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)
744 LYXERR(Debug::FIND, "new_pos: " << new_pos);
745 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
746 LYXERR(Debug::FIND, "t [lyx]: " << t);
747 t = apply_escapes(t, get_regexp_escapes());
748 LYXERR(Debug::FIND, "t [rxp]: " << t);
749 s.replace(pos, new_pos - pos, t);
750 new_pos = pos + t.size();
751 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
752 LYXERR(Debug::FIND, "new_pos: " << new_pos);
753 if (new_pos == s.size())
756 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
757 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
758 LYXERR(Debug::FIND, "end_pos: " << end_pos);
759 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
760 LYXERR(Debug::FIND, "t in regexp : " << t);
761 t = apply_escapes(t, get_lyx_unescapes());
762 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
764 t = apply_escapes(t, get_regexp_latex_escapes());
765 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
767 if (end_pos == s.size()) {
768 s.replace(new_pos, end_pos - new_pos, t);
769 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
772 s.replace(new_pos, end_pos + 13 - new_pos, t);
773 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
774 pos = new_pos + t.size();
775 LYXERR(Debug::FIND, "pos: " << pos);
781 /// Wrapper for lyx::regex_replace with simpler interface
782 bool regex_replace(string const & s, string & t, string const & searchstr,
783 string const & replacestr)
785 lyx::regex e(searchstr, regex_constants::ECMAScript);
787 ostream_iterator<char, char> it(oss);
788 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
789 // tolerate t and s be references to the same variable
790 bool rv = (s != oss.str());
796 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
798 ** Verify that closed braces exactly match open braces. This avoids that, for example,
799 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
802 ** Number of open braces that must remain open at the end for the verification to succeed.
804 bool braces_match(string::const_iterator const & beg,
805 string::const_iterator const & end,
809 string::const_iterator it = beg;
810 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
811 for (; it != end; ++it) {
812 // Skip escaped braces in the count
817 } else if (*it == '{') {
819 } else if (*it == '}') {
820 if (open_pars == 0) {
821 LYXERR(Debug::FIND, "Found unmatched closed brace");
827 if (open_pars != unmatched) {
828 LYXERR(Debug::FIND, "Found " << open_pars
829 << " instead of " << unmatched
830 << " unmatched open braces at the end of count");
833 LYXERR(Debug::FIND, "Braces match as expected");
838 /** The class performing a match between a position in the document and the FindAdvOptions.
840 class MatchStringAdv {
842 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
844 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
845 ** constructor as opt.search, under the opt.* options settings.
848 ** If set, then match is searched only against beginning of text starting at cur.
849 ** If unset, then match is searched anywhere in text starting at cur.
852 ** The length of the matching text, or zero if no match was found.
854 int operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
859 /// first buffer on which search was started
860 lyx::Buffer * const p_first_buf;
862 FindAndReplaceOptions const & opt;
865 /// Auxiliary find method (does not account for opt.matchword)
866 int findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
868 /** Normalize a stringified or latexified LyX paragraph.
872 ** <li>if search is not casesensitive, then lowercase the string;
873 ** <li>remove any newline at begin or end of the string;
874 ** <li>replace any newline in the middle of the string with a simple space;
875 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
878 ** @todo Normalization should also expand macros, if the corresponding
879 ** search option was checked.
881 string normalize(docstring const & s, bool hack_braces) const;
882 // normalized string to search
883 string par_as_string;
884 // regular expression to use for searching
886 // same as regexp, but prefixed with a ".*?"
888 // leading format material as string
889 string lead_as_string;
890 // par_as_string after removal of lead_as_string
891 string par_as_string_nolead;
892 // unmatched open braces in the search string/regexp
894 // number of (.*?) subexpressions added at end of search regexp for closing
895 // environments, math mode, styles, etc...
897 // Are we searching with regular expressions ?
902 static docstring buffer_to_latex(Buffer & buffer)
904 OutputParams runparams(&buffer.params().encoding());
905 odocstringstream ods;
907 runparams.nice = true;
908 runparams.flavor = OutputParams::LATEX;
909 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
910 // No side effect of file copying and image conversion
911 runparams.dryrun = true;
912 runparams.for_search = true;
913 pit_type const endpit = buffer.paragraphs().size();
914 for (pit_type pit = 0; pit != endpit; ++pit) {
915 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
916 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
922 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
925 if (!opt.ignoreformat) {
926 str = buffer_to_latex(buffer);
928 OutputParams runparams(&buffer.params().encoding());
929 runparams.nice = true;
930 runparams.flavor = OutputParams::LATEX;
931 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
932 runparams.dryrun = true;
933 runparams.for_search = true;
934 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
935 Paragraph const & par = buffer.paragraphs().at(pit);
936 LYXERR(Debug::FIND, "Adding to search string: '"
937 << par.asString(pos_type(0), par.size(),
938 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
941 str += par.asString(pos_type(0), par.size(),
942 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
950 /// Return separation pos between the leading material and the rest
951 static size_t identifyLeading(string const & s)
954 // @TODO Support \item[text]
955 // Kornel: Added textsl, textsf, textit, texttt and noun
956 // + allow to search for colored text too
957 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)\\*?)\\{", "")
958 || regex_replace(t, t, REGEX_BOS "\\$", "")
959 || regex_replace(t, t, REGEX_BOS "\\\\\\[ ", "")
960 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
961 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
963 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
968 * Given a latexified string, retrieve some handled features
969 * The features of the regex will later be compared with the features
970 * of the searched text. If the regex features are not a
971 * subset of the analized, then, in not format ignoring search
972 * we can early stop the search in the relevant inset.
974 typedef map<string, bool> Features;
976 static Features identifyFeatures(string const & s)
978 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
979 static regex const valid("^("
981 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
982 "emph|noun|text(bf|md|sl|sf|it|tt)|"
983 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
984 "(u|uu)line|(s|x)out|uwave|"
985 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
986 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
991 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
994 if (sub.str(1).compare("regexp") == 0) {
998 string token = sub.str(1);
1000 if (regex_match(token, sub2, valid)) {
1008 if (sub.str(1).compare("endregexp") == 0) {
1018 * defines values features of a key "\\[a-z]+{"
1023 /* Char type with content discarded
1024 * like \hspace{1cm} */
1026 /* Char, like \backslash */
1028 /* \part, \section*, ... */
1030 /* title, author etc */
1032 /* \foreignlanguage{ngerman}, ... */
1035 * to discard language in content */
1038 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1040 /* fonts, colors, markups, ... */
1042 /* footnotesize, ... large, ...
1043 * Ignore all of them */
1046 /* inputencoding, ...
