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_(true),
72 ignoreColor_(false), ignoreLanguage_(false) {}
74 bool getFamily() { return ignoreFamily_; };
76 bool getSeries() { return ignoreSeries_; };
78 bool getShape() { return ignoreShape_; };
80 bool getUnderline() { return ignoreUnderline_; };
82 bool getMarkUp() { return ignoreMarkUp_; };
84 bool getStrikeOut() { return ignoreStrikeOut_; };
86 bool getSectioning() { return ignoreSectioning_; };
88 bool getFrontMatter() { return ignoreFrontMatter_; };
90 bool getColor() { return ignoreColor_; };
92 bool getLanguage() { return ignoreLanguage_; };
94 void setIgnoreFormat(string type, bool value);
104 bool ignoreUnderline_;
108 bool ignoreStrikeOut_;
110 bool ignoreSectioning_;
112 bool ignoreFrontMatter_;
116 bool ignoreLanguage_;
120 void IgnoreFormats::setIgnoreFormat(string type, bool value)
122 if (type == "color") {
123 ignoreColor_ = value;
125 else if (type == "language") {
126 ignoreLanguage_ = value;
128 else if (type == "sectioning") {
129 ignoreSectioning_ = value;
130 ignoreFrontMatter_ = value;
132 else if (type == "font") {
133 ignoreSeries_ = value;
134 ignoreShape_ = value;
135 ignoreFamily_ = value;
137 else if (type == "series") {
138 ignoreSeries_ = value;
140 else if (type == "shape") {
141 ignoreShape_ = value;
143 else if (type == "family") {
144 ignoreFamily_ = value;
146 else if (type == "markup") {
147 ignoreMarkUp_ = value;
149 else if (type == "underline") {
150 ignoreUnderline_ = value;
152 else if (type == "strike") {
153 ignoreStrikeOut_ = value;
157 // The global variable that can be changed from outside
158 IgnoreFormats ignoreFormats;
161 void setIgnoreFormat(string type, bool value)
163 ignoreFormats.setIgnoreFormat(type, value);
169 bool parse_bool(docstring & howto)
174 howto = split(howto, var, ' ');
179 class MatchString : public binary_function<Paragraph, pos_type, int>
182 MatchString(docstring const & str, bool cs, bool mw)
183 : str(str), case_sens(cs), whole_words(mw)
186 // returns true if the specified string is at the specified position
187 // del specifies whether deleted strings in ct mode will be considered
188 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
190 return par.find(str, case_sens, whole_words, pos, del);
198 // match whole words only
203 int findForward(DocIterator & cur, MatchString const & match,
204 bool find_del = true)
206 for (; cur; cur.forwardChar())
207 if (cur.inTexted()) {
208 int len = match(cur.paragraph(), cur.pos(), find_del);
216 int findBackwards(DocIterator & cur, MatchString const & match,
217 bool find_del = true)
221 if (cur.inTexted()) {
222 int len = match(cur.paragraph(), cur.pos(), find_del);
231 bool searchAllowed(docstring const & str)
234 frontend::Alert::error(_("Search error"), _("Search string is empty"));
241 bool findOne(BufferView * bv, docstring const & searchstr,
242 bool case_sens, bool whole, bool forward,
243 bool find_del = true, bool check_wrap = false)
245 if (!searchAllowed(searchstr))
248 DocIterator cur = forward
249 ? bv->cursor().selectionEnd()
250 : bv->cursor().selectionBegin();
252 MatchString const match(searchstr, case_sens, whole);
254 int match_len = forward
255 ? findForward(cur, match, find_del)
256 : findBackwards(cur, match, find_del);
259 bv->putSelectionAt(cur, match_len, !forward);
260 else if (check_wrap) {
261 DocIterator cur_orig(bv->cursor());
264 q = _("End of file reached while searching forward.\n"
265 "Continue searching from the beginning?");
267 q = _("Beginning of file reached while searching backward.\n"
268 "Continue searching from the end?");
269 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
270 q, 0, 1, _("&Yes"), _("&No"));
271 if (wrap_answer == 0) {
273 bv->cursor().clear();
274 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
276 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
277 bv->cursor().backwardPos();
279 bv->clearSelection();
280 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
283 bv->cursor().setCursor(cur_orig);
287 return match_len > 0;
291 int replaceAll(BufferView * bv,
292 docstring const & searchstr, docstring const & replacestr,
293 bool case_sens, bool whole)
295 Buffer & buf = bv->buffer();
297 if (!searchAllowed(searchstr) || buf.isReadonly())
300 DocIterator cur_orig(bv->cursor());
302 MatchString const match(searchstr, case_sens, whole);
305 int const rsize = replacestr.size();
306 int const ssize = searchstr.size();
309 cur.setCursor(doc_iterator_begin(&buf));
310 int match_len = findForward(cur, match, false);
311 while (match_len > 0) {
312 // Backup current cursor position and font.
313 pos_type const pos = cur.pos();
314 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
316 int striked = ssize -
317 cur.paragraph().eraseChars(pos, pos + match_len,
318 buf.params().track_changes);
319 cur.paragraph().insert(pos, replacestr, font,
320 Change(buf.params().track_changes
322 : Change::UNCHANGED));
323 for (int i = 0; i < rsize + striked; ++i)
326 match_len = findForward(cur, match, false);
329 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
331 cur_orig.fixIfBroken();
332 bv->setCursor(cur_orig);
338 // the idea here is that we are going to replace the string that
339 // is selected IF it is the search string.
