = submath.position(size_t(0));
1487 if ((pos == 0) || (interval.par[pos-1] != '\\')) {
1488 math_end_waiting = true;
1495 // Ignore language if there is math somewhere in pattern-string
1496 if (isPatternString) {
1499 keys["foreignlanguage"].disabled = true;
1500 disableLanguageOverride = true;
1503 disableLanguageOverride = false;
1506 if (disableLanguageOverride) {
1507 keys["foreignlanguage"].disabled = true;
1510 math_pos = mi.getFirstPos();
1511 for (sregex_iterator it(interval.par.begin(), interval.par.end(), rkeys), end; it != end; ++it) {
1513 string key = sub.str(3);
1515 if (sub.str(0)[0] == '\\')
1516 key = sub.str(0)[1];
1520 size_t k_pos = sub.position(size_t(0));
1521 if ((k_pos > 0) && (interval.par[k_pos - 1] == '\\')) {
1522 // Escaped '$', ignoring
1528 if (evaluatingRegexp) {
1529 if (sub.str(1).compare("endregexp") == 0) {
1530 evaluatingRegexp = false;
1531 // found._tokenstart already set
1532 found._dataEnd = sub.position(size_t(0)) + 13;
1533 found._dataStart = found._dataEnd;
1534 found._tokensize = found._dataEnd - found._tokenstart;
1535 found.parenthesiscount = 0;
1539 if (evaluatingMath) {
1540 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1542 evaluatingMath = false;
1544 math_pos = mi.getStartPos();
1546 if (keys.find(key) == keys.end()) {
1547 found = KeyInfo(KeyInfo::isStandard, 0, true);
1548 if (isPatternString) {
1549 found.keytype = KeyInfo::isChar;
1550 found.disabled = false;
1557 if (key.compare("regexp") == 0) {
1558 evaluatingRegexp = true;
1559 found._tokenstart = sub.position(size_t(0));
1560 found._tokensize = 0;
1564 // Handle the other params of key
1565 if (found.keytype == KeyInfo::isIgnored)
1567 else if (found.keytype == KeyInfo::isMath) {
1568 if (size_t(sub.position(size_t(0))) == math_pos) {
1570 found._tokenstart = sub.position(size_t(0));
1571 found._tokensize = mi.getSize();
1572 found._dataEnd = found._tokenstart + found._tokensize;
1573 found._dataStart = found._dataEnd;
1574 found.parenthesiscount = 0;
1575 evaluatingMath = true;
1578 // begin|end of unknown env, discard
1579 // First handle tables
1580 // longtable|tabular
1581 bool discardComment;
1583 found.keytype = KeyInfo::doRemove;
1584 if ((sub.str(5).compare("longtable") == 0) ||
1585 (sub.str(5).compare("tabular") == 0)) {
1586 discardComment = true; /* '%' */
1589 discardComment = false;
1590 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1592 string token = sub.str(5);
1593 if (regex_match(token, sub2, removeArgs)) {
1594 found.keytype = KeyInfo::removeWithArg;
1597 // discard spaces before pos(0)
1598 int pos = sub.position(size_t(0));
1600 for (count = 0; pos - count > 0; count++) {
1601 char c = interval.par[pos-count-1];
1602 if (discardComment) {
1603 if ((c != ' ') && (c != '%'))
1609 found._tokenstart = pos - count;
1610 if (sub.str(1).compare(0, 5, "begin") == 0) {
1611 size_t pos1 = pos + sub.str(0).length();
1612 if (sub.str(5).compare("cjk") == 0) {
1613 pos1 = interval.findclosing(pos1+1, interval.par.length()) + 1;
1614 if ((interval.par[pos1] == '{') && (interval.par[pos1+1] == '}'))
1616 found.keytype = KeyInfo::isMain;
1617 found._dataStart = pos1;
1618 found._dataEnd = interval.par.length();
1619 found.disabled = keys["foreignlanguage"].disabled;
1620 found.used = keys["foreignlanguage"].used;
1621 found._tokensize = pos1 - found._tokenstart;
1622 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1625 while (interval.par[pos1] == '[') {
1626 pos1 = interval.findclosing(pos1+1, interval.par.length(), '[', ']')+1;
1628 if (interval.par[pos1] == '{') {
1629 found._dataEnd = interval.findclosing(pos1+1, interval.par.length()) + 1;
1632 found._dataEnd = pos1;
1634 found._dataStart = found._dataEnd;
1635 found._tokensize = count + found._dataEnd - pos;
1636 found.parenthesiscount = 0;
1637 found.disabled = true;
1641 // Handle "\end{...}"
1642 found._dataStart = pos + sub.str(0).length();
1643 found._dataEnd = found._dataStart;
1644 found._tokensize = count + found._dataEnd - pos;
1645 found.parenthesiscount = 0;
1646 found.disabled = true;
1650 else if (found.keytype != KeyInfo::isRegex) {
1651 found._tokenstart = sub.position(size_t(0));
1652 if (found.parenthesiscount == 0) {
1653 // Probably to be discarded
1654 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1655 char following = interval.par[following_pos];
1656 if (following == ' ')
1657 found.head = "\\" + sub.str(3) + " ";
1658 else if (following == '=') {
1659 // like \uldepth=1000pt
1660 found.head = sub.str(0);
1663 found.head = "\\" + key;
1664 found._tokensize = found.head.length();
1665 found._dataEnd = found._tokenstart + found._tokensize;
1666 found._dataStart = found._dataEnd;
1669 int params = found._tokenstart + key.length() + 1;
1670 if (evaluatingOptional) {
1671 if (size_t(found._tokenstart) > optionalEnd) {
1672 evaluatingOptional = false;
1675 found.disabled = true;
1678 int optend = params;
1679 while (interval.par[optend] == '[') {
1680 // discard optional parameters
1681 optend = interval.findclosing(optend+1, interval.par.length(), '[', ']') + 1;
1683 if (optend > params) {
1684 key += interval.par.substr(params, optend-params);
1685 evaluatingOptional = true;
1686 optionalEnd = optend;
1688 string token = sub.str(5);
1689 int closings = found.parenthesiscount;
1690 if (found.parenthesiscount == 1) {
1691 found.head = "\\" + key + "{";
1693 else if (found.parenthesiscount > 1) {
1695 found.head = sub.str(0) + "{";
1696 closings = found.parenthesiscount - 1;
1699 found.head = "\\" + key + "{";
1702 found._dataStart = found._tokenstart + found.head.length();
1703 if (interval.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1704 found._dataStart += 15;
1706 size_t endpos = interval.findclosing(found._dataStart, interval.par.length(), '{', '}', closings);
1707 if (found.keytype == KeyInfo::isList) {
1708 // Check if it really is list env
1709 static regex const listre("^([a-z]+)$");
1711 if (!regex_match(token, sub2, listre)) {
1712 // Change the key of this entry. It is not in a list/item environment
1713 found.keytype = KeyInfo::endArguments;
1716 if (found.keytype == KeyInfo::noMain) {
1717 evaluatingCode = true;
1719 codeStart = found._dataStart;
1721 else if (evaluatingCode) {
1722 if (size_t(found._dataStart) > codeEnd)
1723 evaluatingCode = false;
1724 else if (found.keytype == KeyInfo::isMain) {
1725 // Disable this key, treate it as standard
1726 found.keytype = KeyInfo::isStandard;
1727 found.disabled = true;
1728 if ((codeEnd == interval.par.length()) &&
1729 (found._tokenstart == codeStart)) {
1730 // trickery, because the code inset starts
1731 // with \selectlanguage ...