1047 * Discard also content, because they do not help in search */
1050 * like remove, but also all arguments */
1052 /* item, listitem */
1054 /* tex, latex, ... like isChar */
1056 /* like \lettrine[lines=5]{}{} */
1058 /* End of arguments marker for lettrine,
1059 * so that they can be ignored */
1069 parenthesiscount(1),
1073 KeyInfo(KeyType type, int parcount, bool disable)
1079 parenthesiscount(parcount),
1088 int parenthesiscount;
1090 bool used; /* by pattern */
1095 Border(int l=0, int u=0) : low(l), upper(u) {};
1100 #define MAXOPENED 30
1102 bool isPatternString;
1104 explicit Intervall(bool isPattern) :
1105 isPatternString(isPattern),
1107 actualdeptindex(0) { depts[0] = 0; closes[0] = 0;};
1110 int depts[MAXOPENED];
1111 int closes[MAXOPENED];
1112 int actualdeptindex;
1113 Border borders[2*MAXOPENED];
1114 int previousNotIgnored(int);
1115 int nextNotIgnored(int);
1116 void handleOpenP(int i);
1117 void handleCloseP(int i, bool closingAllowed);
1118 void resetOpenedP(int openPos);
1119 void addIntervall(int upper);
1120 void addIntervall(int low, int upper); /* if explicit */
1121 void setForDefaultLang(KeyInfo &defLang);
1122 int findclosing(int start, int end, char up, char down, int repeat);
1123 void handleParentheses(int lastpos, bool closingAllowed);
1125 int isOpeningPar(int pos);
1127 void output(ostringstream &os, int lastpos);
1128 // string show(int lastpos);
1131 int Intervall::isOpeningPar(int pos)
1133 if ((pos < 0) || (size_t(pos) >= par.size()))
1135 if (par[pos] != '{')
1137 if (size_t(pos) + 2 >= par.size())
1139 if (par[pos+2] != '}')
1141 if (par[pos+1] == '[' || par[pos+1] == ']')
1146 void Intervall::setForDefaultLang(KeyInfo &defLang)
1148 // Enable the use of first token again
1149 if (ignoreidx >= 0) {
1150 int value = defLang._tokenstart + defLang._tokensize;
1152 if (borders[0].low < value)
1153 borders[0].low = value;
1154 if (borders[0].upper < value)
1155 borders[0].upper = value;
1160 static void checkDepthIndex(int val)
1162 static int maxdepthidx = MAXOPENED-2;
1163 if (val > maxdepthidx) {
1165 LYXERR0("maxdepthidx now " << val);
1169 static void checkIgnoreIdx(int val)
1171 static int maxignoreidx = 2*MAXOPENED - 4;
1172 if (val > maxignoreidx) {
1174 LYXERR0("maxignoreidx now " << val);
1179 * Expand the region of ignored parts of the input latex string
1180 * The region is only relevant in output()
1182 void Intervall::addIntervall(int low, int upper)
1185 if (low == upper) return;
1186 for (idx = ignoreidx+1; idx > 0; --idx) {
1187 if (low > borders[idx-1].upper) {
1191 Border br(low, upper);
1192 if (idx > ignoreidx) {
1195 checkIgnoreIdx(ignoreidx);
1199 // Expand only if one of the new bound is inside the interwall
1200 // We know here that br.low > borders[idx-1].upper
1201 if (br.upper < borders[idx].low) {
1202 // We have to insert at this pos
1203 for (int i = ignoreidx+1; i > idx; --i) {
1204 borders[i] = borders[i-1];
1208 checkIgnoreIdx(ignoreidx);
1211 // Here we know, that we are overlapping
1212 if (br.low > borders[idx].low)
1213 br.low = borders[idx].low;
1214 // check what has to be concatenated
1216 for (int i = idx; i <= ignoreidx; i++) {
1217 if (br.upper >= borders[i].low) {
1219 if (br.upper < borders[i].upper)
1220 br.upper = borders[i].upper;
1226 // count should be >= 1 here
1229 for (int i = idx + count; i <= ignoreidx; i++) {
1230 borders[i-count+1] = borders[i];
1232 ignoreidx -= count - 1;
1238 void Intervall::handleOpenP(int i)
1241 depts[actualdeptindex] = i+1;
1242 closes[actualdeptindex] = -1;
1243 checkDepthIndex(actualdeptindex);
1246 void Intervall::handleCloseP(int i, bool closingAllowed)
1248 if (actualdeptindex <= 0) {
1249 if (! closingAllowed)
1250 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1251 // if we are at the very end
1252 addIntervall(i, i+1);
1255 closes[actualdeptindex] = i+1;
1260 void Intervall::resetOpenedP(int openPos)
1262 // Used as initializer for foreignlanguage entry
1263 actualdeptindex = 1;
1264 depts[1] = openPos+1;
1268 int Intervall::previousNotIgnored(int start)
1270 int idx = 0; /* int intervalls */
1271 for (idx = ignoreidx; idx >= 0; --idx) {
1272 if (start > borders[idx].upper)
1274 if (start >= borders[idx].low)
1275 start = borders[idx].low-1;
1280 int Intervall::nextNotIgnored(int start)
1282 int idx = 0; /* int intervalls */
1283 for (idx = 0; idx <= ignoreidx; idx++) {
1284 if (start < borders[idx].low)
1286 if (start < borders[idx].upper)
1287 start = borders[idx].upper;
1292 typedef map<string, KeyInfo> KeysMap;
1293 typedef vector< KeyInfo> Entries;
1294 static KeysMap keys = map<string, KeyInfo>();
1301 void buildKeys(bool);
1302 void buildEntries(bool);
1303 void makeKey(const string &, KeyInfo, bool isPatternString);
1304 void processRegion(int start, int region_end); /* remove {} parts */
1305 void removeHead(KeyInfo&, int count=0);
1308 LatexInfo(string par, bool isPatternString) : entidx(-1), interval(isPatternString) {
1310 interval.hasTitle = false;
1311 interval.titleValue = "";
1312 buildKeys(isPatternString);
1313 entries = vector<KeyInfo>();
1314 buildEntries(isPatternString);
1318 if (entries.empty()) {
1321 if (entries[0].keytype == KeyInfo::isTitle) {
1322 if (! entries[0].disabled) {
1323 interval.hasTitle = true;
1324 interval.titleValue = entries[0].head;
1327 interval.hasTitle = false;
1328 interval.titleValue = "";
1330 removeHead(entries[0]);
1331 if (entries.size() > 1)
1340 if (int(entries.size()) > entidx) {
1347 bool setNextKey(int idx) {
1348 if ((idx == entidx) && (entidx >= 0)) {
1355 int find(int start, KeyInfo::KeyType keytype) {
1359 while (tmpIdx < int(entries.size())) {
1360 if (entries[tmpIdx].keytype == keytype)
1366 int process(ostringstream &os, KeyInfo &actual);
1367 int dispatch(ostringstream &os, int previousStart, KeyInfo &actual);
1368 // string show(int lastpos) { return interval.show(lastpos);};
1369 int nextNotIgnored(int start) { return interval.nextNotIgnored(start);};
1370 KeyInfo &getKeyInfo(int keyinfo) {
1371 static KeyInfo invalidInfo = KeyInfo();
1372 if ((keyinfo < 0) || ( keyinfo >= int(entries.size())))
1375 return entries[keyinfo];
1377 void setForDefaultLang(KeyInfo &defLang) {interval.setForDefaultLang(defLang);};
1378 void addIntervall(int low, int up) { interval.addIntervall(low, up); };
1382 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1387 for (int i = start; i < end; i += 1 + skip) {
1391 if (c == '\\') skip = 1;
1395 else if (c == down) {
1397 if ((repeat <= 0) || (par[i+1] != up))
1415 vector<MathEntry> entries;
1420 void insert(string wait, size_t start, size_t end) {
1421 MathEntry m = MathEntry();
1423 m.mathStart = start;
1425 m.mathSize = end - start;
1426 entries.push_back(m);
1428 bool empty() { return entries.empty(); };
1429 size_t getEndPos() {
1430 if (entries.empty() || (actualIdx >= entries.size())) {
1433 return entries[actualIdx].mathEnd;
1435 size_t getStartPos() {
1436 if (entries.empty() || (actualIdx >= entries.size())) {
1437 return 100000; /* definitely enough? */
1439 return entries[actualIdx].mathStart;
1441 size_t getFirstPos() {
1443 return getStartPos();
1446 if (entries.empty() || (actualIdx >= entries.size())) {
1449 return entries[actualIdx].mathSize;
1451 void incrEntry() { actualIdx++; };
1454 void LatexInfo::buildEntries(bool isPatternString)
1456 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1457 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1458 static bool disableLanguageOverride = false;
1459 smatch sub, submath;
1460 bool evaluatingRegexp = false;
1462 bool evaluatingMath = false;
1463 bool evaluatingCode = false;
1465 bool evaluatingOptional = false;
1466 size_t optionalEnd = 0;
1469 bool math_end_waiting = false;
1470 size_t math_pos = 10000;
1473 for (sregex_iterator itmath(interval.par.begin(), interval.par.end(), rmath), end; itmath != end; ++itmath) {
1475 if (math_end_waiting) {