340 // if there is a selection, but it is not the search string, then
341 // we basically ignore it. (FIXME We ought to replace only within
343 // if there is no selection, then:
344 // (i) if some search string has been provided, then we find it.
345 // (think of how the dialog works when you hit "replace" the
347 // (ii) if no search string has been provided, then we treat the
348 // word the cursor is in as the search string. (why? i have no
349 // idea.) but this only works in text?
351 // returns the number of replacements made (one, if any) and
352 // whether anything at all was done.
353 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
354 docstring const & replacestr, bool case_sens,
355 bool whole, bool forward, bool findnext)
357 Cursor & cur = bv->cursor();
359 if (!cur.selection()) {
360 // no selection, non-empty search string: find it
361 if (!searchstr.empty()) {
362 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
363 return make_pair(found, 0);
365 // empty search string
368 return make_pair(false, 0);
369 // select current word and treat it as the search string.
370 // This causes a minor bug as undo will restore this selection,
371 // which the user did not create (#8986).
372 cur.innerText()->selectWord(cur, WHOLE_WORD);
373 searchstr = cur.selectionAsString(false);
376 // if we still don't have a search string, report the error
378 if (!searchAllowed(searchstr))
379 return make_pair(false, 0);
381 bool have_selection = cur.selection();
382 docstring const selected = cur.selectionAsString(false);
385 ? searchstr == selected
386 : compare_no_case(searchstr, selected) == 0;
388 // no selection or current selection is not search word:
389 // just find the search word
390 if (!have_selection || !match) {
391 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
392 return make_pair(found, 0);
395 // we're now actually ready to replace. if the buffer is
396 // read-only, we can't, though.
397 if (bv->buffer().isReadonly())
398 return make_pair(false, 0);
400 cap::replaceSelectionWithString(cur, replacestr);
402 cur.pos() += replacestr.length();
403 LASSERT(cur.pos() <= cur.lastpos(),
404 cur.pos() = cur.lastpos());
407 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
409 return make_pair(true, 1);
415 docstring const find2string(docstring const & search,
416 bool casesensitive, bool matchword, bool forward)
420 << int(casesensitive) << ' '
421 << int(matchword) << ' '
427 docstring const replace2string(docstring const & replace,
428 docstring const & search,
429 bool casesensitive, bool matchword,
430 bool all, bool forward, bool findnext)
433 ss << replace << '\n'
435 << int(casesensitive) << ' '
436 << int(matchword) << ' '
438 << int(forward) << ' '
444 bool lyxfind(BufferView * bv, FuncRequest const & ev)
446 if (!bv || ev.action() != LFUN_WORD_FIND)
449 //lyxerr << "find called, cmd: " << ev << endl;
451 // data is of the form
453 // <casesensitive> <matchword> <forward>"
455 docstring howto = split(ev.argument(), search, '\n');
457 bool casesensitive = parse_bool(howto);
458 bool matchword = parse_bool(howto);
459 bool forward = parse_bool(howto);
461 return findOne(bv, search, casesensitive, matchword, forward, true, true);
465 bool lyxreplace(BufferView * bv,
466 FuncRequest const & ev, bool has_deleted)
468 if (!bv || ev.action() != LFUN_WORD_REPLACE)
471 // data is of the form
474 // <casesensitive> <matchword> <all> <forward> <findnext>"
477 docstring howto = split(ev.argument(), rplc, '\n');
478 howto = split(howto, search, '\n');
480 bool casesensitive = parse_bool(howto);
481 bool matchword = parse_bool(howto);
482 bool all = parse_bool(howto);
483 bool forward = parse_bool(howto);
484 bool findnext = howto.empty() ? true : parse_bool(howto);
489 int replace_count = 0;
491 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
492 update = replace_count > 0;
495 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
497 replace_count = rv.second;
500 Buffer const & buf = bv->buffer();
502 // emit message signal.
503 buf.message(_("String not found."));
505 if (replace_count == 0) {
506 buf.message(_("String found."));
507 } else if (replace_count == 1) {
508 buf.message(_("String has been replaced."));
510 docstring const str =
511 bformat(_("%1$d strings have been replaced."), replace_count);
515 } else if (findnext) {
516 // if we have deleted characters, we do not replace at all, but
517 // rather search for the next occurence
518 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
521 bv->message(_("String not found."));
527 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
529 for (; cur; cur.forwardPos())
530 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
534 DocIterator cur_orig(bv->cursor());
535 docstring q = _("End of file reached while searching forward.\n"
536 "Continue searching from the beginning?");
537 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
538 q, 0, 1, _("&Yes"), _("&No"));
539 if (wrap_answer == 0) {
540 bv->cursor().clear();
541 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
542 bv->clearSelection();
543 cur.setCursor(bv->cursor().selectionBegin());
544 if (findNextChange(bv, cur, false))
547 bv->cursor().setCursor(cur_orig);
554 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
556 for (cur.backwardPos(); cur; cur.backwardPos()) {
557 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
562 DocIterator cur_orig(bv->cursor());
563 docstring q = _("Beginning of file reached while searching backward.\n"
564 "Continue searching from the end?");
565 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
566 q, 0, 1, _("&Yes"), _("&No"));
567 if (wrap_answer == 0) {
568 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
569 bv->cursor().backwardPos();
570 bv->clearSelection();
571 cur.setCursor(bv->cursor().selectionBegin());
572 if (findPreviousChange(bv, cur, false))
575 bv->cursor().setCursor(cur_orig);
582 bool selectChange(Cursor & cur, bool forward)
584 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
586 Change ch = cur.paragraph().lookupChange(cur.pos());
588 CursorSlice tip1 = cur.top();
589 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
590 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
591 if (!ch2.isSimilarTo(ch))
594 CursorSlice tip2 = cur.top();
595 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
597 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
598 if (!ch2.isSimilarTo(ch)) {
599 // take a step forward to correctly set the selection
607 cur.bv().mouseSetCursor(cur, false);
609 cur.bv().mouseSetCursor(cur, true);
617 bool findChange(BufferView * bv, bool forward)
620 cur.setCursor(forward ? bv->cursor().selectionEnd()
621 : bv->cursor().selectionBegin());
622 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
623 return selectChange(cur, forward);
628 bool findNextChange(BufferView * bv)
630 return findChange(bv, true);
634 bool findPreviousChange(BufferView * bv)
636 return findChange(bv, false);
643 typedef vector<pair<string, string> > Escapes;
645 /// A map of symbols and their escaped equivalent needed within a regex.