1733 if (entries.size() > 1) {
1734 entries[entries.size()-1]._dataEnd = codeEnd;
1739 if ((endpos == interval.par.length()) &&
1740 (found.keytype == KeyInfo::doRemove)) {
1741 // Missing closing => error in latex-input?
1742 // therefore do not delete remaining data
1743 found._dataStart -= 1;
1744 found._dataEnd = found._dataStart;
1747 found._dataEnd = endpos;
1749 if (isPatternString) {
1750 keys[key].used = true;
1753 entries.push_back(found);
1757 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1759 stringstream s(keysstring);
1761 const char delim = '|';
1762 while (getline(s, key, delim)) {
1763 KeyInfo keyII(keyI);
1764 if (isPatternString) {
1767 else if ( !keys[key].used)
1768 keyII.disabled = true;
1773 void LatexInfo::buildKeys(bool isPatternString)
1776 static bool keysBuilt = false;
1777 if (keysBuilt && !isPatternString) return;
1779 // Known standard keys with 1 parameter.
1780 // Split is done, if not at start of region
1781 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1782 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1783 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1784 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1785 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1786 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
1788 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
1789 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1790 makeKey("section*|subsection*|subsubsection*|paragraph*",
1791 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1792 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1793 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
1795 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
1797 // Split is done, if not at start of region
1798 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
1799 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
1801 // Split is done always.
1802 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
1806 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1807 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1808 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1809 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1811 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1812 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1813 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1815 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1816 // Custom space/skip, remove the content (== length value)
1817 makeKey("vspace|hspace|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
1818 // Found in fr/UserGuide.lyx
1819 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1821 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1822 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1823 // Known macros to remove (including their parameter)
1825 makeKey("inputencoding|shortcut|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
1826 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
1827 // handle like standard keys with 1 parameter.
1828 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
1830 // Macros to remove, but let the parameter survive
1832 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1834 // Remove language spec from content of these insets
1835 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
1837 // Same effect as previous, parameter will survive (because there is no one anyway)
1839 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1840 // Remove table decorations
1841 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
1842 // Discard shape-header.
1843 // For footnote too, because of possible lang settings
1844 // and wrong handling if used 'KeyInfo::noMain'
1845 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1846 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1847 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1848 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1849 makeKey("hphantom|footnote|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1850 // like ('tiny{}' or '\tiny ' ... )
1851 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
1853 // Survives, like known character
1854 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1855 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
1857 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1858 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1859 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1861 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1862 // Remove RTL/LTR marker
1863 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1864 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
1865 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
1866 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
1867 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
1868 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
1869 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1870 if (isPatternString) {
1871 // Allow the first searched string to rebuild the keys too
1875 // no need to rebuild again
1881 * Keep the list of actual opened parentheses actual
1882 * (e.g. depth == 4 means there are 4 '{' not processed yet)
1884 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
1887 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
1891 if (c == '\\') skip = 1;
1892 else if (c == '{') {
1895 else if (c == '}') {
1896 handleCloseP(i, closingAllowed);
1902 string Intervall::show(int lastpos)
1904 int idx = 0; /* int intervalls */
1907 for (idx = 0; idx <= ignoreidx; idx++) {
1908 while (i < lastpos) {
1910 if (i <= borders[idx].low) {
1911 if (borders[idx].low > lastpos)
1912 printsize = lastpos - i;
1914 printsize = borders[idx].low - i;
1915 s += par.substr(i, printsize);
1917 if (i >= borders[idx].low)