1476 size_t pos = submath.position(size_t(0));
1477 if ((math_end == "$") &&
1478 (submath.str(0) == "$") &&
1479 (interval.par[pos-1] != '\\')) {
1480 mi.insert("$", math_pos, pos + 1);
1481 math_end_waiting = false;
1483 else if ((math_end == "\\]") &&
1484 (submath.str(0) == "\\]")) {
1485 mi.insert("\\]", math_pos, pos + 2);
1486 math_end_waiting = false;
1488 else if ((submath.str(1).compare("end") == 0) &&
1489 (submath.str(2).compare(math_end) == 0)) {
1490 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1491 math_end_waiting = false;
1497 if (submath.str(1).compare("begin") == 0) {
1498 math_end_waiting = true;
1499 math_end = submath.str(2);
1500 math_pos = submath.position(size_t(0));
1502 else if (submath.str(0).compare("\\[") == 0) {
1503 math_end_waiting = true;
1505 math_pos = submath.position(size_t(0));
1507 else if (submath.str(0) == "$") {
1508 size_t pos = submath.position(size_t(0));
1509 if ((pos == 0) || (interval.par[pos-1] != '\\')) {
1510 math_end_waiting = true;
1517 // Ignore language if there is math somewhere in pattern-string
1518 if (isPatternString) {
1521 keys["foreignlanguage"].disabled = true;
1522 disableLanguageOverride = true;
1525 disableLanguageOverride = false;
1528 if (disableLanguageOverride) {
1529 keys["foreignlanguage"].disabled = true;
1532 math_pos = mi.getFirstPos();
1533 for (sregex_iterator it(interval.par.begin(), interval.par.end(), rkeys), end; it != end; ++it) {
1535 string key = sub.str(3);
1537 if (sub.str(0)[0] == '\\')
1538 key = sub.str(0)[1];
1542 size_t k_pos = sub.position(size_t(0));
1543 if ((k_pos > 0) && (interval.par[k_pos - 1] == '\\')) {
1544 // Escaped '$', ignoring
1550 if (evaluatingRegexp) {
1551 if (sub.str(1).compare("endregexp") == 0) {
1552 evaluatingRegexp = false;
1553 // found._tokenstart already set
1554 found._dataEnd = sub.position(size_t(0)) + 13;
1555 found._dataStart = found._dataEnd;
1556 found._tokensize = found._dataEnd - found._tokenstart;
1557 found.parenthesiscount = 0;
1558 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1564 if (evaluatingMath) {
1565 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1567 evaluatingMath = false;
1569 math_pos = mi.getStartPos();
1571 if (keys.find(key) == keys.end()) {
1572 found = KeyInfo(KeyInfo::isStandard, 0, true);
1573 if (isPatternString) {
1574 found.keytype = KeyInfo::isChar;
1575 found.disabled = false;
1582 if (key.compare("regexp") == 0) {
1583 evaluatingRegexp = true;
1584 found._tokenstart = sub.position(size_t(0));
1585 found._tokensize = 0;
1589 // Handle the other params of key
1590 if (found.keytype == KeyInfo::isIgnored)
1592 else if (found.keytype == KeyInfo::isMath) {
1593 if (size_t(sub.position(size_t(0))) == math_pos) {
1595 found._tokenstart = sub.position(size_t(0));
1596 found._tokensize = mi.getSize();
1597 found._dataEnd = found._tokenstart + found._tokensize;
1598 found._dataStart = found._dataEnd;
1599 found.parenthesiscount = 0;
1600 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1601 evaluatingMath = true;
1604 // begin|end of unknown env, discard
1605 // First handle tables
1606 // longtable|tabular
1607 bool discardComment;
1609 found.keytype = KeyInfo::doRemove;
1610 if ((sub.str(5).compare("longtable") == 0) ||
1611 (sub.str(5).compare("tabular") == 0)) {
1612 discardComment = true; /* '%' */
1615 discardComment = false;
1616 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1618 string token = sub.str(5);
1619 if (regex_match(token, sub2, removeArgs)) {
1620 found.keytype = KeyInfo::removeWithArg;
1623 // discard spaces before pos(0)
1624 int pos = sub.position(size_t(0));
1626 for (count = 0; pos - count > 0; count++) {
1627 char c = interval.par[pos-count-1];
1628 if (discardComment) {
1629 if ((c != ' ') && (c != '%'))
1635 found._tokenstart = pos - count;
1636 if (sub.str(1).compare(0, 5, "begin") == 0) {
1637 size_t pos1 = pos + sub.str(0).length();
1638 if (sub.str(5).compare("cjk") == 0) {
1639 pos1 = interval.findclosing(pos1+1, interval.par.length()) + 1;
1640 if ((interval.par[pos1] == '{') && (interval.par[pos1+1] == '}'))
1642 found.keytype = KeyInfo::isMain;
1643 found._dataStart = pos1;
1644 found._dataEnd = interval.par.length();
1645 found.disabled = keys["foreignlanguage"].disabled;
1646 found.used = keys["foreignlanguage"].used;
1647 found._tokensize = pos1 - found._tokenstart;
1648 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1651 // Swallow possible optional params
1652 while (interval.par[pos1] == '[') {
1653 pos1 = interval.findclosing(pos1+1, interval.par.length(), '[', ']')+1;
1655 // Swallow also the eventual parameter
1656 if (interval.par[pos1] == '{') {
1657 found._dataEnd = interval.findclosing(pos1+1, interval.par.length()) + 1;
1660 found._dataEnd = pos1;
1662 found._dataStart = found._dataEnd;
1663 found._tokensize = count + found._dataEnd - pos;
1664 found.parenthesiscount = 0;
1665 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1666 found.disabled = true;
1670 // Handle "\end{...}"
1671 found._dataStart = pos + sub.str(0).length();
1672 found._dataEnd = found._dataStart;
1673 found._tokensize = count + found._dataEnd - pos;
1674 found.parenthesiscount = 0;
1675 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1676 found.disabled = true;
1680 else if (found.keytype != KeyInfo::isRegex) {
1681 found._tokenstart = sub.position(size_t(0));
1682 if (found.parenthesiscount == 0) {
1683 // Probably to be discarded
1684 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1685 char following = interval.par[following_pos];
1686 if (following == ' ')
1687 found.head = "\\" + sub.str(3) + " ";
1688 else if (following == '=') {
1689 // like \uldepth=1000pt
1690 found.head = sub.str(0);
1693 found.head = "\\" + key;
1694 found._tokensize = found.head.length();
1695 found._dataEnd = found._tokenstart + found._tokensize;
1696 found._dataStart = found._dataEnd;
1699 int params = found._tokenstart + key.length() + 1;
1700 if (evaluatingOptional) {
1701 if (size_t(found._tokenstart) > optionalEnd) {
1702 evaluatingOptional = false;
1705 found.disabled = true;
1708 int optend = params;
1709 while (interval.par[optend] == '[') {
1710 // discard optional parameters
1711 optend = interval.findclosing(optend+1, interval.par.length(), '[', ']') + 1;
1713 if (optend > params) {
1714 key += interval.par.substr(params, optend-params);
1715 evaluatingOptional = true;
1716 optionalEnd = optend;
1718 string token = sub.str(5);
1719 int closings = found.parenthesiscount;
1720 if (found.parenthesiscount == 1) {
1721 found.head = "\\" + key + "{";
1723 else if (found.parenthesiscount > 1) {
1725 found.head = sub.str(0) + "{";
1726 closings = found.parenthesiscount - 1;
1729 found.head = "\\" + key + "{";
1732 found._tokensize = found.head.length();
1733 found._dataStart = found._tokenstart + found.head.length();
1734 if (interval.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1735 found._dataStart += 15;
1737 size_t endpos = interval.findclosing(found._dataStart, interval.par.length(), '{', '}', closings);
1738 if (found.keytype == KeyInfo::isList) {
1739 // Check if it really is list env
1740 static regex const listre("^([a-z]+)$");
1742 if (!regex_match(token, sub2, listre)) {
1743 // Change the key of this entry. It is not in a list/item environment
1744 found.keytype = KeyInfo::endArguments;
1747 if (found.keytype == KeyInfo::noMain) {
1748 evaluatingCode = true;
1750 codeStart = found._dataStart;
1752 else if (evaluatingCode) {
1753 if (size_t(found._dataStart) > codeEnd)
1754 evaluatingCode = false;
1755 else if (found.keytype == KeyInfo::isMain) {
1756 // Disable this key, treate it as standard
1757 found.keytype = KeyInfo::isStandard;
1758 found.disabled = true;
1759 if ((codeEnd == interval.par.length()) &&
1760 (found._tokenstart == codeStart)) {
1761 // trickery, because the code inset starts
1762 // with \selectlanguage ...