646 /// @note Beware of order
647 Escapes const & get_regexp_escapes()
649 typedef std::pair<std::string, std::string> P;
651 static Escapes escape_map;
652 if (escape_map.empty()) {
653 escape_map.push_back(P("$", "_x_$"));
654 escape_map.push_back(P("{", "_x_{"));
655 escape_map.push_back(P("}", "_x_}"));
656 escape_map.push_back(P("[", "_x_["));
657 escape_map.push_back(P("]", "_x_]"));
658 escape_map.push_back(P("(", "_x_("));
659 escape_map.push_back(P(")", "_x_)"));
660 escape_map.push_back(P("+", "_x_+"));
661 escape_map.push_back(P("*", "_x_*"));
662 escape_map.push_back(P(".", "_x_."));
663 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
664 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
665 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
666 escape_map.push_back(P("_x_", "\\"));
671 /// A map of lyx escaped strings and their unescaped equivalent.
672 Escapes const & get_lyx_unescapes()
674 typedef std::pair<std::string, std::string> P;
676 static Escapes escape_map;
677 if (escape_map.empty()) {
678 escape_map.push_back(P("\\%", "%"));
679 escape_map.push_back(P("\\mathcircumflex ", "^"));
680 escape_map.push_back(P("\\mathcircumflex", "^"));
681 escape_map.push_back(P("\\backslash ", "\\"));
682 escape_map.push_back(P("\\backslash", "\\"));
683 escape_map.push_back(P("\\\\{", "_x_<"));
684 escape_map.push_back(P("\\\\}", "_x_>"));
685 escape_map.push_back(P("\\sim ", "~"));
686 escape_map.push_back(P("\\sim", "~"));
691 /// A map of escapes turning a regexp matching text to one matching latex.
692 Escapes const & get_regexp_latex_escapes()
694 typedef std::pair<std::string, std::string> P;
696 static Escapes escape_map;
697 if (escape_map.empty()) {
698 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
699 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
700 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
701 escape_map.push_back(P("\\[", "\\{\\[\\}"));
702 escape_map.push_back(P("\\]", "\\{\\]\\}"));
703 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
704 escape_map.push_back(P("%", "\\\\\\%"));
705 escape_map.push_back(P("#", "\\\\#"));
710 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
711 ** the found occurrence were escaped.
713 string apply_escapes(string s, Escapes const & escape_map)
715 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
716 Escapes::const_iterator it;
717 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
718 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
719 unsigned int pos = 0;
720 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
721 s.replace(pos, it->first.length(), it->second);
722 LYXERR(Debug::FIND, "After escape: " << s);
723 pos += it->second.length();
724 // LYXERR(Debug::FIND, "pos: " << pos);
727 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
732 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
733 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
734 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
735 string escape_for_regex(string s, bool match_latex)
738 while (pos < s.size()) {
739 size_t new_pos = s.find("\\regexp{", pos);
740 if (new_pos == string::npos)
742 LYXERR(Debug::FIND, "new_pos: " << new_pos);
743 string t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
744 LYXERR(Debug::FIND, "t [lyx]: " << t);
745 t = apply_escapes(t, get_regexp_escapes());
746 LYXERR(Debug::FIND, "t [rxp]: " << t);
747 s.replace(pos, new_pos - pos, t);
748 new_pos = pos + t.size();
749 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
750 LYXERR(Debug::FIND, "new_pos: " << new_pos);
751 if (new_pos == s.size())
753 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
754 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
755 LYXERR(Debug::FIND, "end_pos: " << end_pos);
756 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
757 LYXERR(Debug::FIND, "t in regexp : " << t);
758 t = apply_escapes(t, get_lyx_unescapes());
759 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
761 t = apply_escapes(t, get_regexp_latex_escapes());
762 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
764 if (end_pos == s.size()) {
765 s.replace(new_pos, end_pos - new_pos, t);
766 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
769 s.replace(new_pos, end_pos + 13 - new_pos, t);
770 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
771 pos = new_pos + t.size();
772 LYXERR(Debug::FIND, "pos: " << pos);
778 /// Wrapper for lyx::regex_replace with simpler interface
779 bool regex_replace(string const & s, string & t, string const & searchstr,
780 string const & replacestr)
782 lyx::regex e(searchstr, regex_constants::ECMAScript);
784 ostream_iterator<char, char> it(oss);
785 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
786 // tolerate t and s be references to the same variable
787 bool rv = (s != oss.str());
793 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
795 ** Verify that closed braces exactly match open braces. This avoids that, for example,
796 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
799 ** Number of open braces that must remain open at the end for the verification to succeed.
