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;
64 bool parse_bool(docstring & howto)
69 howto = split(howto, var, ' ');
74 class MatchString : public binary_function<Paragraph, pos_type, int>
77 MatchString(docstring const & str, bool cs, bool mw)
78 : str(str), case_sens(cs), whole_words(mw)
81 // returns true if the specified string is at the specified position
82 // del specifies whether deleted strings in ct mode will be considered
83 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
85 return par.find(str, case_sens, whole_words, pos, del);
93 // match whole words only
98 int findForward(DocIterator & cur, MatchString const & match,
101 for (; cur; cur.forwardChar())
102 if (cur.inTexted()) {
103 int len = match(cur.paragraph(), cur.pos(), find_del);
111 int findBackwards(DocIterator & cur, MatchString const & match,
112 bool find_del = true)
116 if (cur.inTexted()) {
117 int len = match(cur.paragraph(), cur.pos(), find_del);
126 bool searchAllowed(docstring const & str)
129 frontend::Alert::error(_("Search error"), _("Search string is empty"));
136 bool findOne(BufferView * bv, docstring const & searchstr,
137 bool case_sens, bool whole, bool forward,
138 bool find_del = true, bool check_wrap = false)
140 if (!searchAllowed(searchstr))
143 DocIterator cur = forward
144 ? bv->cursor().selectionEnd()
145 : bv->cursor().selectionBegin();
147 MatchString const match(searchstr, case_sens, whole);
149 int match_len = forward
150 ? findForward(cur, match, find_del)
151 : findBackwards(cur, match, find_del);
154 bv->putSelectionAt(cur, match_len, !forward);
155 else if (check_wrap) {
156 DocIterator cur_orig(bv->cursor());
159 q = _("End of file reached while searching forward.\n"
160 "Continue searching from the beginning?");
162 q = _("Beginning of file reached while searching backward.\n"
163 "Continue searching from the end?");
164 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
165 q, 0, 1, _("&Yes"), _("&No"));
166 if (wrap_answer == 0) {
168 bv->cursor().clear();
169 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
171 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
172 bv->cursor().backwardPos();
174 bv->clearSelection();
175 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
178 bv->cursor().setCursor(cur_orig);
182 return match_len > 0;
186 int replaceAll(BufferView * bv,
187 docstring const & searchstr, docstring const & replacestr,
188 bool case_sens, bool whole)
190 Buffer & buf = bv->buffer();
192 if (!searchAllowed(searchstr) || buf.isReadonly())
195 DocIterator cur_orig(bv->cursor());
197 MatchString const match(searchstr, case_sens, whole);
200 int const rsize = replacestr.size();
201 int const ssize = searchstr.size();
204 cur.setCursor(doc_iterator_begin(&buf));
205 int match_len = findForward(cur, match, false);
206 while (match_len > 0) {
207 // Backup current cursor position and font.
208 pos_type const pos = cur.pos();
209 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
211 int striked = ssize -
212 cur.paragraph().eraseChars(pos, pos + match_len,
213 buf.params().track_changes);
214 cur.paragraph().insert(pos, replacestr, font,
215 Change(buf.params().track_changes
217 : Change::UNCHANGED));
218 for (int i = 0; i < rsize + striked; ++i)
221 match_len = findForward(cur, match, false);
224 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
226 cur_orig.fixIfBroken();
227 bv->setCursor(cur_orig);
233 // the idea here is that we are going to replace the string that
234 // is selected IF it is the search string.
235 // if there is a selection, but it is not the search string, then
236 // we basically ignore it. (FIXME We ought to replace only within
238 // if there is no selection, then:
239 // (i) if some search string has been provided, then we find it.
240 // (think of how the dialog works when you hit "replace" the
242 // (ii) if no search string has been provided, then we treat the
243 // word the cursor is in as the search string. (why? i have no
244 // idea.) but this only works in text?
246 // returns the number of replacements made (one, if any) and
247 // whether anything at all was done.
248 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
249 docstring const & replacestr, bool case_sens,
250 bool whole, bool forward, bool findnext)
252 Cursor & cur = bv->cursor();
254 if (!cur.selection()) {
255 // no selection, non-empty search string: find it
256 if (!searchstr.empty()) {
257 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
258 return make_pair(found, 0);
260 // empty search string
263 return make_pair(false, 0);
264 // select current word and treat it as the search string.
265 // This causes a minor bug as undo will restore this selection,
266 // which the user did not create (#8986).
267 cur.innerText()->selectWord(cur, WHOLE_WORD);
268 searchstr = cur.selectionAsString(false);
271 // if we still don't have a search string, report the error
273 if (!searchAllowed(searchstr))
274 return make_pair(false, 0);
276 bool have_selection = cur.selection();
277 docstring const selected = cur.selectionAsString(false);
280 ? searchstr == selected
281 : compare_no_case(searchstr, selected) == 0;
283 // no selection or current selection is not search word:
284 // just find the search word
285 if (!have_selection || !match) {
286 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
287 return make_pair(found, 0);
290 // we're now actually ready to replace. if the buffer is
291 // read-only, we can't, though.
292 if (bv->buffer().isReadonly())
293 return make_pair(false, 0);
295 cap::replaceSelectionWithString(cur, replacestr);
297 cur.pos() += replacestr.length();
298 LASSERT(cur.pos() <= cur.lastpos(),
299 cur.pos() = cur.lastpos());
302 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
304 return make_pair(true, 1);
310 docstring const find2string(docstring const & search,
311 bool casesensitive, bool matchword, bool forward)
315 << int(casesensitive) << ' '
316 << int(matchword) << ' '
322 docstring const replace2string(docstring const & replace,
323 docstring const & search,
324 bool casesensitive, bool matchword,
325 bool all, bool forward, bool findnext)
328 ss << replace << '\n'
330 << int(casesensitive) << ' '
331 << int(matchword) << ' '
333 << int(forward) << ' '
339 bool lyxfind(BufferView * bv, FuncRequest const & ev)
341 if (!bv || ev.action() != LFUN_WORD_FIND)
344 //lyxerr << "find called, cmd: " << ev << endl;
346 // data is of the form
348 // <casesensitive> <matchword> <forward>"
350 docstring howto = split(ev.argument(), search, '\n');
352 bool casesensitive = parse_bool(howto);
353 bool matchword = parse_bool(howto);
354 bool forward = parse_bool(howto);
356 return findOne(bv, search, casesensitive, matchword, forward, true, true);
360 bool lyxreplace(BufferView * bv,
361 FuncRequest const & ev, bool has_deleted)
363 if (!bv || ev.action() != LFUN_WORD_REPLACE)
366 // data is of the form
369 // <casesensitive> <matchword> <all> <forward> <findnext>"
372 docstring howto = split(ev.argument(), rplc, '\n');
373 howto = split(howto, search, '\n');
375 bool casesensitive = parse_bool(howto);
376 bool matchword = parse_bool(howto);
377 bool all = parse_bool(howto);
378 bool forward = parse_bool(howto);
379 bool findnext = howto.empty() ? true : parse_bool(howto);
384 int replace_count = 0;
386 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
387 update = replace_count > 0;
390 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
392 replace_count = rv.second;
395 Buffer const & buf = bv->buffer();
397 // emit message signal.