1764 if (entries.size() > 1) {
1765 entries[entries.size()-1]._dataEnd = codeEnd;
1770 if ((endpos == interval.par.length()) &&
1771 (found.keytype == KeyInfo::doRemove)) {
1772 // Missing closing => error in latex-input?
1773 // therefore do not delete remaining data
1774 found._dataStart -= 1;
1775 found._dataEnd = found._dataStart;
1778 found._dataEnd = endpos;
1780 if (isPatternString) {
1781 keys[key].used = true;
1784 entries.push_back(found);
1788 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1790 stringstream s(keysstring);
1792 const char delim = '|';
1793 while (getline(s, key, delim)) {
1794 KeyInfo keyII(keyI);
1795 if (isPatternString) {
1798 else if ( !keys[key].used)
1799 keyII.disabled = true;
1804 void LatexInfo::buildKeys(bool isPatternString)
1807 static bool keysBuilt = false;
1808 if (keysBuilt && !isPatternString) return;
1810 // Known standard keys with 1 parameter.
1811 // Split is done, if not at start of region
1812 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1813 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1814 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1815 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1816 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1817 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
1819 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
1820 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1821 makeKey("section*|subsection*|subsubsection*|paragraph*",
1822 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1823 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1824 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
1826 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
1828 // Split is done, if not at start of region
1829 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
1830 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
1832 // Split is done always.
1833 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
1837 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1838 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1839 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1840 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1842 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1843 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1844 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1846 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1847 // Custom space/skip, remove the content (== length value)
1848 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
1849 // Found in fr/UserGuide.lyx
1850 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1852 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1853 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1854 // Known macros to remove (including their parameter)
1856 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
1857 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
1858 // handle like standard keys with 1 parameter.
1859 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
1861 // Macros to remove, but let the parameter survive
1863 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1865 // Remove language spec from content of these insets
1866 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
1868 // Same effect as previous, parameter will survive (because there is no one anyway)
1870 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1871 // Remove table decorations
1872 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
1873 // Discard shape-header.
1874 // For footnote or shortcut too, because of lang settings
1875 // and wrong handling if used 'KeyInfo::noMain'
1876 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1877 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1878 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1879 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1880 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1881 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
1882 // like ('tiny{}' or '\tiny ' ... )
1883 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
1885 // Survives, like known character
1886 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1887 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
1889 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1890 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1891 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1893 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1894 // Remove RTL/LTR marker
1895 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1896 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
1897 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
1898 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
1899 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
1900 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
1901 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1902 if (isPatternString) {
1903 // Allow the first searched string to rebuild the keys too
1907 // no need to rebuild again
1913 * Keep the list of actual opened parentheses actual
1914 * (e.g. depth == 4 means there are 4 '{' not processed yet)
1916 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
1919 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
1923 if (c == '\\') skip = 1;
1924 else if (c == '{') {
1927 else if (c == '}') {
1928 handleCloseP(i, closingAllowed);
1934 string Intervall::show(int lastpos)
1936 int idx = 0; /* int intervalls */
1939 for (idx = 0; idx <= ignoreidx; idx++) {
1940 while (i < lastpos) {
1942 if (i <= borders[idx].low) {
1943 if (borders[idx].low > lastpos)
1944 printsize = lastpos - i;
1946 printsize = borders[idx].low - i;
1947 s += par.substr(i, printsize);
1949 if (i >= borders[idx].low)
1950 i = borders[idx].upper;
1953 i = borders[idx].upper;
1959 s += par.substr(i, lastpos-i);
1965 void Intervall::output(ostringstream &os, int lastpos)
1967 // get number of chars to output
1968 int idx = 0; /* int intervalls */
1971 string startTitle = titleValue;
1972 for (idx = 0; idx <= ignoreidx; idx++) {
1974 if (i <= borders[idx].low) {
1976 if (borders[idx].low > lastpos)
1977 printsize = lastpos - i;
1979 printsize = borders[idx].low - i;
1980 if (printsize > 0) {
1981 os << startTitle << par.substr(i, printsize);
1983 printed += printsize;
1986 handleParentheses(i, false);
1987 if (i >= borders[idx].low)
1988 i = borders[idx].upper;
1991 i = borders[idx].upper;
1998 os << startTitle << par.substr(i, lastpos-i);
1999 printed += lastpos-i;
2001 handleParentheses(lastpos, false);
2002 for (int i = actualdeptindex; i > 0; --i) {
2005 if (hasTitle && (printed > 0))
2007 if (! isPatternString)
2009 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2012 void LatexInfo::processRegion(int start, int region_end)
2014 while (start < region_end) { /* Let {[} and {]} survive */
2015 int cnt = interval.isOpeningPar(start);
2017 // Closing is allowed past the region
2018 int closing = interval.findclosing(start+1, interval.par.length());
2019 interval.addIntervall(start, start+1);
2020 interval.addIntervall(closing, closing+1);
2024 start = interval.nextNotIgnored(start+1);
2028 void LatexInfo::removeHead(KeyInfo &actual, int count)
2030 if (actual.parenthesiscount == 0) {
2031 // "{\tiny{} ...}" ==> "{{} ...}"
2032 interval.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2035 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2036 interval.addIntervall(actual._tokenstart - count, actual._dataStart);
2037 interval.addIntervall(actual._dataEnd, actual._dataEnd+1);
2041 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2044 switch (actual.keytype)
2046 case KeyInfo::isTitle: {
2048 nextKeyIdx = getNextKey();
2051 case KeyInfo::cleanToStart: {
2052 actual._dataEnd = actual._dataStart;
2053 nextKeyIdx = getNextKey();
2054 // Search for end of arguments
2055 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2057 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2058 entries[i].disabled = true;
2060 actual._dataEnd = entries[tmpIdx]._dataEnd;
2062 while (interval.par[actual._dataEnd] == ' ')
2064 interval.addIntervall(0, actual._dataEnd+1);
2065 interval.actualdeptindex = 0;
2066 interval.depts[0] = actual._dataEnd+1;
2067 interval.closes[0] = -1;
2070 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2071 if (actual.disabled)
2072 interval.addIntervall(actual._tokenstart, actual._dataEnd);
2074 interval.addIntervall(actual._dataStart, actual._dataEnd);
2077 case KeyInfo::isChar: {
2078 nextKeyIdx = getNextKey();
2081 case KeyInfo::isSize: {
2082 if (actual.disabled || (interval.par[actual._dataStart] != '{') || (interval.par[actual._dataStart-1] == ' ')) {
2083 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2084 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2085 nextKeyIdx = getNextKey();
2087 // Here _dataStart points to '{', so correct it
2088 actual._dataStart += 1;
2089 actual._tokensize += 1;
2090 actual.parenthesiscount = 1;
2091 if (interval.par[actual._dataStart] == '}') {
2092 // Determine the end if used like '{\tiny{}...}'
2093 actual._dataEnd = interval.findclosing(actual._dataStart+1, interval.par.length()) + 1;
2094 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2097 // Determine the end if used like '\tiny{...}'
2098 actual._dataEnd = interval.findclosing(actual._dataStart, interval.par.length()) + 1;
2100 // Split on this key if not at start
2101 int start = interval.nextNotIgnored(previousStart);
2102 if (start < actual._tokenstart) {
2103 interval.output(os, actual._tokenstart);
2104 interval.addIntervall(start, actual._tokenstart);
2106 // discard entry if at end of actual
2107 nextKeyIdx = process(os, actual);
2111 case KeyInfo::endArguments:
2112 // Remove trailing '{}' too
2113 actual._dataStart += 2;
2114 actual._dataEnd += 2;
2115 interval.addIntervall(actual._tokenstart, actual._dataEnd);
2116 nextKeyIdx = getNextKey();
2118 case KeyInfo::noMain:
2120 case KeyInfo::isStandard: {
2121 if (actual.disabled) {
2123 processRegion(actual._dataStart, actual._dataStart+1);
2124 nextKeyIdx = getNextKey();
2126 // Split on this key if not at datastart of calling entry
2127 int start = interval.nextNotIgnored(previousStart);
2128 if (start < actual._tokenstart) {
2129 interval.output(os, actual._tokenstart);
2130 interval.addIntervall(start, actual._tokenstart);
2132 // discard entry if at end of actual
2133 nextKeyIdx = process(os, actual);
2137 case KeyInfo::removeWithArg: {
2138 nextKeyIdx = getNextKey();
2139 // Search for end of arguments
2140 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2142 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2143 entries[i].disabled = true;
2145 actual._dataEnd = entries[tmpIdx]._dataEnd;
2147 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2150 case KeyInfo::doRemove: {
2151 // Remove the key with all parameters and following spaces
2153 for (pos = actual._dataEnd+1; pos < interval.par.length(); pos++) {
2154 if ((interval.par[pos] != ' ') && (interval.par[pos] != '%'))
2157 interval.addIntervall(actual._tokenstart, pos);
2158 nextKeyIdx = getNextKey();
2161 case KeyInfo::isList: {
2162 // Discard space before _tokenstart
2164 for (count = 0; count < actual._tokenstart; count++) {
2165 if (interval.par[actual._tokenstart-count-1] != ' ')
2168 nextKeyIdx = getNextKey();
2169 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2171 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2173 // How else can we catch this one?