801 bool braces_match(string::const_iterator const & beg,
802 string::const_iterator const & end,
806 string::const_iterator it = beg;
807 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
808 for (; it != end; ++it) {
809 // Skip escaped braces in the count
814 } else if (*it == '{') {
816 } else if (*it == '}') {
817 if (open_pars == 0) {
818 LYXERR(Debug::FIND, "Found unmatched closed brace");
824 if (open_pars != unmatched) {
825 LYXERR(Debug::FIND, "Found " << open_pars
826 << " instead of " << unmatched
827 << " unmatched open braces at the end of count");
830 LYXERR(Debug::FIND, "Braces match as expected");
835 /** The class performing a match between a position in the document and the FindAdvOptions.
837 class MatchStringAdv {
839 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
841 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
842 ** constructor as opt.search, under the opt.* options settings.
845 ** If set, then match is searched only against beginning of text starting at cur.
846 ** If unset, then match is searched anywhere in text starting at cur.
849 ** The length of the matching text, or zero if no match was found.
851 int operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
856 /// first buffer on which search was started
857 lyx::Buffer * const p_first_buf;
859 FindAndReplaceOptions const & opt;
862 /// Auxiliary find method (does not account for opt.matchword)
863 int findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
865 /** Normalize a stringified or latexified LyX paragraph.
869 ** <li>if search is not casesensitive, then lowercase the string;
870 ** <li>remove any newline at begin or end of the string;
871 ** <li>replace any newline in the middle of the string with a simple space;
872 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
875 ** @todo Normalization should also expand macros, if the corresponding
876 ** search option was checked.
878 string normalize(docstring const & s, bool hack_braces) const;
879 // normalized string to search
880 string par_as_string;
881 // regular expression to use for searching
883 // same as regexp, but prefixed with a ".*"
885 // leading format material as string
886 string lead_as_string;
887 // par_as_string after removal of lead_as_string
888 string par_as_string_nolead;
889 // unmatched open braces in the search string/regexp
891 // number of (.*?) subexpressions added at end of search regexp for closing
892 // environments, math mode, styles, etc...
894 // Are we searching with regular expressions ?
899 static docstring buffer_to_latex(Buffer & buffer)
901 OutputParams runparams(&buffer.params().encoding());
902 odocstringstream ods;
904 runparams.nice = true;
905 runparams.flavor = OutputParams::LATEX;
906 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
907 // No side effect of file copying and image conversion
908 runparams.dryrun = true;
909 runparams.for_search = true;
910 pit_type const endpit = buffer.paragraphs().size();
911 for (pit_type pit = 0; pit != endpit; ++pit) {
912 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
913 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
919 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
922 if (!opt.ignoreformat) {
923 str = buffer_to_latex(buffer);
925 OutputParams runparams(&buffer.params().encoding());
926 runparams.nice = true;
927 runparams.flavor = OutputParams::LATEX;
928 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
929 runparams.dryrun = true;
930 runparams.for_search = true;
931 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
932 Paragraph const & par = buffer.paragraphs().at(pit);
933 LYXERR(Debug::FIND, "Adding to search string: '"
934 << par.asString(pos_type(0), par.size(),
935 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
938 str += par.asString(pos_type(0), par.size(),
939 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
947 /// Return separation pos between the leading material and the rest
948 static size_t identifyLeading(string const & s)
951 // @TODO Support \item[text]
952 // Kornel: Added textsl, textsf, textit, texttt and noun
953 // + allow to search for colored text too
954 while (regex_replace(t, t, REGEX_BOS "\\\\(((emph|noun|minisec|text(bf|md|sl|sf|it|tt))|((textcolor|foreignlanguage)\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
955 || regex_replace(t, t, REGEX_BOS "\\$", "")
956 || regex_replace(t, t, REGEX_BOS "\\\\\\[ ", "")
957 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
958 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
960 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
965 * Given a latexified string, retrieve some handled features
966 * The features of the regex will later be compared with the features
967 * of the searched text. If the regex features are not a
968 * subset of the analized, then, in not format ignoring search
969 * we can early stop the search in the relevant inset.