398 buf.message(_("String not found."));
400 if (replace_count == 0) {
401 buf.message(_("String found."));
402 } else if (replace_count == 1) {
403 buf.message(_("String has been replaced."));
405 docstring const str =
406 bformat(_("%1$d strings have been replaced."), replace_count);
410 } else if (findnext) {
411 // if we have deleted characters, we do not replace at all, but
412 // rather search for the next occurence
413 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
416 bv->message(_("String not found."));
422 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
424 for (; cur; cur.forwardPos())
425 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
429 DocIterator cur_orig(bv->cursor());
430 docstring q = _("End of file reached while searching forward.\n"
431 "Continue searching from the beginning?");
432 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
433 q, 0, 1, _("&Yes"), _("&No"));
434 if (wrap_answer == 0) {
435 bv->cursor().clear();
436 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
437 bv->clearSelection();
438 cur.setCursor(bv->cursor().selectionBegin());
439 if (findNextChange(bv, cur, false))
442 bv->cursor().setCursor(cur_orig);
449 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
451 for (cur.backwardPos(); cur; cur.backwardPos()) {
452 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
457 DocIterator cur_orig(bv->cursor());
458 docstring q = _("Beginning of file reached while searching backward.\n"
459 "Continue searching from the end?");
460 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
461 q, 0, 1, _("&Yes"), _("&No"));
462 if (wrap_answer == 0) {
463 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
464 bv->cursor().backwardPos();
465 bv->clearSelection();
466 cur.setCursor(bv->cursor().selectionBegin());
467 if (findPreviousChange(bv, cur, false))
470 bv->cursor().setCursor(cur_orig);
477 bool selectChange(Cursor & cur, bool forward)
479 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
481 Change ch = cur.paragraph().lookupChange(cur.pos());
483 CursorSlice tip1 = cur.top();
484 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
485 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
486 if (!ch2.isSimilarTo(ch))
489 CursorSlice tip2 = cur.top();
490 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
492 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
493 if (!ch2.isSimilarTo(ch)) {
494 // take a step forward to correctly set the selection
502 cur.bv().mouseSetCursor(cur, false);
504 cur.bv().mouseSetCursor(cur, true);
512 bool findChange(BufferView * bv, bool forward)
515 cur.setCursor(forward ? bv->cursor().selectionEnd()
516 : bv->cursor().selectionBegin());
517 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
518 return selectChange(cur, forward);
523 bool findNextChange(BufferView * bv)
525 return findChange(bv, true);
529 bool findPreviousChange(BufferView * bv)
531 return findChange(bv, false);
538 typedef vector<pair<string, string> > Escapes;
540 /// A map of symbols and their escaped equivalent needed within a regex.
541 /// @note Beware of order
542 Escapes const & get_regexp_escapes()
544 typedef std::pair<std::string, std::string> P;
546 static Escapes escape_map;
547 if (escape_map.empty()) {
548 escape_map.push_back(P("$", "_x_$"));
549 escape_map.push_back(P("{", "_x_{"));
550 escape_map.push_back(P("}", "_x_}"));
551 escape_map.push_back(P("[", "_x_["));
552 escape_map.push_back(P("]", "_x_]"));
553 escape_map.push_back(P("(", "_x_("));
554 escape_map.push_back(P(")", "_x_)"));
555 escape_map.push_back(P("+", "_x_+"));
556 escape_map.push_back(P("*", "_x_*"));
557 escape_map.push_back(P(".", "_x_."));
558 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
559 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
560 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
561 escape_map.push_back(P("_x_", "\\"));
566 /// A map of lyx escaped strings and their unescaped equivalent.
567 Escapes const & get_lyx_unescapes()
569 typedef std::pair<std::string, std::string> P;
571 static Escapes escape_map;
572 if (escape_map.empty()) {
573 escape_map.push_back(P("\\%", "%"));
574 escape_map.push_back(P("\\mathcircumflex ", "^"));
575 escape_map.push_back(P("\\mathcircumflex", "^"));
576 escape_map.push_back(P("\\backslash ", "\\"));
577 escape_map.push_back(P("\\backslash", "\\"));
578 escape_map.push_back(P("\\\\{", "_x_<"));
579 escape_map.push_back(P("\\\\}", "_x_>"));
580 escape_map.push_back(P("\\sim ", "~"));
581 escape_map.push_back(P("\\sim", "~"));
586 /// A map of escapes turning a regexp matching text to one matching latex.
587 Escapes const & get_regexp_latex_escapes()
589 typedef std::pair<std::string, std::string> P;
591 static Escapes escape_map;
592 if (escape_map.empty()) {
593 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
594 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
595 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
596 escape_map.push_back(P("\\[", "\\{\\[\\}"));
597 escape_map.push_back(P("\\]", "\\{\\]\\}"));
598 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
599 escape_map.push_back(P("%", "\\\\\\%"));
604 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
605 ** the found occurrence were escaped.
607 string apply_escapes(string s, Escapes const & escape_map)
609 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
610 Escapes::const_iterator it;
611 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
612 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
613 unsigned int pos = 0;
614 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
615 s.replace(pos, it->first.length(), it->second);
616 LYXERR(Debug::FIND, "After escape: " << s);
617 pos += it->second.length();
618 // LYXERR(Debug::FIND, "pos: " << pos);
621 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
626 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
627 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
628 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
629 string escape_for_regex(string s, bool match_latex)
632 while (pos < s.size()) {
633 size_t new_pos = s.find("\\regexp{", pos);
634 if (new_pos == string::npos)
636 LYXERR(Debug::FIND, "new_pos: " << new_pos);
637 string t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
638 LYXERR(Debug::FIND, "t [lyx]: " << t);
639 t = apply_escapes(t, get_regexp_escapes());
640 LYXERR(Debug::FIND, "t [rxp]: " << t);
641 s.replace(pos, new_pos - pos, t);
642 new_pos = pos + t.size();
643 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
644 LYXERR(Debug::FIND, "new_pos: " << new_pos);
645 if (new_pos == s.size())
647 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
648 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
649 LYXERR(Debug::FIND, "end_pos: " << end_pos);
650 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
651 LYXERR(Debug::FIND, "t in regexp : " << t);
652 t = apply_escapes(t, get_lyx_unescapes());
653 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
655 t = apply_escapes(t, get_regexp_latex_escapes());
656 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
658 if (end_pos == s.size()) {
659 s.replace(new_pos, end_pos - new_pos, t);
660 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
663 s.replace(new_pos, end_pos + 13 - new_pos, t);
664 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
665 pos = new_pos + t.size();
666 LYXERR(Debug::FIND, "pos: " << pos);
672 /// Wrapper for lyx::regex_replace with simpler interface
673 bool regex_replace(string const & s, string & t, string const & searchstr,
674 string const & replacestr)
676 lyx::regex e(searchstr, regex_constants::ECMAScript);
678 ostream_iterator<char, char> it(oss);
679 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
680 // tolerate t and s be references to the same variable
681 bool rv = (s != oss.str());
687 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
689 ** Verify that closed braces exactly match open braces. This avoids that, for example,
690 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
693 ** Number of open braces that must remain open at the end for the verification to succeed.