1918 i = borders[idx].upper;
1921 i = borders[idx].upper;
1927 s += par.substr(i, lastpos-i);
1933 void Intervall::output(ostringstream &os, int lastpos)
1935 // get number of chars to output
1936 int idx = 0; /* int intervalls */
1939 string startTitle = titleValue;
1940 for (idx = 0; idx <= ignoreidx; idx++) {
1942 if (i <= borders[idx].low) {
1944 if (borders[idx].low > lastpos)
1945 printsize = lastpos - i;
1947 printsize = borders[idx].low - i;
1948 if (printsize > 0) {
1949 os << startTitle << par.substr(i, printsize);
1951 printed += printsize;
1954 handleParentheses(i, false);
1955 if (i >= borders[idx].low)
1956 i = borders[idx].upper;
1959 i = borders[idx].upper;
1966 os << startTitle << par.substr(i, lastpos-i);
1967 printed += lastpos-i;
1969 handleParentheses(lastpos, false);
1970 for (int i = actualdeptindex; i > 0; --i) {
1973 if (hasTitle && (printed > 0))
1975 if (! isPatternString)
1977 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
1980 void LatexInfo::processRegion(int start, int region_end)
1982 while (start < region_end) { /* Let {[} and {]} survive */
1983 if ((interval.par[start] == '{') &&
1984 (interval.par[start+1] != ']') &&
1985 (interval.par[start+1] != '[')) {
1986 // Closing is allowed past the region
1987 int closing = interval.findclosing(start+1, interval.par.length());
1988 interval.addIntervall(start, start+1);
1989 interval.addIntervall(closing, closing+1);
1991 start = interval.nextNotIgnored(start+1);
1995 void LatexInfo::removeHead(KeyInfo &actual, int count)
1997 if (actual.parenthesiscount == 0) {
1998 // "{\tiny{} ...}" ==> "{{} ...}"
1999 interval.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2002 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2003 interval.addIntervall(actual._tokenstart - count, actual._dataStart);
2004 interval.addIntervall(actual._dataEnd, actual._dataEnd+1);
2008 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2011 switch (actual.keytype)
2013 case KeyInfo::isTitle: {
2015 nextKeyIdx = getNextKey();
2018 case KeyInfo::cleanToStart: {
2019 actual._dataEnd = actual._dataStart;
2020 nextKeyIdx = getNextKey();
2021 // Search for end of arguments
2022 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2024 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2025 entries[i].disabled = true;
2027 actual._dataEnd = entries[tmpIdx]._dataEnd;
2029 while (interval.par[actual._dataEnd] == ' ')
2031 interval.addIntervall(0, actual._dataEnd+1);
2032 interval.actualdeptindex = 0;
2033 interval.depts[0] = actual._dataEnd+1;
2034 interval.closes[0] = -1;
2037 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2038 if (actual.disabled)
2039 interval.addIntervall(actual._tokenstart, actual._dataEnd);
2041 interval.addIntervall(actual._dataStart, actual._dataEnd);
2044 case KeyInfo::isChar: {
2045 nextKeyIdx = getNextKey();
2048 case KeyInfo::isSize: {
2049 if (actual.disabled || (interval.par[actual._dataStart] != '{') || (interval.par[actual._dataStart-1] == ' ')) {
2050 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2051 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2052 nextKeyIdx = getNextKey();
2054 // Here _dataStart points to '{', so correct it
2055 actual._dataStart += 1;
2056 actual._tokensize += 1;
2057 actual.parenthesiscount = 1;
2058 if (interval.par[actual._dataStart] == '}') {
2059 // Determine the end if used like '{\tiny{}...}'
2060 actual._dataEnd = interval.findclosing(actual._dataStart+1, interval.par.length()) + 1;
2061 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2064 // Determine the end if used like '\tiny{...}'
2065 actual._dataEnd = interval.findclosing(actual._dataStart, interval.par.length()) + 1;
2067 // Split on this key if not at start
2068 int start = interval.nextNotIgnored(previousStart);
2069 if (start < actual._tokenstart) {
2070 interval.output(os, actual._tokenstart);
2071 interval.addIntervall(start, actual._tokenstart);
2073 // discard entry if at end of actual
2074 nextKeyIdx = process(os, actual);
2078 case KeyInfo::endArguments:
2079 // Remove trailing '{}' too
2080 actual._dataStart += 2;
2081 actual._dataEnd += 2;
2082 interval.addIntervall(actual._tokenstart, actual._dataEnd);
2083 nextKeyIdx = getNextKey();
2085 case KeyInfo::noMain:
2087 case KeyInfo::isStandard: {
2088 if (actual.disabled) {
2090 processRegion(actual._dataStart, actual._dataStart+1);
2091 nextKeyIdx = getNextKey();
2093 // Split on this key if not at datastart of calling entry
2094 int start = interval.nextNotIgnored(previousStart);
2095 if (start < actual._tokenstart) {
2096 interval.output(os, actual._tokenstart);
2097 interval.addIntervall(start, actual._tokenstart);
2099 // discard entry if at end of actual
2100 nextKeyIdx = process(os, actual);
2104 case KeyInfo::removeWithArg: {
2105 nextKeyIdx = getNextKey();
2106 // Search for end of arguments
2107 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2109 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2110 entries[i].disabled = true;
2112 actual._dataEnd = entries[tmpIdx]._dataEnd;
2114 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2117 case KeyInfo::doRemove: {
2118 // Remove the key with all parameters and following spaces
2120 for (pos = actual._dataEnd+1; pos < interval.par.length(); pos++) {
2121 if ((interval.par[pos] != ' ') && (interval.par[pos] != '%'))
2124 interval.addIntervall(actual._tokenstart, pos);
2125 nextKeyIdx = getNextKey();
2128 case KeyInfo::isList: {
2129 // Discard space before _tokenstart
2131 for (count = 0; count < actual._tokenstart; count++) {
2132 if (interval.par[actual._tokenstart-count-1] != ' ')
2135 nextKeyIdx = getNextKey();
2136 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2138 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2140 // How else can we catch this one?