2174 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2175 entries[i].disabled = true;
2177 actual._dataEnd = entries[tmpIdx]._dataEnd;
2179 else if (nextKeyIdx > 0) {
2180 // Ignore any lang entries inside data region
2181 for (int i = nextKeyIdx; i < int(entries.size()) && entries[i]._tokenstart < actual._dataEnd; i++) {
2182 if (entries[i].keytype == KeyInfo::isMain)
2183 entries[i].disabled = true;
2186 if (actual.disabled) {
2187 interval.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2190 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2192 if (interval.par[actual._dataEnd+1] == '[') {
2193 int posdown = interval.findclosing(actual._dataEnd+2, interval.par.length(), '[', ']');
2194 if ((interval.par[actual._dataEnd+2] == '{') &&
2195 (interval.par[posdown-1] == '}')) {
2196 interval.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2197 interval.addIntervall(posdown-1, posdown+1);
2200 interval.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2201 interval.addIntervall(posdown, posdown+1);
2203 int blk = interval.nextNotIgnored(actual._dataEnd+1);
2204 if (blk > posdown) {
2205 // Discard at most 1 space after empty item
2207 for (count = 0; count < 1; count++) {
2208 if (interval.par[blk+count] != ' ')
2212 interval.addIntervall(blk, blk+count);
2217 case KeyInfo::isSectioning: {
2218 // Discard spaces before _tokenstart
2220 int val = actual._tokenstart;
2221 for (count = 0; count < actual._tokenstart;) {
2222 val = interval.previousNotIgnored(val-1);
2223 if (interval.par[val] != ' ')
2226 count = actual._tokenstart - val;
2229 if (actual.disabled) {
2230 removeHead(actual, count);
2231 nextKeyIdx = getNextKey();
2233 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2234 nextKeyIdx = process(os, actual);
2238 case KeyInfo::isMath: {
2239 // Same as regex, use the content unchanged
2240 nextKeyIdx = getNextKey();
2243 case KeyInfo::isRegex: {
2244 // DO NOT SPLIT ON REGEX
2246 nextKeyIdx = getNextKey();
2249 case KeyInfo::isIgnored: {
2250 // Treat like a character for now
2251 nextKeyIdx = getNextKey();
2254 case KeyInfo::isMain: {
2255 if (interval.par.substr(actual._dataStart, 2) == "% ")
2256 interval.addIntervall(actual._dataStart, actual._dataStart+2);
2257 if (actual._tokenstart > 0) {
2258 int prev = interval.previousNotIgnored(actual._tokenstart - 1);
2259 if ((prev >= 0) && interval.par[prev] == '%')
2260 interval.addIntervall(prev, prev+1);
2262 if (actual.disabled) {
2264 if ((interval.par.substr(actual._dataStart, 3) == " \\[") ||
2265 (interval.par.substr(actual._dataStart, 8) == " \\begin{")) {
2266 // Discard also the space before math-equation
2267 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2269 nextKeyIdx = getNextKey();
2270 // interval.resetOpenedP(actual._dataStart-1);
2273 if (actual._tokenstart < 26) {
2274 // for the first (and maybe dummy) language
2275 interval.setForDefaultLang(actual);
2277 interval.resetOpenedP(actual._dataStart-1);
2281 case KeyInfo::invalid:
2282 // This cannot happen, already handled
2285 // LYXERR0("Unhandled keytype");
2286 nextKeyIdx = getNextKey();
2293 int LatexInfo::process(ostringstream &os, KeyInfo &actual )
2295 int end = interval.nextNotIgnored(actual._dataEnd);
2296 int oldStart = actual._dataStart;
2297 int nextKeyIdx = getNextKey();
2299 if ((nextKeyIdx < 0) ||
2300 (entries[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2301 (entries[nextKeyIdx].keytype == KeyInfo::invalid)) {
2302 if (oldStart <= end) {
2303 processRegion(oldStart, end);
2308 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2310 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2311 (void) dispatch(os, actual._dataStart, nextKey);
2312 end = nextKey._tokenstart;
2315 processRegion(oldStart, nextKey._tokenstart);
2316 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2318 oldStart = nextKey._dataEnd+1;
2320 // now nextKey is either invalid or is outside of actual._dataEnd
2321 // output the remaining and discard myself
2322 if (oldStart <= end) {
2323 processRegion(oldStart, end);
2325 if (interval.par[end] == '}') {
2327 // This is the normal case.
2328 // But if using the firstlanguage, the closing may be missing
2330 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2332 if (actual._dataEnd < end)
2333 output_end = interval.nextNotIgnored(actual._dataEnd);
2335 output_end = interval.nextNotIgnored(end);
2336 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2337 interval.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2339 // Remove possible empty data
2340 int dstart = interval.nextNotIgnored(actual._dataStart);
2341 while (interval.isOpeningPar(dstart) == 1) {
2342 interval.addIntervall(dstart, dstart+1);
2343 int dend = interval.findclosing(dstart+1, output_end);
2344 interval.addIntervall(dend, dend+1);
2345 dstart = interval.nextNotIgnored(dstart+1);
2347 if (dstart < output_end)
2348 interval.output(os, output_end);
2349 interval.addIntervall(actual._tokenstart, end);
2353 string splitOnKnownMacros(string par, bool isPatternString)
2356 LatexInfo li(par, isPatternString);
2357 // LYXERR0("Berfore split: " << par);
2358 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2360 DummyKey._tokensize = 0;
2361 DummyKey._dataStart = 0;
2362 DummyKey._dataEnd = par.length();
2363 DummyKey.disabled = true;
2364 int firstkeyIdx = li.getFirstKey();
2366 if (firstkeyIdx >= 0) {
2367 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2368 DummyKey._tokenstart = firstKey._tokenstart;
2370 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2371 // Use dummy firstKey
2372 firstKey = DummyKey;
2373 (void) li.setNextKey(firstkeyIdx);
2376 if (par.substr(firstKey._dataStart, 2) == "% ")
2377 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2379 nextkeyIdx = li.process(os, firstKey);
2380 while (nextkeyIdx >= 0) {
2381 // Check for a possible gap between the last
2382 // entry and this one
2383 int datastart = li.nextNotIgnored(firstKey._dataStart);
2384 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2385 if ((nextKey._tokenstart > datastart)) {
2387 firstKey._dataStart = datastart;
2388 firstKey._dataEnd = par.length();
2389 (void) li.setNextKey(nextkeyIdx);
2390 // Fake the last opened parenthesis
2391 li.setForDefaultLang(firstKey);
2392 nextkeyIdx = li.process(os, firstKey);
2395 if (nextKey.keytype != KeyInfo::isMain) {
2396 firstKey._dataStart = datastart;
2397 firstKey._dataEnd = nextKey._dataEnd+1;
2398 (void) li.setNextKey(nextkeyIdx);
2399 li.setForDefaultLang(firstKey);
2400 nextkeyIdx = li.process(os, firstKey);
2403 nextkeyIdx = li.process(os, nextKey);
2407 // Handle the remaining
2408 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2409 firstKey._dataEnd = par.length();
2411 if ((firstKey._dataStart < firstKey._dataEnd) &&
2412 (par[firstKey._dataStart] != '}')) {
2413 li.setForDefaultLang(firstKey);
2414 (void) li.process(os, firstKey);
2418 // return string definitelly impossible to match
2419 s = "\\foreignlanguage{ignore}{ }";
2423 s = par; /* no known macros found */
2424 // LYXERR0("After split: " << s);
2429 * Try to unify the language specs in the latexified text.