695 bool braces_match(string::const_iterator const & beg,
696 string::const_iterator const & end,
700 string::const_iterator it = beg;
701 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
702 for (; it != end; ++it) {
703 // Skip escaped braces in the count
708 } else if (*it == '{') {
710 } else if (*it == '}') {
711 if (open_pars == 0) {
712 LYXERR(Debug::FIND, "Found unmatched closed brace");
718 if (open_pars != unmatched) {
719 LYXERR(Debug::FIND, "Found " << open_pars
720 << " instead of " << unmatched
721 << " unmatched open braces at the end of count");
724 LYXERR(Debug::FIND, "Braces match as expected");
729 /** The class performing a match between a position in the document and the FindAdvOptions.
731 class MatchStringAdv {
733 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
735 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
736 ** constructor as opt.search, under the opt.* options settings.
739 ** If set, then match is searched only against beginning of text starting at cur.
740 ** If unset, then match is searched anywhere in text starting at cur.
743 ** The length of the matching text, or zero if no match was found.
745 int operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
750 /// first buffer on which search was started
751 lyx::Buffer * const p_first_buf;
753 FindAndReplaceOptions const & opt;
756 /// Auxiliary find method (does not account for opt.matchword)
757 int findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
759 /** Normalize a stringified or latexified LyX paragraph.
763 ** <li>if search is not casesensitive, then lowercase the string;
764 ** <li>remove any newline at begin or end of the string;
765 ** <li>replace any newline in the middle of the string with a simple space;
766 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
769 ** @todo Normalization should also expand macros, if the corresponding
770 ** search option was checked.
772 string normalize(docstring const & s, bool hack_braces) const;
773 // normalized string to search
774 string par_as_string;
775 // regular expression to use for searching
777 // same as regexp, but prefixed with a ".*"
779 // leading format material as string
780 string lead_as_string;
781 // par_as_string after removal of lead_as_string
782 string par_as_string_nolead;
783 // unmatched open braces in the search string/regexp
785 // number of (.*?) subexpressions added at end of search regexp for closing
786 // environments, math mode, styles, etc...
788 // Are we searching with regular expressions ?
793 static docstring buffer_to_latex(Buffer & buffer)
795 OutputParams runparams(&buffer.params().encoding());
796 odocstringstream ods;
798 runparams.nice = true;
799 runparams.flavor = OutputParams::LATEX;
800 runparams.linelen = 80; //lyxrc.plaintext_linelen;
801 // No side effect of file copying and image conversion
802 runparams.dryrun = true;
803 runparams.for_search = true;
804 pit_type const endpit = buffer.paragraphs().size();
805 for (pit_type pit = 0; pit != endpit; ++pit) {
806 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
807 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
813 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
816 if (!opt.ignoreformat) {
817 str = buffer_to_latex(buffer);
819 OutputParams runparams(&buffer.params().encoding());
820 runparams.nice = true;
821 runparams.flavor = OutputParams::LATEX;
822 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
823 runparams.dryrun = true;
824 runparams.for_search = true;
825 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
826 Paragraph const & par = buffer.paragraphs().at(pit);
827 LYXERR(Debug::FIND, "Adding to search string: '"
828 << par.asString(pos_type(0), par.size(),
829 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
832 str += par.asString(pos_type(0), par.size(),
833 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
841 /// Return separation pos between the leading material and the rest
842 static size_t identifyLeading(string const & s)
845 // @TODO Support \item[text]
846 // Kornel: Added textsl, textsf, textit, texttt and noun
847 // + allow to seach for colored text too
848 while (regex_replace(t, t, REGEX_BOS "\\\\(((emph|noun|text(bf|sl|sf|it|tt))|((textcolor|foreignlanguage)\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part)\\*?)\\{", "")
849 || regex_replace(t, t, REGEX_BOS "\\$", "")
850 || regex_replace(t, t, REGEX_BOS "\\\\\\[ ", "")
851 || regex_replace(t, t, REGEX_BOS "\\\\item ", "")
852 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
854 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
859 * Given a latexified string, retrieve some handled features
860 * The features of the regex will later be compared with the features
861 * of the searched text. If the regex features are not a
862 * subset of the analized, then, in not format ignoring search
863 * we can early stop the search in the relevant inset.
865 typedef map<string, bool> Features;
867 static Features identifyFeatures(string const & s)
869 static regex const feature("\\\\(([a-z]+(\\{([a-z]+)\\}|\\*)?))\\{");
870 static regex const valid("^(((emph|noun|text(bf|sl|sf|it|tt)|(textcolor|foreignlanguage)\\{[a-z]+\\})|item |(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part)\\*?)$");
875 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
878 if (sub.str(1).compare("regexp") == 0) {
882 string token = sub.str(1);
884 if (regex_match(token, sub2, valid)) {
892 if (sub.str(1).compare("endregexp") == 0) {
902 * Faster search for the related closing parenthesis
904 static int findclosing(string p, int start, int end)
908 int lastunclosed = start-1;
909 for (int i = start; i < end; i += 1 + skip) {
913 if (c == '\\') skip = 1;
919 if (depth == 0) return(i);
923 return(0 - lastunclosed);
927 * Discard any info for char sizes for now.