2141 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2142 entries[i].disabled = true;
2144 actual._dataEnd = entries[tmpIdx]._dataEnd;
2146 else if (nextKeyIdx > 0) {
2147 // Ignore any lang entries inside data region
2148 for (int i = nextKeyIdx; i < int(entries.size()) && entries[i]._tokenstart < actual._dataEnd; i++) {
2149 if (entries[i].keytype == KeyInfo::isMain)
2150 entries[i].disabled = true;
2153 if (actual.disabled) {
2154 interval.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2157 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2159 if (interval.par[actual._dataEnd+1] == '[') {
2160 int posdown = interval.findclosing(actual._dataEnd+2, interval.par.length(), '[', ']');
2161 if ((interval.par[actual._dataEnd+2] == '{') &&
2162 (interval.par[posdown-1] == '}')) {
2163 interval.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2164 interval.addIntervall(posdown-1, posdown+1);
2167 interval.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2168 interval.addIntervall(posdown, posdown+1);
2170 int blk = interval.nextNotIgnored(actual._dataEnd+1);
2171 if (blk > posdown) {
2172 // Discard at most 1 space after empty item
2174 for (count = 0; count < 1; count++) {
2175 if (interval.par[blk+count] != ' ')
2179 interval.addIntervall(blk, blk+count);
2184 case KeyInfo::isSectioning: {
2185 // Discard spaces before _tokenstart
2187 int val = actual._tokenstart;
2188 for (count = 0; count < actual._tokenstart;) {
2189 val = interval.previousNotIgnored(val-1);
2190 if (interval.par[val] != ' ')
2193 count = actual._tokenstart - val;
2196 if (actual.disabled) {
2197 removeHead(actual, count);
2198 nextKeyIdx = getNextKey();
2200 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2201 nextKeyIdx = process(os, actual);
2205 case KeyInfo::isMath: {
2206 // Same as regex, use the content unchanged
2207 nextKeyIdx = getNextKey();
2210 case KeyInfo::isRegex: {
2211 // DO NOT SPLIT ON REGEX
2213 nextKeyIdx = getNextKey();
2216 case KeyInfo::isIgnored: {
2217 // Treat like a character for now
2218 nextKeyIdx = getNextKey();
2221 case KeyInfo::isMain: {
2222 if (interval.par.substr(actual._dataStart, 2) == "% ")
2223 interval.addIntervall(actual._dataStart, actual._dataStart+2);
2224 if (actual.disabled) {
2226 if ((interval.par.substr(actual._dataStart, 3) == " \\[") ||
2227 (interval.par.substr(actual._dataStart, 8) == " \\begin{")) {
2228 // Discard also the space before math-equation
2229 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2231 nextKeyIdx = getNextKey();
2232 // interval.resetOpenedP(actual._dataStart-1);
2235 if (actual._tokenstart == 0) {
2236 // for the first (and maybe dummy) language
2237 interval.setForDefaultLang(actual._tokenstart + actual._tokensize);
2239 interval.resetOpenedP(actual._dataStart-1);
2243 case KeyInfo::invalid:
2244 // This cannot happen, already handled
2247 // LYXERR0("Unhandled keytype");
2248 nextKeyIdx = getNextKey();
2255 int LatexInfo::process(ostringstream &os, KeyInfo &actual )
2257 int end = interval.nextNotIgnored(actual._dataEnd);
2258 int oldStart = actual._dataStart;
2259 int nextKeyIdx = getNextKey();
2261 if ((nextKeyIdx < 0) ||
2262 (entries[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2263 (entries[nextKeyIdx].keytype == KeyInfo::invalid)) {
2264 if (oldStart <= end) {
2265 processRegion(oldStart, end);
2270 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2272 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2273 (void) dispatch(os, actual._dataStart, nextKey);
2274 end = nextKey._tokenstart;
2277 processRegion(oldStart, nextKey._tokenstart);
2278 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2280 oldStart = nextKey._dataEnd+1;
2282 // now nextKey is either invalid or is outside of actual._dataEnd
2283 // output the remaining and discard myself
2284 if (oldStart <= end) {
2285 processRegion(oldStart, end);
2287 if (interval.par[end] == '}') {
2289 // This is the normal case.
2290 // But if using the firstlanguage, the closing may be missing
2292 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2294 if (actual._dataEnd < end)
2295 output_end = interval.nextNotIgnored(actual._dataEnd);
2297 output_end = interval.nextNotIgnored(end);
2298 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2299 interval.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2301 // Remove possible empty data
2302 int dstart = interval.nextNotIgnored(actual._dataStart);
2303 while ((dstart < output_end) && (interval.par[dstart] == '{')) {
2304 interval.addIntervall(dstart, dstart+1);
2305 int dend = interval.findclosing(dstart+1, output_end);
2306 interval.addIntervall(dend, dend+1);
2307 dstart = interval.nextNotIgnored(dstart+1);
2309 if (dstart < output_end)
2310 interval.output(os, output_end);
2311 interval.addIntervall(actual._tokenstart, end);
2315 string splitOnKnownMacros(string par, bool isPatternString)
2318 LatexInfo li(par, isPatternString);
2319 // LYXERR0("Berfore split: " << par);
2320 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2322 DummyKey._tokensize = 0;
2323 DummyKey._dataStart = 0;
2324 DummyKey._dataEnd = par.length();
2325 DummyKey.disabled = true;
2326 int firstkeyIdx = li.getFirstKey();
2328 if (firstkeyIdx >= 0) {
2329 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2330 DummyKey._tokenstart = firstKey._tokenstart;
2332 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2333 // Use dummy firstKey
2334 firstKey = DummyKey;
2335 (void) li.setNextKey(firstkeyIdx);
2338 if (par.substr(firstKey._dataStart, 2) == "% ")
2339 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2341 nextkeyIdx = li.process(os, firstKey);
2342 while (nextkeyIdx >= 0) {
2343 // Check for a possible gap between the last
2344 // entry and this one
2345 int datastart = li.nextNotIgnored(firstKey._dataStart);
2346 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2347 if ((nextKey._tokenstart > datastart)) {
2349 firstKey._dataStart = datastart;
2350 firstKey._dataEnd = par.length();
2351 (void) li.setNextKey(nextkeyIdx);
2352 if (firstKey._tokensize > 0) {
2353 // Fake the last opened parenthesis
2354 li.setForDefaultLang(firstKey._tokensize);
2356 nextkeyIdx = li.process(os, firstKey);
2359 if (nextKey.keytype != KeyInfo::isMain) {
2360 firstKey._dataStart = datastart;
2361 firstKey._dataEnd = nextKey._dataEnd+1;
2362 (void) li.setNextKey(nextkeyIdx);
2363 if (firstKey._tokensize > 0)
2364 li.setForDefaultLang(firstKey._tokensize);
2365 nextkeyIdx = li.process(os, firstKey);
2368 nextkeyIdx = li.process(os, nextKey);
2372 // Handle the remaining
2373 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2374 firstKey._dataEnd = par.length();
2376 if ((firstKey._dataStart < firstKey._dataEnd) &&
2377 (par[firstKey._dataStart] != '}')) {
2378 if (firstKey._tokensize > 0)
2379 li.setForDefaultLang(firstKey._tokensize);
2380 (void) li.process(os, firstKey);
2384 // return string definitelly impossible to match
2385 s = "\\foreignlanguage{ignore}{ }";
2389 s = par; /* no known macros found */
2390 // LYXERR0("After split: " << s);
2395 * Try to unify the language specs in the latexified text.