971 typedef map<string, bool> Features;
973 static Features identifyFeatures(string const & s)
975 static regex const feature("\\\\(([a-z]+(\\{([a-z]+)\\}|\\*)?))\\{");
976 static regex const valid("^(((emph|noun|text(bf|md|sl|sf|it|tt)|(textcolor|foreignlanguage|item)\\{[a-z]+\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)$");
981 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
984 if (sub.str(1).compare("regexp") == 0) {
988 string token = sub.str(1);
990 if (regex_match(token, sub2, valid)) {
998 if (sub.str(1).compare("endregexp") == 0) {
1008 * defines values features of a key "\\[a-z]+{"
1015 isMain, /* for \\foreignlanguage */
1023 isIgnored /* to be ignored by creating infos */
1028 parenthesiscount(1),
1032 KeyInfo(KeyType type, int parcount, bool disable)
1034 parenthesiscount(parcount),
1043 int parenthesiscount;
1045 bool used; /* by pattern */
1050 Border(int l=0, int u=0) : low(l), upper(u) {};
1055 #define MAXOPENED 30
1057 bool isPatternString;
1059 Intervall(bool isPattern) : isPatternString(isPattern), ignoreidx(-1), actualdeptindex(0) { depts[0] = 0;};
1062 int depts[MAXOPENED];
1063 int closes[MAXOPENED];
1064 int actualdeptindex;
1065 Border borders[2*MAXOPENED];
1066 // int previousNotIgnored(int);
1067 int nextNotIgnored(int);
1068 void handleOpenP(int i);
1069 void handleCloseP(int i, bool closingAllowed);
1070 void resetOpenedP(int openPos);
1071 void addIntervall(int upper);
1072 void addIntervall(int low, int upper); /* if explicit */
1073 void setForDefaultLang(int upTo);
1074 int findclosing(int start, int end, char up, char down);
1075 void handleParentheses(int lastpos, bool closingAllowed);
1076 void output(ostringstream &os, int lastpos);
1077 // string show(int lastpos);
1080 void Intervall::setForDefaultLang(int upTo)
1082 // Enable the use of first token again
1083 if (ignoreidx >= 0) {
1084 if (borders[0].low < upTo)
1085 borders[0].low = upTo;
1086 if (borders[0].upper < upTo)
1087 borders[0].upper = upTo;
1091 static void checkDepthIndex(int val)
1093 static int maxdepthidx = MAXOPENED-2;
1094 if (val > maxdepthidx) {
1096 LYXERR0("maxdepthidx now " << val);
1100 static void checkIgnoreIdx(int val)
1102 static int maxignoreidx = 2*MAXOPENED - 4;
1103 if (val > maxignoreidx) {
1105 LYXERR0("maxignoreidx now " << val);
1110 * Expand the region of ignored parts of the input latex string
1111 * The region is only relevant in output()
1113 void Intervall::addIntervall(int low, int upper)
1116 if (low == upper) return;
1117 for (idx = ignoreidx+1; idx > 0; --idx) {
1118 if (low > borders[idx-1].upper) {
1122 Border br(low, upper);
1123 if (idx > ignoreidx) {
1126 checkIgnoreIdx(ignoreidx);
1130 // Expand only if one of the new bound is inside the interwall
1131 // We know here that br.low > borders[idx-1].upper
1132 if (br.upper < borders[idx].low) {
1133 // We have to insert at this pos
1134 for (int i = ignoreidx+1; i > idx; --i) {
1135 borders[i] = borders[i-1];
1139 checkIgnoreIdx(ignoreidx);
1142 // Here we know, that we are overlapping
1143 if (br.low > borders[idx].low)
1144 br.low = borders[idx].low;
1145 // check what has to be concatenated
1147 for (int i = idx; i <= ignoreidx; i++) {
1148 if (br.upper >= borders[i].low) {
1150 if (br.upper < borders[i].upper)
1151 br.upper = borders[i].upper;
1157 // count should be >= 1 here
1160 for (int i = idx + count; i <= ignoreidx; i++) {
1161 borders[i-count+1] = borders[i];
1163 ignoreidx -= count - 1;
1169 void Intervall::handleOpenP(int i)
1172 depts[actualdeptindex] = i+1;
1173 closes[actualdeptindex] = -1;
1174 checkDepthIndex(actualdeptindex);
1177 void Intervall::handleCloseP(int i, bool closingAllowed)
1179 if (actualdeptindex <= 0) {
1180 if (! closingAllowed)
1181 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1182 // if we are at the very end
1183 addIntervall(i, i+1);
1186 closes[actualdeptindex] = i+1;
1191 void Intervall::resetOpenedP(int openPos)
1193 // Used as initializer for foreignlanguage entry
1194 actualdeptindex = 1;
1195 depts[1] = openPos+1;
1200 int Intervall::previousNotIgnored(int start)
1202 int idx = 0; /* int intervalls */
1203 for (idx = ignoreidx; idx >= 0; --idx) {
1204 if (start > borders[idx].upper)
1206 if (start >= borders[idx].low)
1207 start = borders[idx].low-1;
1213 int Intervall::nextNotIgnored(int start)
1215 int idx = 0; /* int intervalls */
1216 for (idx = 0; idx <= ignoreidx; idx++) {
1217 if (start < borders[idx].low)
1219 if (start < borders[idx].upper)
1220 start = borders[idx].upper;
1225 typedef map<string, KeyInfo> KeysMap;
1226 typedef vector< KeyInfo> Entries;
1227 static KeysMap keys = map<string, KeyInfo>();
1234 void buildKeys(bool);
1235 void buildEntries(bool);
1236 void makeKey(const string &, KeyInfo, bool isPatternString);
1237 void processRegion(int start, int region_end); /* remove {} parts */
1238 void removeHead(KeyInfo&, int count=0);
1241 LatexInfo(string par, bool isPatternString) : interval(isPatternString) {
1243 buildKeys(isPatternString);
1244 entries = vector<KeyInfo>();
1245 buildEntries(isPatternString);
1249 if (entries.empty()) {
1256 if (int(entries.size()) > entidx) {
1263 bool setNextKey(int idx) {
1264 if ((idx == entidx) && (entidx >= 0)) {
1271 int process(ostringstream &os, KeyInfo &actual);
1272 int dispatch(ostringstream &os, int previousStart, KeyInfo &actual);
1273 // string show(int lastpos) { return interval.show(lastpos);};