929 static string removefontinfo(string par)
931 // Remove fontsizes, inputencoding
934 static regex const sizescodings("(\\\\(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|inputencoding\\{[^\\}]*})(\b|(\\{(\\{\\})?\\})?(%\\n)?))");
935 for (sregex_iterator it(par.begin(), par.end(), sizescodings), end; it != end; ++it) {
937 string token = sub.str(1);
938 fpars.push_back(token);
940 for (list<string>::const_iterator li = fpars.begin(); li != fpars.end(); ++li) {
944 int ic; // Position of closing part e.g. '}'
945 while ((f = par.find(token, firstpos)) >= 0) {
946 size_t ssize = token.length();
947 int parcount = 0; // how many '{}' can be removed
954 // here '{' preceedes
955 ic = findclosing(par, f + ssize, par.length());
958 else if ((f == 2) && (par[f-2] == '{')) {
959 if ((ic < 0) || (par[ic+1] == '}'))
963 } else while (f > parcount + 1) {
964 if (par[f-2-parcount] != '{')
967 if ((ic > 0) && (par[ic+parcount] != '}'))
975 ic = par.length() - parcount;
976 par = par.substr(0, f-parcount) + par.substr(f+ssize, ic+parcount-f-ssize) + par.substr(ic+parcount);
992 /*LangInfo(LangInfo &orig) :
994 maxoffset(orig.maxoffset),
995 search(orig.search) {valid = Invalid;}
997 LangInfo(string par, string search1 = "", int start = 0, int end = -1)
1004 _tokenstart = start;
1005 if (end > int(par.length())) {
1006 maxoffset = par.length();
1011 maxoffset = par.length();
1012 if (!search1.empty())
1015 bool nextInfo(); // of the same type, from the last start in the same reagion
1016 bool firstInfo(string search, int datastart);
1017 void setDataEnd(int value);
1018 void setDataStart(int value);
1019 int getDataStart() { return _dataStart;};
1020 string name() { return _search;};
1021 string lasttoken() { if (valid == Valid) return _foundtoken; else return "";};
1022 int getStart() { return _tokenstart;};
1023 int getTokenEnd() { return _tokenend;};
1024 int getEnd() { return _dataEnd;};
1025 bool isValid() { return (valid == Valid); };
1026 void process(ostringstream &os);
1027 void output(ostringstream &os, int);
1028 void addIntervall(int upper);
1029 void addIntervall(int low, int upper); /* if explicit */
1030 void handleParentheses(int lastpos);
1031 string show(int lastpos);
1044 int actualdeptindex;
1045 int ignoreIntervalls[10][2];
1049 void LangInfo::setDataEnd(int dataend)
1051 if (dataend < _tokenend) {
1052 _dataEnd = _tokenend;
1053 // cout << "Wrong data start, too low\n";
1055 else if (size_t(dataend) > par.length()) {
1056 // cout << "Wrong data start, too high\n";
1057 _dataEnd = par.length();
1063 void LangInfo::setDataStart(int datastart)
1065 if (datastart < _tokenend) {
1066 _dataStart = _tokenend;
1067 // cout << "Wrong data start, too low\n";
1069 else if (size_t(datastart) > par.length()) {
1070 // cout << "Wrong data start, too high\n";
1071 _dataStart = par.length();
1074 _dataStart = datastart;
1075 //cout << "found entry at " << _tokenstart << "\n";
1076 actualdeptindex = 1; /* == Number of open brases */
1077 depts[0] = _dataStart;
1079 depts[1] = _dataStart;
1081 ignoreIntervalls[ignoreidx][0] = _dataStart;
1082 if ((par[_dataStart] == '{') && (par[_dataStart+1] == '}')) {
1083 // First candidates to be ignored
1084 ignoreIntervalls[ignoreidx][1] = _dataStart+2;
1087 ignoreIntervalls[ignoreidx][1] = _dataStart;
1090 void LangInfo::handleParentheses(int lastpos)
1093 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
1097 if (c == '\\') skip = 1;
1098 else if (c == '{') {
1100 depts[actualdeptindex] = i+1;
1101 closes[actualdeptindex] = -1;
1103 else if (c == '}') {
1104 if (actualdeptindex <= 0) {
1105 LYXERR0("ERROR ERROR ERROR"); /* should never happen! */
1108 closes[actualdeptindex] = i+1;
1115 void LangInfo::addIntervall(int low, int upper)
1118 if (low == upper) return;
1119 for (idx = ignoreidx+1; idx > 0; --idx) {
1120 if (low > ignoreIntervalls[idx-1][1]) {
1124 if (idx > ignoreidx) {
1125 ignoreIntervalls[idx][0] = low;
1126 ignoreIntervalls[idx][1] = upper;
1129 // Expand only if one of the new bounds is inside the interwall
1130 if (((low <= ignoreIntervalls[idx][1]) && (low >= ignoreIntervalls[idx][0])) ||
1131 ((upper <= ignoreIntervalls[idx][1]) && (upper >= ignoreIntervalls[idx][0]))) {
1132 if (low < ignoreIntervalls[idx][0])
1133 ignoreIntervalls[idx][0] = low;
1134 if (upper > ignoreIntervalls[idx][1])
1135 ignoreIntervalls[idx][1] = upper;
1138 ignoreidx = idx; /* because upper is in all cases bigger */
1141 void LangInfo::addIntervall(int upper)
1144 if (actualdeptindex >= 0)
1145 low = depts[actualdeptindex]; /* the position of last unclosed '{' */
1147 LYXERR0("ERROR ERROR ERROR2");
1150 addIntervall(low, upper);
1153 string LangInfo::show(int lastpos)
1157 os << par.substr(_tokenstart, _tokenend - _tokenstart);
1159 for (int i = _dataStart; i < lastpos;) {
1160 if (i <= ignoreIntervalls[idx][0]) {
1161 os << par.substr(i, ignoreIntervalls[idx][0] - i);
1162 i = ignoreIntervalls[idx][1];
1165 if (idx > ignoreidx) {
1166 os << par.substr(i, lastpos-i);
1170 for (int i = actualdeptindex; i > 0; --i)
1175 void LangInfo::output(ostringstream &os, int lastpos)
1177 // get number of chars to output
1178 int idx = 0; /* int intervalls */
1180 for (int i = _dataStart; i < lastpos;) {
1181 if (i <= ignoreIntervalls[idx][0]) {
1182 count += ignoreIntervalls[idx][0] - i;
1183 i = ignoreIntervalls[idx][1];
1186 if (idx > ignoreidx) {
1191 //cout << "Number of output chars would be " << count + actualdeptindex << "\n";
1193 // Now the acual data
1194 os << par.substr(_tokenstart, _tokenend - _tokenstart);
1196 for (int i = _dataStart; i < lastpos;) {
1197 if (i <= ignoreIntervalls[idx][0]) {
1198 os << par.substr(i, ignoreIntervalls[idx][0] - i);
1199 i = ignoreIntervalls[idx][1];
1202 if (idx > ignoreidx) {
1203 os << par.substr(i, lastpos-i);