2396 * Resulting modified string is set to "", if
2397 * the searched tex does not contain all the features in the search pattern
2399 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2401 static Features regex_f;
2402 static int missed = 0;
2403 static bool regex_with_format = false;
2405 int parlen = par.length();
2407 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2410 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2411 // Happens to be there in case of description or labeling environment
2416 // Split the latex input into pieces which
2417 // can be digested by our search engine
2418 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2419 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2420 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2423 result = par.substr(0, parlen);
2424 if (isPatternString) {
2427 regex_f = identifyFeatures(result);
2428 string features = "";
2429 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2430 string a = it->first;
2431 regex_with_format = true;
2432 features += " " + a;
2433 // LYXERR0("Identified regex format:" << a);
2435 LYXERR(Debug::FIND, "Identified Features" << features);
2438 } else if (regex_with_format) {
2439 Features info = identifyFeatures(result);
2440 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2441 string a = it->first;
2442 bool b = it->second;
2443 if (b && ! info[a]) {
2445 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2451 // LYXERR0("No regex formats");
2457 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2458 static int identifyClosing(string & t)
2460 int open_braces = 0;
2462 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2463 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2465 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]" REGEX_EOS, "$1"))
2467 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2469 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2479 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2480 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2482 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2483 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2484 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2485 // When using regexp, braces are hacked already by escape_for_regex()
2486 par_as_string = normalize(ds, !use_regexp);
2488 close_wildcards = 0;
2490 size_t lead_size = 0;
2491 // correct the language settings
2492 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2493 if (opt.ignoreformat) {
2495 // if par_as_string_nolead were emty,
2496 // the following call to findAux will always *find* the string
2497 // in the checked data, and thus always using the slow
2498 // examining of the current text part.
2499 par_as_string_nolead = par_as_string;
2502 lead_size = identifyLeading(par_as_string);
2503 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2504 lead_as_string = par_as_string.substr(0, lead_size);
2505 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2509 open_braces = identifyClosing(par_as_string);
2510 identifyClosing(par_as_string_nolead);
2511 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2512 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2514 string lead_as_regexp;
2515 if (lead_size > 0) {
2516 // @todo No need to search for \regexp{} insets in leading material
2517 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2518 par_as_string = par_as_string_nolead;
2519 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2520 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2522 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2523 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2524 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2526 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2527 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2528 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2529 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2530 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2531 || regex_replace(par_as_string, par_as_string,
2532 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2533 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2534 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2538 if (!opt.ignoreformat) {
2539 // Remove extra '\}' at end
2540 while ( regex_replace(par_as_string, par_as_string, "(.*)\\\\}$", "$1")) {
2545 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2546 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2547 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2548 // replace '[^...]' with '[^...\}\{\\]'
2549 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2550 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2551 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2553 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2556 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2557 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2558 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2559 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2561 // If entered regexp must match at begin of searched string buffer
2562 // Kornel: Added parentheses to use $1 for size of the leading string
2566 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2567 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2568 // so the convert has no effect in that case
2569 for (int i = 8; i > 0; --i) {
2570 string orig = "\\\\" + std::to_string(i);
2571 string dest = "\\" + std::to_string(i+1);
2572 while (regex_replace(par_as_string, par_as_string, orig, dest));
2574 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2575 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2577 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2578 regexp = lyx::regex(regexp_str);
2580 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2581 regexp2 = lyx::regex(regexp2_str);
2586 // Count number of characters in string
2590 // \\[a-zA-Z]+ ==> 1
2591 static int computeSize(string s, int len)
2597 for (int i = 0; i < len; i += skip, count++) {
2600 if (isalpha(s[i+1])) {
2601 for (int j = 2; i+j < len; j++) {
2602 if (! isalpha(s[i+j])) {
2605 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2607 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2615 else if (s[i] == '{') {
2621 else if (s[i] == '-') {
2622 if (s[i+1] == '-') {
2638 int MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2641 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2644 docstring docstr = stringifyFromForSearch(opt, cur, len);
2645 string str = normalize(docstr, true);
2646 if (!opt.ignoreformat) {
2647 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2649 if (str.empty()) return(-1);
2650 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2651 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2654 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2655 regex const *p_regexp;
2656 regex_constants::match_flag_type flags;
2658 flags = regex_constants::match_continuous;
2661 flags = regex_constants::match_default;
2662 p_regexp = ®exp2;
2664 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2665 if (re_it == sregex_iterator())
2667 match_results<string::const_iterator> const & m = *re_it;
2669 if (0) { // Kornel Benko: DO NOT CHECKK
2670 // Check braces on the segment that matched the entire regexp expression,
2671 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2672 if (!braces_match(m[0].first, m[0].second, open_braces))
2676 // Check braces on segments that matched all (.*?) subexpressions,
2677 // except the last "padding" one inserted by lyx.