1274 int nextNotIgnored(int start) { return interval.nextNotIgnored(start);};
1275 KeyInfo &getKeyInfo(int keyinfo) {
1276 static KeyInfo invalidInfo = KeyInfo();
1277 if ((keyinfo < 0) || ( keyinfo >= int(entries.size())))
1280 return entries[keyinfo];
1282 void setForDefaultLang(int upTo) {interval.setForDefaultLang(upTo);};
1283 void addIntervall(int low, int up) { interval.addIntervall(low, up); };
1287 int Intervall::findclosing(int start, int end, char up = '{', char down = '}')
1291 for (int i = start; i < end; i += 1 + skip) {
1295 if (c == '\\') skip = 1;
1299 else if (c == down) {
1300 if (depth == 0) return i;
1316 vector<MathEntry> entries;
1321 void insert(string wait, size_t start, size_t end) {
1322 MathEntry m = MathEntry();
1324 m.mathStart = start;
1326 m.mathSize = end - start;
1327 entries.push_back(m);
1329 bool evaluating(size_t pos) {
1330 while (actualIdx < entries.size()) {
1331 if (pos < entries[actualIdx].mathStart)
1333 if (pos < entries[actualIdx].mathEnd)
1339 bool empty() { return entries.empty(); };
1340 size_t getEndPos() {
1341 if (entries.empty() || (actualIdx >= entries.size())) {
1344 return entries[actualIdx].mathEnd;
1346 size_t getStartPos() {
1347 if (entries.empty() || (actualIdx >= entries.size())) {
1348 return 100000; /* definitely enough? */
1350 return entries[actualIdx].mathStart;
1352 size_t getFirstPos() {
1354 return getStartPos();
1357 if (entries.empty() || (actualIdx >= entries.size())) {
1360 return entries[actualIdx].mathSize;
1362 void incrEntry() { actualIdx++; };
1365 void LatexInfo::buildEntries(bool isPatternString)
1367 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1368 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1369 static bool disableLanguageOverride = false;
1370 smatch sub, submath;
1371 bool evaluatingRegexp = false;
1373 bool evaluatingMath = false;
1375 bool math_end_waiting = false;
1376 size_t math_pos = 10000;
1379 for (sregex_iterator itmath(interval.par.begin(), interval.par.end(), rmath), end; itmath != end; ++itmath) {
1381 if (math_end_waiting) {
1382 size_t pos = submath.position(size_t(0));
1383 if ((math_end == "$") &&
1384 (submath.str(0) == "$") &&
1385 (interval.par[pos-1] != '\\')) {
1386 mi.insert("$", math_pos, pos + 1);
1387 math_end_waiting = false;
1389 else if ((math_end == "\\]") &&
1390 (submath.str(0) == "\\]")) {
1391 mi.insert("\\]", math_pos, pos + 2);
1392 math_end_waiting = false;
1394 else if ((submath.str(1).compare("end") == 0) &&
1395 (submath.str(2).compare(math_end) == 0)) {
1396 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1397 math_end_waiting = false;
1403 if (submath.str(1).compare("begin") == 0) {
1404 math_end_waiting = true;
1405 math_end = submath.str(2);
1406 math_pos = submath.position(size_t(0));
1408 else if (submath.str(0).compare("\\[") == 0) {
1409 math_end_waiting = true;
1411 math_pos = submath.position(size_t(0));
1413 else if (submath.str(0) == "$") {
1414 size_t pos = submath.position(size_t(0));
1415 if ((pos == 0) || (interval.par[pos-1] != '\\')) {
1416 math_end_waiting = true;
1423 // Ignore language if there is math somewhere in pattern-string
1424 if (isPatternString) {
1427 keys["foreignlanguage"].disabled = true;
1428 disableLanguageOverride = true;
1431 disableLanguageOverride = false;
1434 if (disableLanguageOverride) {
1435 keys["foreignlanguage"].disabled = true;
1438 math_pos = mi.getFirstPos();
1439 for (sregex_iterator it(interval.par.begin(), interval.par.end(), rkeys), end; it != end; ++it) {
1441 string key = sub.str(3);
1443 if (sub.str(0)[0] == '\\')
1444 key = sub.str(0)[1];
1448 if (evaluatingRegexp) {
1449 if (sub.str(1).compare("endregexp") == 0) {
1450 evaluatingRegexp = false;
1451 // found._tokenstart already set
1452 found._dataEnd = sub.position(size_t(0)) + 13;
1453 found._dataStart = found._dataEnd;
1454 found._tokensize = found._dataEnd - found._tokenstart;
1455 found.parenthesiscount = 0;
1459 if (evaluatingMath) {
1460 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1462 evaluatingMath = false;
1464 math_pos = mi.getStartPos();
1466 if (keys.find(key) == keys.end()) {
1467 LYXERR(Debug::FIND, "Found unknown key " << sub.str(0));
1471 if (key.compare("regexp") == 0) {
1472 evaluatingRegexp = true;
1473 found._tokenstart = sub.position(size_t(0));
1474 found._tokensize = 0;
1478 // Handle the other params of key
1479 if (found.keytype == KeyInfo::isIgnored)
1481 else if (found.keytype == KeyInfo::isMath) {
1482 if (size_t(sub.position(size_t(0))) == math_pos) {
1484 found._tokenstart = sub.position(size_t(0));
1485 found._tokensize = mi.getSize();
1486 found._dataEnd = found._tokenstart + found._tokensize;
1487 found._dataStart = found._dataEnd;
1488 found.parenthesiscount = 0;
1489 evaluatingMath = true;
1492 // begin|end of unknown env, discard
1493 // First handle tables
1494 // longtable|tabular
1495 bool discardComment;
1496 if ((sub.str(5).compare("longtable") == 0) ||
1497 (sub.str(5).compare("tabular") == 0)) {
1498 discardComment = true; /* '%' */
1501 discardComment = false;
1503 // discard spaces before pos(0)
1504 int pos = sub.position(size_t(0));
1506 for (count = 0; pos - count > 0; count++) {
1507 char c = interval.par[pos-count-1];
1508 if (discardComment) {
1509 if ((c != ' ') && (c != '%'))
1515 found.keytype = KeyInfo::doRemove;
1516 found._tokenstart = pos - count;
1517 if (sub.str(1).compare(0, 5, "begin") == 0) {