1207 for (int i = actualdeptindex; i > 0; --i)
1210 handleParentheses(lastpos);
1213 bool LangInfo::nextInfo()
1215 int start = _tokenstart;
1217 // cout << par << "\n";
1218 if (valid == Invalid)
1219 _dataEnd = _tokenstart;
1220 else if (valid == LastValid)
1222 // cout << "Start search at " << _tokenclose << " for \"" << _search << "\n";
1223 size_t foundstart = par.find(_search, _dataEnd);
1224 if (foundstart == string::npos) {
1227 return false; // not found
1229 if (foundstart >= maxoffset)
1232 int closelang = findclosing(par, start + _search.length(), maxoffset);
1235 if (size_t(closelang) >= maxoffset)
1237 if (par[closelang] != '}')
1240 _foundtoken = par.substr(start, closelang - start + 2);
1241 _tokenstart = start;
1242 _tokenend = closelang+2;
1243 setDataStart(_tokenend);
1244 closelang = findclosing(par, _dataStart, maxoffset);
1245 if (closelang < 0) {
1246 _dataEnd = maxoffset;
1250 _dataEnd = closelang;
1256 bool LangInfo::firstInfo(string search1, int datastart)
1258 if (!search1.empty()) {
1259 if (_search.compare(search1) != 0) {
1260 _tokenstart = datastart;
1268 void LangInfo::process(ostringstream &os)
1270 LangInfo color(*this);
1271 (void) color.firstInfo("\\textcolor{", _dataStart);
1272 while (color.isValid() && (color.getStart() < _dataEnd)) {
1273 bool isEmpty = false;
1274 if (color.getDataStart() == color.getEnd()) {
1275 // Empty, e.g. par[color.getDataStart()] == '}'
1278 else if ((par[color.getDataStart()] == '{') && (par[color.getDataStart()+1] == '}')) {
1279 // color starts with '{}', discard it
1280 if (color.getDataStart()+2 == color.getEnd())
1283 // discard the first '{}'
1284 addIntervall(color.getDataStart(), color.getDataStart()+2);
1288 // it is emty, so ignore and go to next color
1289 addIntervall(color.getStart(), color.getEnd()+1);
1292 if (par[color.getStart()-1] != '{') {
1293 output(os, color.getStart());
1294 addIntervall(color.getStart());
1296 // Check if color empty
1297 output(os, color.getEnd()+1);
1298 addIntervall(color.getEnd()+1);
1300 for (int i = color.getEnd()+1; par[i] == '}'; i++) {
1301 handleParentheses(i+1);
1306 if (par[_dataEnd] != '}')
1307 output(os, _dataEnd-1);
1309 output(os, _dataEnd);
1313 * Called only if the par starts with lang spec
1316 string splitForColors(string par) {
1318 LangInfo firstLanguage(par, "\\foreignlanguage{");
1319 if (firstLanguage.firstInfo("\\foreignlanguage{", 0)) {
1320 LangInfo nextLanguage(firstLanguage);
1321 nextLanguage.setDataEnd(firstLanguage.getDataStart());
1322 if (nextLanguage.firstInfo("\\foreignlanguage{", firstLanguage.getTokenEnd())) {
1323 firstLanguage.setDataEnd(nextLanguage.getStart());
1325 firstLanguage.process(os);
1326 while (nextLanguage.isValid()) {
1327 nextLanguage.process(os);
1328 // To handle the gap, we need the end of last languuage to start of next
1329 int gapstart = nextLanguage.getEnd()+1;
1331 nextLanguage.nextInfo();
1332 if (nextLanguage.isValid())
1333 gapend = nextLanguage.getStart();
1335 gapend = par.length();
1336 // Now handle the gap, if there is one
1337 if (gapend > gapstart) {
1338 // cout << "Gap found, size = " << gapend - gapstart << "\n";
1339 firstLanguage.setDataEnd(gapend);
1340 firstLanguage.setDataStart(gapstart);
1341 firstLanguage.process(os);
1345 string s = os.str();
1350 * Try to unify the language specs in the latexified text.
1351 * Resulting modified string is set to "", if
1352 * the searched tex does not contain all the features in the search pattern
1354 static string correctlanguagesetting(string par, bool from_regex, bool withformat)
1356 static Features regex_f;
1357 static int missed = 0;
1358 static bool regex_with_format = false;
1360 int parlen = par.length();
1362 while ((parlen > 0) && (par[parlen-1] == '\n')) {
1365 string result = removefontinfo(par.substr(0, parlen));
1366 result = splitForColors(result);
1367 LYXERR(Debug::FIND, "Converted: \"" << result << "\"");
1368 bool handle_colors = false;
1372 regex_f = identifyFeatures(result);
1373 string features = "";
1374 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
1375 string a = it->first;
1376 regex_with_format = true;
1377 if (a.compare(0,10,"textcolor{") == 0)
1378 handle_colors = true;
1379 features += " " + a;
1380 // LYXERR0("Identified regex format:" << a);
1382 LYXERR(Debug::FIND, "Identified Features" << features);
1385 } else if (regex_with_format) {
1386 Features info = identifyFeatures(result);
1387 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
1388 string a = it->first;
1389 bool b = it->second;
1390 if (b && ! info[a]) {
1392 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
1395 else if (a.compare(0,10,"textcolor{") == 0)
1396 handle_colors = true;
1400 // LYXERR0("No regex formats");
1402 // remove possible disturbing macros
1403 while (regex_replace(result, result, "\\\\(noindent )", ""))
1405 // Either not found language spec,or is single and closed spec or empty
1408 static regex const parreg("((\\n)?\\\\[a-z]+par)\\{");
1412 for (sregex_iterator it(result.begin(), result.end(), parreg), end; it != end; ++it) {
1414 string token = sub.str(1);
1415 pars.push_back(token);
1417 for (list<string>::const_iterator li = pars.begin(); li != pars.end(); ++li) {
1419 int ti = result.find(token);
1420 int tokensize = token.size() + 1;
1422 int tc = findclosing(result, ti + tokensize, result.size());
1424 result = result.substr(0, ti) + result.substr(ti + tokensize, tc - ti -tokensize) + result.substr(tc+1);
1428 if (handle_colors) {
1429 while (regex_replace(result, result, "(\\{\\\\textcolor\\{[a-z]+\\}\\{)\\s*\\{\\}\\s*", "$1"));
1430 while (regex_replace(result, result, "\\{\\\\textcolor\\{[a-z]+\\}\\{\\s*\\}\\s*\\}", ""));
1436 // Remove trailing closure of math, macros and environments, so to catch parts of them.