2430 * Resulting modified string is set to "", if
2431 * the searched tex does not contain all the features in the search pattern
2433 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2435 static Features regex_f;
2436 static int missed = 0;
2437 static bool regex_with_format = false;
2439 int parlen = par.length();
2441 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2444 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2445 // Happens to be there in case of description or labeling environment
2450 // Split the latex input into pieces which
2451 // can be digested by our search engine
2452 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2453 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2454 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2457 result = par.substr(0, parlen);
2458 if (isPatternString) {
2461 regex_f = identifyFeatures(result);
2462 string features = "";
2463 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2464 string a = it->first;
2465 regex_with_format = true;
2466 features += " " + a;
2467 // LYXERR0("Identified regex format:" << a);
2469 LYXERR(Debug::FIND, "Identified Features" << features);
2472 } else if (regex_with_format) {
2473 Features info = identifyFeatures(result);
2474 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2475 string a = it->first;
2476 bool b = it->second;
2477 if (b && ! info[a]) {
2479 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2485 // LYXERR0("No regex formats");
2491 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2492 static int identifyClosing(string & t)
2494 int open_braces = 0;
2496 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2497 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2499 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]" REGEX_EOS, "$1"))
2501 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2503 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2513 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2514 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2516 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2517 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2518 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2519 // When using regexp, braces are hacked already by escape_for_regex()
2520 par_as_string = normalize(ds, !use_regexp);
2522 close_wildcards = 0;
2524 size_t lead_size = 0;
2525 // correct the language settings
2526 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2527 if (opt.ignoreformat) {
2529 // if par_as_string_nolead were emty,
2530 // the following call to findAux will always *find* the string
2531 // in the checked data, and thus always using the slow
2532 // examining of the current text part.
2533 par_as_string_nolead = par_as_string;
2536 lead_size = identifyLeading(par_as_string);
2537 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2538 lead_as_string = par_as_string.substr(0, lead_size);
2539 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2543 open_braces = identifyClosing(par_as_string);
2544 identifyClosing(par_as_string_nolead);
2545 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2546 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2548 string lead_as_regexp;
2549 if (lead_size > 0) {
2550 // @todo No need to search for \regexp{} insets in leading material
2551 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2552 par_as_string = par_as_string_nolead;
2553 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2554 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2556 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2557 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2558 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2560 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2561 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2562 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2563 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2564 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2565 || regex_replace(par_as_string, par_as_string,
2566 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2567 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2568 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2572 if (!opt.ignoreformat) {
2573 // Remove extra '\}' at end if not part of \{\.\}
2574 size_t lng = par_as_string.size();
2576 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2578 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2587 if (lng < par_as_string.size())
2588 par_as_string = par_as_string.substr(0,lng);
2591 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2592 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2593 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2594 // replace '[^...]' with '[^...\}\{\\]'
2595 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2596 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2597 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2599 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2602 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2603 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2604 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2605 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2607 // If entered regexp must match at begin of searched string buffer
2608 // Kornel: Added parentheses to use $1 for size of the leading string
2612 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2613 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2614 // so the convert has no effect in that case
2615 for (int i = 8; i > 0; --i) {
2616 string orig = "\\\\" + std::to_string(i);
2617 string dest = "\\" + std::to_string(i+1);
2618 while (regex_replace(par_as_string, par_as_string, orig, dest));
2620 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2621 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2623 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2624 regexp = lyx::regex(regexp_str);
2626 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2627 regexp2 = lyx::regex(regexp2_str);
2632 // Count number of characters in string
2636 // \\[a-zA-Z]+ ==> 1
2637 static int computeSize(string s, int len)
2643 for (int i = 0; i < len; i += skip, count++) {
2646 if (isalpha(s[i+1])) {
2647 for (int j = 2; i+j < len; j++) {
2648 if (! isalpha(s[i+j])) {
2651 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2653 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2661 else if (s[i] == '{') {
2667 else if (s[i] == '-') {
2668 if (s[i+1] == '-') {
2684 int MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2687 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2690 docstring docstr = stringifyFromForSearch(opt, cur, len);
2691 string str = normalize(docstr, true);
2692 if (!opt.ignoreformat) {
2693 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2695 if (str.empty()) return(-1);
2696 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2697 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2700 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2701 regex const *p_regexp;
2702 regex_constants::match_flag_type flags;
2704 flags = regex_constants::match_continuous;
2707 flags = regex_constants::match_default;
2708 p_regexp = ®exp2;
2710 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2711 if (re_it == sregex_iterator())
2713 match_results<string::const_iterator> const & m = *re_it;
2715 if (0) { // Kornel Benko: DO NOT CHECKK
2716 // Check braces on the segment that matched the entire regexp expression,
2717 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2718 if (!braces_match(m[0].first, m[0].second, open_braces))
2722 // Check braces on segments that matched all (.*?) subexpressions,
2723 // except the last "padding" one inserted by lyx.