1518 size_t pos1 = pos + sub.str(0).length();
1519 if (interval.par[pos1] == '[') {
1520 pos1 = interval.findclosing(pos1+1, interval.par.length(), '[', ']')+1;
1522 if (interval.par[pos1] == '{') {
1523 found._dataEnd = interval.findclosing(pos1+1, interval.par.length()) + 1;
1526 found._dataEnd = pos1;
1528 found._dataStart = found._dataEnd;
1531 found._dataStart = pos + sub.str(0).length();
1532 found._dataEnd = found._dataStart;
1534 found._tokensize = count + found._dataEnd - pos;
1535 found.parenthesiscount = 0;
1536 found.disabled = true;
1539 else if (found.keytype != KeyInfo::isRegex) {
1540 found._tokenstart = sub.position(size_t(0));
1541 if (found.parenthesiscount == 0) {
1542 // Probably to be discarded
1543 char following = interval.par[sub.position(size_t(0)) + sub.str(3).length() + 1];
1544 if (following == ' ')
1545 found.head = "\\" + sub.str(3) + " ";
1546 else if (following == '=') {
1547 // like \uldepth=1000pt
1548 found.head = sub.str(0);
1551 found.head = "\\" + key;
1552 found._tokensize = found.head.length();
1553 found._dataEnd = found._tokenstart + found._tokensize;
1554 found._dataStart = found._dataEnd;
1557 if (found.parenthesiscount == 1) {
1558 found.head = "\\" + key + "{";
1560 else if (found.parenthesiscount == 2) {
1561 found.head = sub.str(0) + "{";
1563 found._tokensize = found.head.length();
1564 found._dataStart = found._tokenstart + found.head.length();
1565 size_t endpos = interval.findclosing(found._dataStart, interval.par.length());
1566 if ((endpos == interval.par.length()) &&
1567 (found.keytype == KeyInfo::doRemove)) {
1568 // Missing closing => error in latex-input?
1569 // therefore do not delete remaining data
1570 found._dataStart -= 1;
1571 found._dataEnd = found._dataStart;
1574 found._dataEnd = endpos;
1575 if (isPatternString) {
1576 keys[key].used = true;
1580 entries.push_back(found);
1584 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1586 stringstream s(keysstring);
1588 const char delim = '|';
1589 while (getline(s, key, delim)) {
1590 KeyInfo keyII(keyI);
1591 if (isPatternString) {
1594 else if ( !keys[key].used)
1595 keyII.disabled = true;
1600 void LatexInfo::buildKeys(bool isPatternString)
1603 static bool keysBuilt = false;
1604 if (keysBuilt && !isPatternString) return;
1606 // Know standard keys with 1 parameter.
1607 // Split is done, if not at start of region
1608 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1609 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1610 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1611 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1612 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1613 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
1615 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
1616 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1617 makeKey("section*|subsection*|subsubsection*|paragraph*",
1618 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1619 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1620 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getFrontMatter()), isPatternString);
1622 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
1624 // Split is done, if not at start of region
1625 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
1627 // Split is done always.
1628 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
1632 makeKey("backslash|textbackslash|textasciicircum|textasciitilde|ldots", KeyInfo(KeyInfo::isChar, 1, false), isPatternString);
1633 // Found in fr/UserGuide.lyx
1634 makeKey("og|fg|textvisiblespace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1636 // Known macros to remove (including their parameter)
1638 makeKey("inputencoding|shortcut|label|ref", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
1640 // Macros to remove, but let the parameter survive
1642 makeKey("url|href|menuitem|footnote|code|index|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1644 // Same effect as previous, parameter will survive (because there is no one anyway)
1646 makeKey("noindent", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1647 // Remove table decorations
1648 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
1649 // Discard shape-header
1650 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1651 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1652 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1653 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1654 // like ('tiny{}' or '\tiny ' ... )
1655 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, true), isPatternString);
1657 // Survives, like known character
1658 makeKey("lyx|latex|tex", KeyInfo(KeyInfo::isIgnored, 0, false), isPatternString);
1659 makeKey("item", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
1661 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1662 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1663 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1665 makeKey("par|uldepth|ULdepth|protect", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
1667 if (isPatternString) {
1668 // Allow the first searched string to rebuild the keys too
1672 // no need to rebuild again
1678 * Keep the list of actual opened parentheses actual
1679 * (e.g. depth == 4 means there are 4 '{' not processed yet)
1681 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
1684 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
1688 if (c == '\\') skip = 1;