1437 static int identifyClosing(string & t)
1439 int open_braces = 0;
1441 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
1442 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
1444 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]" REGEX_EOS, "$1"))
1446 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
1448 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
1458 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
1459 : p_buf(&buf), p_first_buf(&buf), opt(opt)
1461 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
1462 docstring const & ds = stringifySearchBuffer(find_buf, opt);
1463 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
1464 // When using regexp, braces are hacked already by escape_for_regex()
1465 par_as_string = normalize(ds, !use_regexp);
1467 close_wildcards = 0;
1469 size_t lead_size = 0;
1470 // correct the language settings
1471 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
1472 if (opt.ignoreformat) {
1474 // if par_as_string_nolead were emty,
1475 // the following call to findAux will always *find* the string
1476 // in the checked data, and thus always using the slow
1477 // examining of the current text part.
1478 par_as_string_nolead = par_as_string;
1481 lead_size = identifyLeading(par_as_string);
1482 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
1483 lead_as_string = par_as_string.substr(0, lead_size);
1484 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
1488 open_braces = identifyClosing(par_as_string);
1489 identifyClosing(par_as_string_nolead);
1490 LYXERR(Debug::FIND, "Open braces: " << open_braces);
1491 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
1493 string lead_as_regexp;
1494 if (lead_size > 0) {
1495 // @todo No need to search for \regexp{} insets in leading material
1496 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
1497 par_as_string = par_as_string_nolead;
1498 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
1499 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
1501 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
1502 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
1503 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
1505 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
1506 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
1507 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
1508 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
1509 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
1510 || regex_replace(par_as_string, par_as_string,
1511 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
1512 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
1513 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
1517 if (!opt.ignoreformat) {
1518 // Remove extra '\}' at end
1519 while ( regex_replace(par_as_string, par_as_string, "(.*)\\\\}$", "$1")) {
1523 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
1524 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
1525 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
1526 // replace '[^...]' with '[^...\}\{\\]'
1527 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
1528 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
1529 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
1531 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
1533 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
1534 LYXERR(Debug::FIND, "Open braces: " << open_braces);
1535 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
1536 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
1538 // If entered regexp must match at begin of searched string buffer
1539 // Kornel: Added parentheses to use $1 for size of the leading string
1543 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
1544 // Unfortunately is '\1', '\2', etc not working for strings with extra format
1545 // so the convert has no effect in that case
1546 for (int i = 8; i > 0; --i) {
1547 string orig = "\\\\" + std::to_string(i);
1548 string dest = "\\" + std::to_string(i+1);
1549 while (regex_replace(par_as_string, par_as_string, orig, dest));
1551 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
1552 regexp2_str = "(" + lead_as_regexp + ").*" + par_as_string;
1554 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
1555 regexp = lyx::regex(regexp_str);
1557 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
1558 regexp2 = lyx::regex(regexp2_str);
1563 int MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
1566 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
1569 docstring docstr = stringifyFromForSearch(opt, cur, len);
1570 string str = normalize(docstr, true);
1571 if (str.empty()) return(-1);
1572 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
1573 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
1576 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
1577 regex const *p_regexp;
1578 regex_constants::match_flag_type flags;
1580 flags = regex_constants::match_continuous;
1583 flags = regex_constants::match_default;
1584 p_regexp = ®exp2;
1586 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
1587 if (re_it == sregex_iterator())
1589 match_results<string::const_iterator> const & m = *re_it;
1591 if (0) { // Kornel Benko: DO NOT CHECKK
1592 // Check braces on the segment that matched the entire regexp expression,
1593 // plus the last subexpression, if a (.*?) was inserted in the constructor.
1594 if (!braces_match(m[0].first, m[0].second, open_braces))
1598 // Check braces on segments that matched all (.*?) subexpressions,
1599 // except the last "padding" one inserted by lyx.
1600 for (size_t i = 1; i < m.size() - 1; ++i)
1601 if (!braces_match(m[i].first, m[i].second, open_braces))
1604 // Exclude from the returned match length any length
1605 // due to close wildcards added at end of regexp
1606 // and also the length of the leading (e.g. '\emph{')
1608 // Whole found string, including the leading: m[0].second - m[0].first
1609 // Size of the leading string: m[1].second - m[1].first
1610 int leadingsize = 0;
1612 leadingsize = m[1].second - m[1].first;
1614 for (size_t i = 0; i < m.size(); i++) {
1615 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
1617 if (close_wildcards == 0)
1618 result = m[0].second - m[0].first;
1621 result = m[m.size() - close_wildcards].first - m[0].first;
1623 if (result > leadingsize)
1624 result -= leadingsize;
1630 // else !use_regexp: but all code paths above return
1631 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
1632 << par_as_string << "', str='" << str << "'");
1633 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
1634 << lead_as_string << "', par_as_string_nolead='"
1635 << par_as_string_nolead << "'");
1638 LYXERR(Debug::FIND, "size=" << par_as_string.size()
1639 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
1640 if (str.substr(0, par_as_string.size()) == par_as_string)
1641 return par_as_string.size();
1643 size_t pos = str.find(par_as_string_nolead);
1644 if (pos != string::npos)
1645 return par_as_string.size();
1651 int MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
1653 int res = findAux(cur, len, at_begin);
1655 "res=" << res << ", at_begin=" << at_begin
1656 << ", matchword=" << opt.matchword
1657 << ", inTexted=" << cur.inTexted());
1658 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
1660 Paragraph const & par = cur.paragraph();
1661 bool ws_left = (cur.pos() > 0)
1662 ? par.isWordSeparator(cur.pos() - 1)
1664 bool ws_right = (cur.pos() + res < par.size())
1665 ? par.isWordSeparator(cur.pos() + res)
1668 "cur.pos()=" << cur.pos() << ", res=" << res
1669 << ", separ: " << ws_left << ", " << ws_right
1671 if (ws_left && ws_right)
1677 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
1680 if (! opt.casesensitive)
1681 t = lyx::to_utf8(lowercase(s));
1683 t = lyx::to_utf8(s);
1684 // Remove \n at begin
1685 while (!t.empty() && t[0] == '\n')
1688 while (!t.empty() && t[t.size() - 1] == '\n')
1689 t = t.substr(0, t.size() - 1);
1691 // Replace all other \n with spaces
1692 while ((pos = t.find("\n")) != string::npos)
1693 t.replace(pos, 1, " ");
1694 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
1695 // Kornel: Added textsl, textsf, textit, texttt and noun
1696 // + allow to seach for colored text too
1697 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
1698 while (regex_replace(t, t, "\\\\((emph|noun|text(bf|sl|sf|it|tt|color\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part)\\*?)(\\{(\\{\\})?\\})+", ""))
1699 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
1701 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor)\\{[a-z]+\\}(\\{(\\\\item |\\{\\})?\\})+", ""));
1702 // FIXME - check what preceeds the brace
1704 if (opt.ignoreformat)
1705 while (regex_replace(t, t, "\\{", "_x_<")
1706 || regex_replace(t, t, "\\}", "_x_>"))
1707 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
1709 while (regex_replace(t, t, "\\\\\\{", "_x_<")
1710 || regex_replace(t, t, "\\\\\\}", "_x_>"))
1711 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
1718 docstring stringifyFromCursor(DocIterator const & cur, int len)
1720 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
1721 if (cur.inTexted()) {
1722 Paragraph const & par = cur.paragraph();
1723 // TODO what about searching beyond/across paragraph breaks ?