2678 for (size_t i = 1; i < m.size() - 1; ++i)
2679 if (!braces_match(m[i].first, m[i].second, open_braces))
2682 // Exclude from the returned match length any length
2683 // due to close wildcards added at end of regexp
2684 // and also the length of the leading (e.g. '\emph{')
2686 // Whole found string, including the leading: m[0].second - m[0].first
2687 // Size of the leading string: m[1].second - m[1].first
2688 int leadingsize = 0;
2690 leadingsize = m[1].second - m[1].first;
2692 for (size_t i = 0; i < m.size(); i++) {
2693 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2695 if (close_wildcards == 0)
2696 result = m[0].second - m[0].first;
2699 result = m[m.size() - close_wildcards].first - m[0].first;
2701 size_t pos = m.position(size_t(0));
2702 // Ignore last closing characters
2703 while (result > 0) {
2704 if (str[pos+result-1] == '}')
2709 if (result > leadingsize)
2710 result -= leadingsize;
2713 return computeSize(str.substr(pos+leadingsize,result), result);
2716 // else !use_regexp: but all code paths above return
2717 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2718 << par_as_string << "', str='" << str << "'");
2719 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2720 << lead_as_string << "', par_as_string_nolead='"
2721 << par_as_string_nolead << "'");
2724 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2725 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2726 if (str.substr(0, par_as_string.size()) == par_as_string)
2727 return par_as_string.size();
2729 size_t pos = str.find(par_as_string_nolead);
2730 if (pos != string::npos)
2731 return par_as_string.size();
2737 int MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
2739 int res = findAux(cur, len, at_begin);
2741 "res=" << res << ", at_begin=" << at_begin
2742 << ", matchword=" << opt.matchword
2743 << ", inTexted=" << cur.inTexted());
2744 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
2746 if ((len > 0) && (res < len))
2748 Paragraph const & par = cur.paragraph();
2749 bool ws_left = (cur.pos() > 0)
2750 ? par.isWordSeparator(cur.pos() - 1)
2752 bool ws_right = (cur.pos() + len < par.size())
2753 ? par.isWordSeparator(cur.pos() + len)
2756 "cur.pos()=" << cur.pos() << ", res=" << res
2757 << ", separ: " << ws_left << ", " << ws_right
2760 if (ws_left && ws_right) {
2761 // Check for word separators inside the found 'word'
2762 for (int i = 0; i < len; i++) {
2763 if (par.isWordSeparator(cur.pos() + i))
2772 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
2775 if (! opt.casesensitive)
2776 t = lyx::to_utf8(lowercase(s));
2778 t = lyx::to_utf8(s);
2779 // Remove \n at begin
2780 while (!t.empty() && t[0] == '\n')
2783 while (!t.empty() && t[t.size() - 1] == '\n')
2784 t = t.substr(0, t.size() - 1);
2786 // Handle all other '\n'
2787 while ((pos = t.find("\n")) != string::npos) {
2788 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
2790 if (std::isalnum(t[pos+1])) {
2791 t.replace(pos-2, 3, " ");
2794 t.replace(pos-2, 3, "");
2797 else if (!std::isalnum(t[pos+1]) || !std::isalnum(t[pos-1])) {
2798 // '\n' adjacent to non-alpha-numerics, discard
2799 t.replace(pos, 1, "");
2802 // Replace all other \n with spaces
2803 t.replace(pos, 1, " ");
2806 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
2807 // Kornel: Added textsl, textsf, textit, texttt and noun
2808 // + allow to seach for colored text too
2809 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
2810 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
2811 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2812 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
2813 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2815 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
2816 // FIXME - check what preceeds the brace
2818 if (opt.ignoreformat)
2819 while (regex_replace(t, t, "\\{", "_x_<")
2820 || regex_replace(t, t, "\\}", "_x_>"))
2821 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2823 while (regex_replace(t, t, "\\\\\\{", "_x_<")
2824 || regex_replace(t, t, "\\\\\\}", "_x_>"))
2825 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2832 docstring stringifyFromCursor(DocIterator const & cur, int len)
2834 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
2835 if (cur.inTexted()) {
2836 Paragraph const & par = cur.paragraph();
2837 // TODO what about searching beyond/across paragraph breaks ?
2838 // TODO Try adding a AS_STR_INSERTS as last arg
2839 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
2840 int(par.size()) : cur.pos() + len;
2841 OutputParams runparams(&cur.buffer()->params().encoding());
2842 runparams.nice = true;
2843 runparams.flavor = OutputParams::LATEX;
2844 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
2845 // No side effect of file copying and image conversion
2846 runparams.dryrun = true;
2847 LYXERR(Debug::FIND, "Stringifying with cur: "
2848 << cur << ", from pos: " << cur.pos() << ", end: " << end);
2849 return par.asString(cur.pos(), end,
2850 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
2852 } else if (cur.inMathed()) {
2853 CursorSlice cs = cur.top();
2854 MathData md = cs.cell();
2855 MathData::const_iterator it_end =
2856 (( len == -1 || cs.pos() + len > int(md.size()))
2858 : md.begin() + cs.pos() + len );
2860 for (MathData::const_iterator it = md.begin() + cs.pos();
2863 docstring s = asString(md2);
2864 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
2867 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
2872 /** Computes the LaTeX export of buf starting from cur and ending len positions
2873 * after cur, if len is positive, or at the paragraph or innermost inset end
2876 docstring latexifyFromCursor(DocIterator const & cur, int len)
2878 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
2879 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
2880 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
2881 Buffer const & buf = *cur.buffer();
2883 odocstringstream ods;
2885 OutputParams runparams(&buf.params().encoding());
2886 runparams.nice = false;
2887 runparams.flavor = OutputParams::LATEX;
2888 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
2889 // No side effect of file copying and image conversion
2890 runparams.dryrun = true;
2891 runparams.for_search = true;
2893 if (cur.inTexted()) {
2894 // @TODO what about searching beyond/across paragraph breaks ?