2724 for (size_t i = 1; i < m.size() - 1; ++i)
2725 if (!braces_match(m[i].first, m[i].second, open_braces))
2728 // Exclude from the returned match length any length
2729 // due to close wildcards added at end of regexp
2730 // and also the length of the leading (e.g. '\emph{')
2732 // Whole found string, including the leading: m[0].second - m[0].first
2733 // Size of the leading string: m[1].second - m[1].first
2734 int leadingsize = 0;
2736 leadingsize = m[1].second - m[1].first;
2738 for (size_t i = 0; i < m.size(); i++) {
2739 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2741 if (close_wildcards == 0)
2742 result = m[0].second - m[0].first;
2745 result = m[m.size() - close_wildcards].first - m[0].first;
2747 size_t pos = m.position(size_t(0));
2748 // Ignore last closing characters
2749 while (result > 0) {
2750 if (str[pos+result-1] == '}')
2755 if (result > leadingsize)
2756 result -= leadingsize;
2759 return computeSize(str.substr(pos+leadingsize,result), result);
2762 // else !use_regexp: but all code paths above return
2763 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2764 << par_as_string << "', str='" << str << "'");
2765 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2766 << lead_as_string << "', par_as_string_nolead='"
2767 << par_as_string_nolead << "'");
2770 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2771 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2772 if (str.substr(0, par_as_string.size()) == par_as_string)
2773 return par_as_string.size();
2775 size_t pos = str.find(par_as_string_nolead);
2776 if (pos != string::npos)
2777 return par_as_string.size();
2783 int MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
2785 int res = findAux(cur, len, at_begin);
2787 "res=" << res << ", at_begin=" << at_begin
2788 << ", matchword=" << opt.matchword
2789 << ", inTexted=" << cur.inTexted());
2790 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
2792 if ((len > 0) && (res < len))
2794 Paragraph const & par = cur.paragraph();
2795 bool ws_left = (cur.pos() > 0)
2796 ? par.isWordSeparator(cur.pos() - 1)
2798 bool ws_right = (cur.pos() + len < par.size())
2799 ? par.isWordSeparator(cur.pos() + len)
2802 "cur.pos()=" << cur.pos() << ", res=" << res
2803 << ", separ: " << ws_left << ", " << ws_right
2806 if (ws_left && ws_right) {
2807 // Check for word separators inside the found 'word'
2808 for (int i = 0; i < len; i++) {
2809 if (par.isWordSeparator(cur.pos() + i))
2818 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
2821 if (! opt.casesensitive)
2822 t = lyx::to_utf8(lowercase(s));
2824 t = lyx::to_utf8(s);
2825 // Remove \n at begin
2826 while (!t.empty() && t[0] == '\n')
2829 while (!t.empty() && t[t.size() - 1] == '\n')
2830 t = t.substr(0, t.size() - 1);
2832 // Handle all other '\n'
2833 while ((pos = t.find("\n")) != string::npos) {
2834 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
2836 if (std::isalnum(t[pos+1])) {
2837 t.replace(pos-2, 3, " ");
2840 t.replace(pos-2, 3, "");
2843 else if (!std::isalnum(t[pos+1]) || !std::isalnum(t[pos-1])) {
2844 // '\n' adjacent to non-alpha-numerics, discard
2845 t.replace(pos, 1, "");
2848 // Replace all other \n with spaces
2849 t.replace(pos, 1, " ");
2852 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
2853 // Kornel: Added textsl, textsf, textit, texttt and noun
2854 // + allow to seach for colored text too
2855 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
2856 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
2857 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2858 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
2859 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2861 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
2862 // FIXME - check what preceeds the brace
2864 if (opt.ignoreformat)
2865 while (regex_replace(t, t, "\\{", "_x_<")
2866 || regex_replace(t, t, "\\}", "_x_>"))
2867 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2869 while (regex_replace(t, t, "\\\\\\{", "_x_<")
2870 || regex_replace(t, t, "\\\\\\}", "_x_>"))
2871 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2878 docstring stringifyFromCursor(DocIterator const & cur, int len)
2880 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
2881 if (cur.inTexted()) {
2882 Paragraph const & par = cur.paragraph();
2883 // TODO what about searching beyond/across paragraph breaks ?
2884 // TODO Try adding a AS_STR_INSERTS as last arg
2885 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
2886 int(par.size()) : cur.pos() + len;
2887 OutputParams runparams(&cur.buffer()->params().encoding());
2888 runparams.nice = true;
2889 runparams.flavor = OutputParams::LATEX;
2890 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
2891 // No side effect of file copying and image conversion
2892 runparams.dryrun = true;
2893 LYXERR(Debug::FIND, "Stringifying with cur: "
2894 << cur << ", from pos: " << cur.pos() << ", end: " << end);
2895 return par.asString(cur.pos(), end,
2896 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
2898 } else if (cur.inMathed()) {
2899 CursorSlice cs = cur.top();
2900 MathData md = cs.cell();
2901 MathData::const_iterator it_end =
2902 (( len == -1 || cs.pos() + len > int(md.size()))
2904 : md.begin() + cs.pos() + len );
2906 for (MathData::const_iterator it = md.begin() + cs.pos();
2909 docstring s = asString(md2);
2910 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
2913 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
2918 /** Computes the LaTeX export of buf starting from cur and ending len positions
2919 * after cur, if len is positive, or at the paragraph or innermost inset end
2922 docstring latexifyFromCursor(DocIterator const & cur, int len)
2924 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
2925 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
2926 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
2927 Buffer const & buf = *cur.buffer();
2929 odocstringstream ods;
2931 OutputParams runparams(&buf.params().encoding());
2932 runparams.nice = false;
2933 runparams.flavor = OutputParams::LATEX;
2934 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
2935 // No side effect of file copying and image conversion
2936 runparams.dryrun = true;
2937 runparams.for_search = true;
2939 if (cur.inTexted()) {
2940 // @TODO what about searching beyond/across paragraph breaks ?
2941 pos_type endpos = cur.paragraph().size();
2942 if (len != -1 && endpos > cur.pos() + len)
2943 endpos = cur.pos() + len;
2944 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
2945 string(), cur.pos(), endpos);
2946 string s = lyx::to_utf8(ods.str());
2947 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
2948 return(lyx::from_utf8(s));
2949 } else if (cur.inMathed()) {
2950 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
2951 for (int s = cur.depth() - 1; s >= 0; --s) {
2952 CursorSlice const & cs = cur[s];
2953 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
2955 cs.asInsetMath()->asHullInset()->header_write(ws);
2960 CursorSlice const & cs = cur.top();
2961 MathData md = cs.cell();
2962 MathData::const_iterator it_end =
2963 ((len == -1 || cs.pos() + len > int(md.size()))
2965 : md.begin() + cs.pos() + len);
2967 for (MathData::const_iterator it = md.begin() + cs.pos();
2971 ods << asString(md2);
2972 // Retrieve the math environment type, and add '$' or '$]'
2973 // or others (\end{equation}) accordingly
2974 for (int s = cur.depth() - 1; s >= 0; --s) {
2975 CursorSlice const & cs2 = cur[s];
2976 InsetMath * inset = cs2.asInsetMath();
2977 if (inset && inset->asHullInset()) {
2979 inset->asHullInset()->footer_write(ws);
2983 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
2985 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
2991 /** Finalize an advanced find operation, advancing the cursor to the innermost
2992 ** position that matches, plus computing the length of the matching text to
2995 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
2997 // Search the foremost position that matches (avoids find of entire math
2998 // inset when match at start of it)
3000 DocIterator old_cur(cur.buffer());
3002 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3006 } while (cur && cur.depth() > d && match(cur) > 0);
3008 int max_match = match(cur); /* match valid only if not searching whole words */
3009 if (max_match <= 0) return 0;
3010 LYXERR(Debug::FIND, "Ok");
3012 // Compute the match length
3014 if (cur.pos() + len > cur.lastpos())
3016 if (match.opt.matchword) {
3017 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3018 while (cur.pos() + len <= cur.lastpos() && match(cur, len) <= 0) {
3020 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3022 // Length of matched text (different from len param)
3023 int old_match = match(cur, len);
3027 // Greedy behaviour while matching regexps
3028 while ((new_match = match(cur, len + 1)) > old_match) {
3030 old_match = new_match;
3031 LYXERR(Debug::FIND, "verifying match with len = " << len);
3038 int maxl = cur.lastpos() - cur.pos();
3039 // Greedy behaviour while matching regexps
3040 while (maxl > minl) {
3041 int actual_match = match(cur, len);
3042 if (actual_match >= max_match) {
3043 // actual_match > max_match _can_ happen,
3044 // if the search area splits
3045 // some following word so that the regex
3046 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3048 // This means, the len value is too big
3050 len = (int)((maxl + minl)/2);
3053 // (actual_match < max_match)
3055 len = (int)((maxl + minl)/2);
3059 // Search for real start of matched characters
3064 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3065 if (cur.depth() < old_cur.depth()) {
3067 LYXERR0("cur.depth() < old_cur.depth(), this should never happen");
3070 if (cur.pos() != old_cur.pos()) {
3071 // OK, forwarded 1 pos in actual inset
3072 actual_match = match(cur, len-1);
3073 if (actual_match == max_match) {
3074 // Ha, got it! The shorter selection has the same match length
3079 // OK, the shorter selection matches less chars, revert to previous value
3085 LYXERR0("cur.pos() == old_cur.pos(), this should never happen");
3086 actual_match = match(cur, len);
3087 if (actual_match == max_match)
3097 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3101 while (!theApp()->longOperationCancelled() && cur) {
3102 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3103 int match_len = match(cur, -1, false);
3104 LYXERR(Debug::FIND, "match_len: " << match_len);
3105 if (match_len > 0) {
3106 int match_len_zero_count = 0;
3107 for (; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3108 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3109 int match_len3 = match(cur, 1);
3112 int match_len2 = match(cur);
3113 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3114 if (match_len2 > 0) {
3115 // Sometimes in finalize we understand it wasn't a match
3116 // and we need to continue the outest loop
3117 int len = findAdvFinalize(cur, match);
3122 if (match_len2 >= 0) {
3123 if (match_len2 == 0)
3124 match_len_zero_count++;
3126 match_len_zero_count = 0;
3129 if (++match_len_zero_count > 3) {
3130 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3131 match_len_zero_count = 0;
3139 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3140 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3143 // This should exit nested insets, if any, or otherwise undefine the currsor.