1724 // TODO Try adding a AS_STR_INSERTS as last arg
1725 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
1726 int(par.size()) : cur.pos() + len;
1727 OutputParams runparams(&cur.buffer()->params().encoding());
1728 runparams.nice = true;
1729 runparams.flavor = OutputParams::LATEX;
1730 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
1731 // No side effect of file copying and image conversion
1732 runparams.dryrun = true;
1733 LYXERR(Debug::FIND, "Stringifying with cur: "
1734 << cur << ", from pos: " << cur.pos() << ", end: " << end);
1735 return par.asString(cur.pos(), end,
1736 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
1738 } else if (cur.inMathed()) {
1740 CursorSlice cs = cur.top();
1741 MathData md = cs.cell();
1742 MathData::const_iterator it_end =
1743 (( len == -1 || cs.pos() + len > int(md.size()))
1745 : md.begin() + cs.pos() + len );
1746 for (MathData::const_iterator it = md.begin() + cs.pos();
1748 s = s + asString(*it);
1749 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
1752 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
1757 /** Computes the LaTeX export of buf starting from cur and ending len positions
1758 * after cur, if len is positive, or at the paragraph or innermost inset end
1761 docstring latexifyFromCursor(DocIterator const & cur, int len)
1763 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
1764 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
1765 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
1766 Buffer const & buf = *cur.buffer();
1768 odocstringstream ods;
1770 OutputParams runparams(&buf.params().encoding());
1771 runparams.nice = false;
1772 runparams.flavor = OutputParams::LATEX;
1773 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
1774 // No side effect of file copying and image conversion
1775 runparams.dryrun = true;
1776 runparams.for_search = true;
1778 if (cur.inTexted()) {
1779 // @TODO what about searching beyond/across paragraph breaks ?
1780 pos_type endpos = cur.paragraph().size();
1781 if (len != -1 && endpos > cur.pos() + len)
1782 endpos = cur.pos() + len;
1783 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
1784 string(), cur.pos(), endpos);
1785 string s = correctlanguagesetting(lyx::to_utf8(ods.str()), false, false);
1786 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
1787 return(lyx::from_utf8(s));
1788 } else if (cur.inMathed()) {
1789 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
1790 for (int s = cur.depth() - 1; s >= 0; --s) {
1791 CursorSlice const & cs = cur[s];
1792 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
1794 cs.asInsetMath()->asHullInset()->header_write(ws);
1799 CursorSlice const & cs = cur.top();
1800 MathData md = cs.cell();
1801 MathData::const_iterator it_end =
1802 ((len == -1 || cs.pos() + len > int(md.size()))
1804 : md.begin() + cs.pos() + len);
1805 for (MathData::const_iterator it = md.begin() + cs.pos();
1807 ods << asString(*it);
1809 // Retrieve the math environment type, and add '$' or '$]'
1810 // or others (\end{equation}) accordingly
1811 for (int s = cur.depth() - 1; s >= 0; --s) {
1812 CursorSlice const & cs2 = cur[s];
1813 InsetMath * inset = cs2.asInsetMath();
1814 if (inset && inset->asHullInset()) {
1816 inset->asHullInset()->footer_write(ws);
1820 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
1822 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
1828 /** Finalize an advanced find operation, advancing the cursor to the innermost
1829 ** position that matches, plus computing the length of the matching text to
1832 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
1834 // Search the foremost position that matches (avoids find of entire math
1835 // inset when match at start of it)
1837 DocIterator old_cur(cur.buffer());
1839 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
1843 } while (cur && cur.depth() > d && match(cur) > 0);
1845 if (match(cur) <= 0) return 0;
1846 LYXERR(Debug::FIND, "Ok");
1848 // Compute the match length
1850 if (cur.pos() + len > cur.lastpos())
1852 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
1853 while (cur.pos() + len <= cur.lastpos() && match(cur, len) <= 0) {
1855 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
1857 // Length of matched text (different from len param)
1858 int old_len = match(cur, len);
1859 if (old_len < 0) old_len = 0;
1861 // Greedy behaviour while matching regexps
1862 while ((new_len = match(cur, len + 1)) > old_len) {
1865 LYXERR(Debug::FIND, "verifying match with len = " << len);
1872 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
1876 while (!theApp()->longOperationCancelled() && cur) {
1877 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
1878 int match_len = match(cur, -1, false);
1879 LYXERR(Debug::FIND, "match_len: " << match_len);
1880 if (match_len > 0) {
1881 int match_len_zero_count = 0;
1882 for (; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
1883 LYXERR(Debug::FIND, "Advancing cur: " << cur);
1884 int match_len2 = match(cur);
1885 LYXERR(Debug::FIND, "match_len2: " << match_len2);
1886 if (match_len2 > 0) {
1887 // Sometimes in finalize we understand it wasn't a match
1888 // and we need to continue the outest loop
1889 int len = findAdvFinalize(cur, match);
1894 if (match_len2 >= 0) {
1895 if (match_len2 == 0)
1896 match_len_zero_count++;
1898 match_len_zero_count = 0;
1901 if (++match_len_zero_count > 3) {
1902 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
1903 match_len_zero_count = 0;
1911 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
1912 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
1915 // This should exit nested insets, if any, or otherwise undefine the currsor.
1916 cur.pos() = cur.lastpos();
1917 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
1925 /// Find the most backward consecutive match within same paragraph while searching backwards.