2895 pos_type endpos = cur.paragraph().size();
2896 if (len != -1 && endpos > cur.pos() + len)
2897 endpos = cur.pos() + len;
2898 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
2899 string(), cur.pos(), endpos);
2900 string s = lyx::to_utf8(ods.str());
2901 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
2902 return(lyx::from_utf8(s));
2903 } else if (cur.inMathed()) {
2904 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
2905 for (int s = cur.depth() - 1; s >= 0; --s) {
2906 CursorSlice const & cs = cur[s];
2907 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
2909 cs.asInsetMath()->asHullInset()->header_write(ws);
2914 CursorSlice const & cs = cur.top();
2915 MathData md = cs.cell();
2916 MathData::const_iterator it_end =
2917 ((len == -1 || cs.pos() + len > int(md.size()))
2919 : md.begin() + cs.pos() + len);
2921 for (MathData::const_iterator it = md.begin() + cs.pos();
2925 ods << asString(md2);
2926 // Retrieve the math environment type, and add '$' or '$]'
2927 // or others (\end{equation}) accordingly
2928 for (int s = cur.depth() - 1; s >= 0; --s) {
2929 CursorSlice const & cs2 = cur[s];
2930 InsetMath * inset = cs2.asInsetMath();
2931 if (inset && inset->asHullInset()) {
2933 inset->asHullInset()->footer_write(ws);
2937 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
2939 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
2945 /** Finalize an advanced find operation, advancing the cursor to the innermost
2946 ** position that matches, plus computing the length of the matching text to
2949 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
2951 // Search the foremost position that matches (avoids find of entire math
2952 // inset when match at start of it)
2954 DocIterator old_cur(cur.buffer());
2956 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
2960 } while (cur && cur.depth() > d && match(cur) > 0);
2962 int max_match = match(cur); /* match valid only if not searching whole words */
2963 if (max_match <= 0) return 0;
2964 LYXERR(Debug::FIND, "Ok");
2966 // Compute the match length
2968 if (cur.pos() + len > cur.lastpos())
2970 if (match.opt.matchword) {
2971 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
2972 while (cur.pos() + len <= cur.lastpos() && match(cur, len) <= 0) {
2974 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
2976 // Length of matched text (different from len param)
2977 int old_match = match(cur, len);
2981 // Greedy behaviour while matching regexps
2982 while ((new_match = match(cur, len + 1)) > old_match) {
2984 old_match = new_match;
2985 LYXERR(Debug::FIND, "verifying match with len = " << len);
2992 int maxl = cur.lastpos() - cur.pos();
2993 // Greedy behaviour while matching regexps
2994 while (maxl > minl) {
2995 int actual_match = match(cur, len);
2996 if (actual_match >= max_match) {
2997 // actual_match > max_match _can_ happen,
2998 // if the search area splits
2999 // some following word so that the regex
3000 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3002 // This means, the len value is too big
3004 len = (int)((maxl + minl)/2);
3007 // (actual_match < max_match)
3009 len = (int)((maxl + minl)/2);
3013 // Search for real start of matched characters
3018 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3019 if (cur.depth() < old_cur.depth()) {
3021 LYXERR0("cur.depth() < old_cur.depth(), this should never happen");
3024 if (cur.pos() != old_cur.pos()) {
3025 // OK, forwarded 1 pos in actual inset
3026 actual_match = match(cur, len-1);
3027 if (actual_match == max_match) {
3028 // Ha, got it! The shorter selection has the same match length
3033 // OK, the shorter selection matches less chars, revert to previous value
3039 LYXERR0("cur.pos() == old_cur.pos(), this should never happen");
3040 actual_match = match(cur, len);
3041 if (actual_match == max_match)
3051 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3055 while (!theApp()->longOperationCancelled() && cur) {
3056 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3057 int match_len = match(cur, -1, false);
3058 LYXERR(Debug::FIND, "match_len: " << match_len);
3059 if (match_len > 0) {
3060 int match_len_zero_count = 0;
3061 for (; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3062 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3063 int match_len3 = match(cur, 1);
3066 int match_len2 = match(cur);
3067 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3068 if (match_len2 > 0) {
3069 // Sometimes in finalize we understand it wasn't a match
3070 // and we need to continue the outest loop
3071 int len = findAdvFinalize(cur, match);
3076 if (match_len2 >= 0) {
3077 if (match_len2 == 0)
3078 match_len_zero_count++;
3080 match_len_zero_count = 0;
3083 if (++match_len_zero_count > 3) {
3084 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3085 match_len_zero_count = 0;
3093 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3094 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3097 // This should exit nested insets, if any, or otherwise undefine the currsor.
3098 cur.pos() = cur.lastpos();
3099 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3107 /// Find the most backward consecutive match within same paragraph while searching backwards.