3144 cur.pos() = cur.lastpos();
3145 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3153 /// Find the most backward consecutive match within same paragraph while searching backwards.
3154 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3156 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3157 DocIterator tmp_cur = cur;
3158 int len = findAdvFinalize(tmp_cur, match);
3159 Inset & inset = cur.inset();
3160 for (; cur != cur_begin; cur.backwardPos()) {
3161 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3162 DocIterator new_cur = cur;
3163 new_cur.backwardPos();
3164 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur))
3166 int new_len = findAdvFinalize(new_cur, match);
3171 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3177 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3181 // Backup of original position
3182 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3183 if (cur == cur_begin)
3186 DocIterator cur_orig(cur);
3187 bool pit_changed = false;
3190 bool found_match = match(cur, -1, false);
3194 cur.pos() = cur.lastpos();
3196 cur.pos() = cur_orig.pos();
3197 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3198 DocIterator cur_prev_iter;
3200 found_match = match(cur);
3201 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3202 << found_match << ", cur: " << cur);
3204 return findMostBackwards(cur, match);
3206 // Stop if begin of document reached
3207 if (cur == cur_begin)
3209 cur_prev_iter = cur;
3213 if (cur == cur_begin)
3220 } while (!theApp()->longOperationCancelled());
3228 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3229 DocIterator const & cur, int len)
3231 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3233 if (!opt.ignoreformat)
3234 return latexifyFromCursor(cur, len);
3236 return stringifyFromCursor(cur, len);
3240 FindAndReplaceOptions::FindAndReplaceOptions(
3241 docstring const & find_buf_name, bool casesensitive,
3242 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
3243 docstring const & repl_buf_name, bool keep_case,
3244 SearchScope scope, SearchRestriction restr)
3245 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
3246 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
3247 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope), restr(restr)
3255 /** Check if 'len' letters following cursor are all non-lowercase */
3256 static bool allNonLowercase(Cursor const & cur, int len)
3258 pos_type beg_pos = cur.selectionBegin().pos();
3259 pos_type end_pos = cur.selectionBegin().pos() + len;
3260 if (len > cur.lastpos() + 1 - beg_pos) {
3261 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3262 len = cur.lastpos() + 1 - beg_pos;
3263 end_pos = beg_pos + len;
3265 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3266 if (isLowerCase(cur.paragraph().getChar(pos)))
3272 /** Check if first letter is upper case and second one is lower case */
3273 static bool firstUppercase(Cursor const & cur)
3276 pos_type pos = cur.selectionBegin().pos();
3277 if (pos >= cur.lastpos() - 1) {
3278 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3281 ch1 = cur.paragraph().getChar(pos);
3282 ch2 = cur.paragraph().getChar(pos + 1);
3283 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3284 LYXERR(Debug::FIND, "firstUppercase(): "
3285 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3286 << ch2 << "(" << char(ch2) << ")"
3287 << ", result=" << result << ", cur=" << cur);
3292 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3294 ** \fixme What to do with possible further paragraphs in replace buffer ?
3296 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3298 ParagraphList::iterator pit = buffer.paragraphs().begin();
3299 LASSERT(pit->size() >= 1, /**/);
3300 pos_type right = pos_type(1);
3301 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3302 right = pit->size();
3303 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3309 static void findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3311 Cursor & cur = bv->cursor();
3312 if (opt.repl_buf_name == docstring()
3313 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3314 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3317 DocIterator sel_beg = cur.selectionBegin();
3318 DocIterator sel_end = cur.selectionEnd();
3319 if (&sel_beg.inset() != &sel_end.inset()
3320 || sel_beg.pit() != sel_end.pit()
3321 || sel_beg.idx() != sel_end.idx())
3323 int sel_len = sel_end.pos() - sel_beg.pos();
3324 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3325 << ", sel_len: " << sel_len << endl);
3328 LASSERT(sel_len > 0, return);
3330 if (!matchAdv(sel_beg, sel_len))
3333 // Build a copy of the replace buffer, adapted to the KeepCase option
3334 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3336 repl_buffer_orig.write(oss);
3337 string lyx = oss.str();
3338 Buffer repl_buffer("", false);
3339 repl_buffer.setUnnamed(true);
3340 LASSERT(repl_buffer.readString(lyx), return);
3341 if (opt.keep_case && sel_len >= 2) {
3342 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3343 if (cur.inTexted()) {
3344 if (firstUppercase(cur))
3345 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3346 else if (allNonLowercase(cur, sel_len))
3347 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3350 cap::cutSelection(cur, false);
3351 if (cur.inTexted()) {
3352 repl_buffer.changeLanguage(
3353 repl_buffer.language(),
3354 cur.getFont().language());
3355 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3356 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3357 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3358 repl_buffer.params().documentClassPtr(),
3359 bv->buffer().errorList("Paste"));
3360 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3361 sel_len = repl_buffer.paragraphs().begin()->size();
3362 } else if (cur.inMathed()) {
3363 odocstringstream ods;
3365 OutputParams runparams(&repl_buffer.params().encoding());
3366 runparams.nice = false;
3367 runparams.flavor = OutputParams::LATEX;
3368 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3369 runparams.dryrun = true;
3370 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3371 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3372 docstring repl_latex = ods.str();
3373 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3375 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3376 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3377 repl_latex = from_utf8(s);
3378 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3379 MathData ar(cur.buffer());
3380 asArray(repl_latex, ar, Parse::NORMAL);
3382 sel_len = ar.size();
3383 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3385 if (cur.pos() >= sel_len)
3386 cur.pos() -= sel_len;
3389 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3390 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3391 bv->processUpdateFlags(Update::Force);
3395 /// Perform a FindAdv operation.
3396 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3401 // e.g., when invoking word-findadv from mini-buffer wither with
3402 // wrong options syntax or before ever opening advanced F&R pane
3403 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3407 MatchStringAdv matchAdv(bv->buffer(), opt);
3408 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3410 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3411 findAdvReplace(bv, opt, matchAdv);
3414 match_len = findForwardAdv(cur, matchAdv);
3416 match_len = findBackwardsAdv(cur, matchAdv);
3418 // This may only be raised by lyx::regex()
3419 bv->message(_("Invalid regular expression!"));
3423 if (match_len == 0) {
3424 bv->message(_("Match not found!"));
3428 bv->message(_("Match found!"));
3430 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3431 bv->putSelectionAt(cur, match_len, !opt.forward);
3437 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3439 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3440 << opt.casesensitive << ' '
3441 << opt.matchword << ' '
3442 << opt.forward << ' '
3443 << opt.expandmacros << ' '
3444 << opt.ignoreformat << ' '
3445 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3446 << opt.keep_case << ' '
3447 << int(opt.scope) << ' '
3450 LYXERR(Debug::FIND, "built: " << os.str());
3456 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3458 LYXERR(Debug::FIND, "parsing");
3462 while (line != "EOSS") {
3466 if (is.eof()) // Tolerate malformed request
3470 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3471 opt.find_buf_name = from_utf8(s);
3472 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat;
3473 is.get(); // Waste space before replace string
3476 while (line != "EOSS") {
3480 if (is.eof()) // Tolerate malformed request
3484 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3485 opt.repl_buf_name = from_utf8(s);
3486 is >> opt.keep_case;
3489 opt.scope = FindAndReplaceOptions::SearchScope(i);
3491 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3493 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3494 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3495 << opt.scope << ' ' << opt.restr);