1926 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
1928 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
1929 DocIterator tmp_cur = cur;
1930 int len = findAdvFinalize(tmp_cur, match);
1931 Inset & inset = cur.inset();
1932 for (; cur != cur_begin; cur.backwardPos()) {
1933 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
1934 DocIterator new_cur = cur;
1935 new_cur.backwardPos();
1936 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur))
1938 int new_len = findAdvFinalize(new_cur, match);
1943 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
1949 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
1953 // Backup of original position
1954 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
1955 if (cur == cur_begin)
1958 DocIterator cur_orig(cur);
1959 bool pit_changed = false;
1962 bool found_match = match(cur, -1, false);
1966 cur.pos() = cur.lastpos();
1968 cur.pos() = cur_orig.pos();
1969 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
1970 DocIterator cur_prev_iter;
1972 found_match = match(cur);
1973 LYXERR(Debug::FIND, "findBackAdv3: found_match="
1974 << found_match << ", cur: " << cur);
1976 return findMostBackwards(cur, match);
1978 // Stop if begin of document reached
1979 if (cur == cur_begin)
1981 cur_prev_iter = cur;
1985 if (cur == cur_begin)
1992 } while (!theApp()->longOperationCancelled());
2000 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
2001 DocIterator const & cur, int len)
2003 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
2005 if (!opt.ignoreformat)
2006 return latexifyFromCursor(cur, len);
2008 return stringifyFromCursor(cur, len);
2012 FindAndReplaceOptions::FindAndReplaceOptions(
2013 docstring const & find_buf_name, bool casesensitive,
2014 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
2015 docstring const & repl_buf_name, bool keep_case,
2016 SearchScope scope, SearchRestriction restr)
2017 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
2018 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
2019 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope), restr(restr)
2027 /** Check if 'len' letters following cursor are all non-lowercase */
2028 static bool allNonLowercase(Cursor const & cur, int len)
2030 pos_type beg_pos = cur.selectionBegin().pos();
2031 pos_type end_pos = cur.selectionBegin().pos() + len;
2032 if (len > cur.lastpos() + 1 - beg_pos) {
2033 LYXERR(Debug::FIND, "This should not happen, more debug needed");
2034 len = cur.lastpos() + 1 - beg_pos;
2035 end_pos = beg_pos + len;
2037 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
2038 if (isLowerCase(cur.paragraph().getChar(pos)))
2044 /** Check if first letter is upper case and second one is lower case */
2045 static bool firstUppercase(Cursor const & cur)
2048 pos_type pos = cur.selectionBegin().pos();
2049 if (pos >= cur.lastpos() - 1) {
2050 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
2053 ch1 = cur.paragraph().getChar(pos);
2054 ch2 = cur.paragraph().getChar(pos + 1);
2055 bool result = isUpperCase(ch1) && isLowerCase(ch2);
2056 LYXERR(Debug::FIND, "firstUppercase(): "
2057 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
2058 << ch2 << "(" << char(ch2) << ")"
2059 << ", result=" << result << ", cur=" << cur);
2064 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
2066 ** \fixme What to do with possible further paragraphs in replace buffer ?
2068 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
2070 ParagraphList::iterator pit = buffer.paragraphs().begin();
2071 LASSERT(pit->size() >= 1, /**/);
2072 pos_type right = pos_type(1);
2073 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
2074 right = pit->size();
2075 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
2081 static void findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
2083 Cursor & cur = bv->cursor();
2084 if (opt.repl_buf_name == docstring()
2085 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
2086 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
2089 DocIterator sel_beg = cur.selectionBegin();
2090 DocIterator sel_end = cur.selectionEnd();
2091 if (&sel_beg.inset() != &sel_end.inset()
2092 || sel_beg.pit() != sel_end.pit()
2093 || sel_beg.idx() != sel_end.idx())
2095 int sel_len = sel_end.pos() - sel_beg.pos();
2096 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
2097 << ", sel_len: " << sel_len << endl);
2100 LASSERT(sel_len > 0, return);
2102 if (!matchAdv(sel_beg, sel_len))
2105 // Build a copy of the replace buffer, adapted to the KeepCase option
2106 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
2108 repl_buffer_orig.write(oss);
2109 string lyx = oss.str();
2110 Buffer repl_buffer("", false);
2111 repl_buffer.setUnnamed(true);
2112 LASSERT(repl_buffer.readString(lyx), return);
2113 if (opt.keep_case && sel_len >= 2) {
2114 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
2115 if (cur.inTexted()) {
2116 if (firstUppercase(cur))
2117 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
2118 else if (allNonLowercase(cur, sel_len))
2119 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
2122 cap::cutSelection(cur, false);
2123 if (cur.inTexted()) {
2124 repl_buffer.changeLanguage(
2125 repl_buffer.language(),
2126 cur.getFont().language());
2127 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
2128 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
2129 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
2130 repl_buffer.params().documentClassPtr(),
2131 bv->buffer().errorList("Paste"));
2132 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
2133 sel_len = repl_buffer.paragraphs().begin()->size();
2134 } else if (cur.inMathed()) {
2135 odocstringstream ods;
2137 OutputParams runparams(&repl_buffer.params().encoding());
2138 runparams.nice = false;
2139 runparams.flavor = OutputParams::LATEX;
2140 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
2141 runparams.dryrun = true;
2142 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
2143 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
2144 docstring repl_latex = ods.str();
2145 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
2147 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
2148 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
2149 repl_latex = from_utf8(s);
2150 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
2151 MathData ar(cur.buffer());
2152 asArray(repl_latex, ar, Parse::NORMAL);
2154 sel_len = ar.size();
2155 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
2157 if (cur.pos() >= sel_len)
2158 cur.pos() -= sel_len;
2161 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
2162 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
2163 bv->processUpdateFlags(Update::Force);
2167 /// Perform a FindAdv operation.
2168 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
2173 // e.g., when invoking word-findadv from mini-buffer wither with
2174 // wrong options syntax or before ever opening advanced F&R pane
2175 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
2179 MatchStringAdv matchAdv(bv->buffer(), opt);
2180 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
2182 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
2183 findAdvReplace(bv, opt, matchAdv);
2186 match_len = findForwardAdv(cur, matchAdv);
2188 match_len = findBackwardsAdv(cur, matchAdv);
2190 // This may only be raised by lyx::regex()
2191 bv->message(_("Invalid regular expression!"));
2195 if (match_len == 0) {
2196 bv->message(_("Match not found!"));
2200 bv->message(_("Match found!"));
2202 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
2203 bv->putSelectionAt(cur, match_len, !opt.forward);
2209 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
2211 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
2212 << opt.casesensitive << ' '
2213 << opt.matchword << ' '
2214 << opt.forward << ' '
2215 << opt.expandmacros << ' '
2216 << opt.ignoreformat << ' '
2217 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
2218 << opt.keep_case << ' '
2219 << int(opt.scope) << ' '
2222 LYXERR(Debug::FIND, "built: " << os.str());
2228 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
2230 LYXERR(Debug::FIND, "parsing");
2234 while (line != "EOSS") {
2238 if (is.eof()) // Tolerate malformed request
2242 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
2243 opt.find_buf_name = from_utf8(s);
2244 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat;
2245 is.get(); // Waste space before replace string
2248 while (line != "EOSS") {
2252 if (is.eof()) // Tolerate malformed request
2256 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
2257 opt.repl_buf_name = from_utf8(s);
2258 is >> opt.keep_case;
2261 opt.scope = FindAndReplaceOptions::SearchScope(i);
2263 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
2265 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
2266 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
2267 << opt.scope << ' ' << opt.restr);