3108 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3110 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3111 DocIterator tmp_cur = cur;
3112 int len = findAdvFinalize(tmp_cur, match);
3113 Inset & inset = cur.inset();
3114 for (; cur != cur_begin; cur.backwardPos()) {
3115 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3116 DocIterator new_cur = cur;
3117 new_cur.backwardPos();
3118 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur))
3120 int new_len = findAdvFinalize(new_cur, match);
3125 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3131 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3135 // Backup of original position
3136 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3137 if (cur == cur_begin)
3140 DocIterator cur_orig(cur);
3141 bool pit_changed = false;
3144 bool found_match = match(cur, -1, false);
3148 cur.pos() = cur.lastpos();
3150 cur.pos() = cur_orig.pos();
3151 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3152 DocIterator cur_prev_iter;
3154 found_match = match(cur);
3155 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3156 << found_match << ", cur: " << cur);
3158 return findMostBackwards(cur, match);
3160 // Stop if begin of document reached
3161 if (cur == cur_begin)
3163 cur_prev_iter = cur;
3167 if (cur == cur_begin)
3174 } while (!theApp()->longOperationCancelled());
3182 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3183 DocIterator const & cur, int len)
3185 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3187 if (!opt.ignoreformat)
3188 return latexifyFromCursor(cur, len);
3190 return stringifyFromCursor(cur, len);
3194 FindAndReplaceOptions::FindAndReplaceOptions(
3195 docstring const & find_buf_name, bool casesensitive,
3196 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
3197 docstring const & repl_buf_name, bool keep_case,
3198 SearchScope scope, SearchRestriction restr)
3199 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
3200 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
3201 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope), restr(restr)
3209 /** Check if 'len' letters following cursor are all non-lowercase */
3210 static bool allNonLowercase(Cursor const & cur, int len)
3212 pos_type beg_pos = cur.selectionBegin().pos();
3213 pos_type end_pos = cur.selectionBegin().pos() + len;
3214 if (len > cur.lastpos() + 1 - beg_pos) {
3215 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3216 len = cur.lastpos() + 1 - beg_pos;
3217 end_pos = beg_pos + len;
3219 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3220 if (isLowerCase(cur.paragraph().getChar(pos)))
3226 /** Check if first letter is upper case and second one is lower case */
3227 static bool firstUppercase(Cursor const & cur)
3230 pos_type pos = cur.selectionBegin().pos();
3231 if (pos >= cur.lastpos() - 1) {
3232 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3235 ch1 = cur.paragraph().getChar(pos);
3236 ch2 = cur.paragraph().getChar(pos + 1);
3237 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3238 LYXERR(Debug::FIND, "firstUppercase(): "
3239 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3240 << ch2 << "(" << char(ch2) << ")"
3241 << ", result=" << result << ", cur=" << cur);
3246 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3248 ** \fixme What to do with possible further paragraphs in replace buffer ?
3250 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3252 ParagraphList::iterator pit = buffer.paragraphs().begin();
3253 LASSERT(pit->size() >= 1, /**/);
3254 pos_type right = pos_type(1);
3255 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3256 right = pit->size();
3257 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3263 static void findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3265 Cursor & cur = bv->cursor();
3266 if (opt.repl_buf_name == docstring()
3267 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3268 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3271 DocIterator sel_beg = cur.selectionBegin();
3272 DocIterator sel_end = cur.selectionEnd();
3273 if (&sel_beg.inset() != &sel_end.inset()
3274 || sel_beg.pit() != sel_end.pit()
3275 || sel_beg.idx() != sel_end.idx())
3277 int sel_len = sel_end.pos() - sel_beg.pos();
3278 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3279 << ", sel_len: " << sel_len << endl);
3282 LASSERT(sel_len > 0, return);
3284 if (!matchAdv(sel_beg, sel_len))
3287 // Build a copy of the replace buffer, adapted to the KeepCase option
3288 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3290 repl_buffer_orig.write(oss);
3291 string lyx = oss.str();
3292 Buffer repl_buffer("", false);
3293 repl_buffer.setUnnamed(true);
3294 LASSERT(repl_buffer.readString(lyx), return);
3295 if (opt.keep_case && sel_len >= 2) {
3296 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3297 if (cur.inTexted()) {
3298 if (firstUppercase(cur))
3299 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3300 else if (allNonLowercase(cur, sel_len))
3301 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3304 cap::cutSelection(cur, false);
3305 if (cur.inTexted()) {
3306 repl_buffer.changeLanguage(
3307 repl_buffer.language(),
3308 cur.getFont().language());
3309 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3310 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3311 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3312 repl_buffer.params().documentClassPtr(),
3313 bv->buffer().errorList("Paste"));
3314 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3315 sel_len = repl_buffer.paragraphs().begin()->size();
3316 } else if (cur.inMathed()) {
3317 odocstringstream ods;
3319 OutputParams runparams(&repl_buffer.params().encoding());
3320 runparams.nice = false;
3321 runparams.flavor = OutputParams::LATEX;
3322 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3323 runparams.dryrun = true;
3324 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3325 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3326 docstring repl_latex = ods.str();
3327 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3329 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3330 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3331 repl_latex = from_utf8(s);
3332 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3333 MathData ar(cur.buffer());
3334 asArray(repl_latex, ar, Parse::NORMAL);
3336 sel_len = ar.size();
3337 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3339 if (cur.pos() >= sel_len)
3340 cur.pos() -= sel_len;
3343 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3344 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3345 bv->processUpdateFlags(Update::Force);
3349 /// Perform a FindAdv operation.
3350 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3355 // e.g., when invoking word-findadv from mini-buffer wither with
3356 // wrong options syntax or before ever opening advanced F&R pane
3357 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3361 MatchStringAdv matchAdv(bv->buffer(), opt);
3362 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3364 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3365 findAdvReplace(bv, opt, matchAdv);
3368 match_len = findForwardAdv(cur, matchAdv);
3370 match_len = findBackwardsAdv(cur, matchAdv);
3372 // This may only be raised by lyx::regex()
3373 bv->message(_("Invalid regular expression!"));
3377 if (match_len == 0) {
3378 bv->message(_("Match not found!"));
3382 bv->message(_("Match found!"));
3384 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3385 bv->putSelectionAt(cur, match_len, !opt.forward);
3391 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3393 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3394 << opt.casesensitive << ' '
3395 << opt.matchword << ' '
3396 << opt.forward << ' '
3397 << opt.expandmacros << ' '
3398 << opt.ignoreformat << ' '
3399 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3400 << opt.keep_case << ' '
3401 << int(opt.scope) << ' '
3404 LYXERR(Debug::FIND, "built: " << os.str());
3410 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3412 LYXERR(Debug::FIND, "parsing");
3416 while (line != "EOSS") {
3420 if (is.eof()) // Tolerate malformed request
3424 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3425 opt.find_buf_name = from_utf8(s);
3426 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat;
3427 is.get(); // Waste space before replace string
3430 while (line != "EOSS") {
3434 if (is.eof()) // Tolerate malformed request
3438 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3439 opt.repl_buf_name = from_utf8(s);
3440 is >> opt.keep_case;
3443 opt.scope = FindAndReplaceOptions::SearchScope(i);
3445 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3447 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3448 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3449 << opt.scope << ' ' << opt.restr);