1689 else if (c == '{') {
1692 else if (c == '}') {
1693 handleCloseP(i, closingAllowed);
1699 string Intervall::show(int lastpos)
1701 int idx = 0; /* int intervalls */
1705 for (idx = 0; idx <= ignoreidx; idx++) {
1706 while (i < lastpos) {
1708 if (i <= borders[idx].low) {
1709 if (borders[idx].low > lastpos)
1710 printsize = lastpos - i;
1712 printsize = borders[idx].low - i;
1713 s += par.substr(i, printsize);
1715 if (i >= borders[idx].low)
1716 i = borders[idx].upper;
1719 i = borders[idx].upper;
1725 s += par.substr(i, lastpos-i);
1731 void Intervall::output(ostringstream &os, int lastpos)
1733 // get number of chars to output
1734 int idx = 0; /* int intervalls */
1736 for (idx = 0; idx <= ignoreidx; idx++) {
1739 if (i <= borders[idx].low) {
1740 if (borders[idx].low > lastpos)
1741 printsize = lastpos - i;
1743 printsize = borders[idx].low - i;
1744 os << par.substr(i, printsize);
1746 handleParentheses(i, false);
1747 if (i >= borders[idx].low)
1748 i = borders[idx].upper;
1751 i = borders[idx].upper;
1758 os << par.substr(i, lastpos-i);
1760 handleParentheses(lastpos, false);
1761 for (int i = actualdeptindex; i > 0; --i) {
1764 if (! isPatternString)
1766 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
1769 void LatexInfo::processRegion(int start, int region_end)
1771 while (start < region_end) { /* Let {[} and {]} survive */
1772 if ((interval.par[start] == '{') &&
1773 (interval.par[start+1] != ']') &&
1774 (interval.par[start+1] != '[')) {
1775 // Closing is allowed past the region
1776 int closing = interval.findclosing(start+1, interval.par.length());
1777 interval.addIntervall(start, start+1);
1778 interval.addIntervall(closing, closing+1);
1780 start = interval.nextNotIgnored(start+1);
1784 void LatexInfo::removeHead(KeyInfo &actual, int count)
1786 if (actual.parenthesiscount == 0) {
1787 // "{\tiny{} ...}" ==> "{{} ...}"
1788 interval.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
1791 // Remove header hull, that is "\url{abcd}" ==> "abcd"
1792 interval.addIntervall(actual._tokenstart, actual._dataStart);
1793 interval.addIntervall(actual._dataEnd, actual._dataEnd+1);
1797 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
1800 switch (actual.keytype)
1802 case KeyInfo::isChar: {
1803 nextKeyIdx = getNextKey();
1806 case KeyInfo::isSize: {
1807 if (actual.disabled) {
1809 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
1810 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
1811 nextKeyIdx = getNextKey();
1813 // Split on this key if not at start
1814 int start = interval.nextNotIgnored(previousStart);
1815 if (start < actual._tokenstart) {
1816 interval.output(os, actual._tokenstart);
1817 interval.addIntervall(start, actual._tokenstart);
1819 // discard entry if at end of actual
1820 nextKeyIdx = process(os, actual);
1824 case KeyInfo::isStandard: {
1825 if (actual.disabled) {
1827 processRegion(actual._dataStart, actual._dataStart+1);
1828 nextKeyIdx = getNextKey();
1830 // Split on this key if not at start
1831 int start = interval.nextNotIgnored(previousStart);
1832 if (start < actual._tokenstart) {
1833 interval.output(os, actual._tokenstart);
1834 interval.addIntervall(start, actual._tokenstart);
1836 // discard entry if at end of actual
1837 nextKeyIdx = process(os, actual);
1841 case KeyInfo::doRemove: {
1842 // Remove the key with all parameters and following spaces
1844 for (pos = actual._dataEnd+1; pos < interval.par.length(); pos++) {
1845 if (interval.par[pos] != ' ')
1848 interval.addIntervall(actual._tokenstart, pos);
1849 nextKeyIdx = getNextKey();
1852 case KeyInfo::isList: {
1853 // Discard space before _tokenstart
1855 for (count = 0; count < actual._tokenstart; count++) {
1856 if (interval.par[actual._tokenstart-count-1] != ' ')
1859 if (actual.disabled) {
1860 interval.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
1863 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
1865 // Discard extra parentheses '[]'
1866 if (interval.par[actual._dataEnd+1] == '[') {
1867 int posdown = interval.findclosing(actual._dataEnd+2, interval.par.length(), '[', ']');
1868 if ((interval.par[actual._dataEnd+2] == '{') &&
1869 (interval.par[posdown-1] == '}')) {
1870 interval.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
1871 interval.addIntervall(posdown-1, posdown+1);
1874 interval.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
1875 interval.addIntervall(posdown, posdown+1);
1877 int blk = interval.nextNotIgnored(actual._dataEnd+1);
1878 if (blk > posdown) {
1879 // Discard at most 1 space after empty item
1881 for (count = 0; count < 1; count++) {
1882 if (interval.par[blk+count] != ' ')
1886 interval.addIntervall(blk, blk+count);
1889 nextKeyIdx = getNextKey();
1892 case KeyInfo::isSectioning: {
1893 // Discard spaces before _tokenstart
1895 for (count = 0; count < actual._tokenstart; count++) {
1896 if (interval.par[actual._tokenstart-count-1] != ' ')
1899 if (actual.disabled) {
1900 removeHead(actual, count);
1901 nextKeyIdx = getNextKey();
1903 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
1904 nextKeyIdx = process(os, actual);
1908 case KeyInfo::isMath: {
1909 // Same as regex, use the content unchanged
1910 nextKeyIdx = getNextKey();
1913 case KeyInfo::isRegex: {
1914 // DO NOT SPLIT ON REGEX
1916 nextKeyIdx = getNextKey();
1919 case KeyInfo::isIgnored: {
1920 // Treat like a character for now
1921 nextKeyIdx = getNextKey();
1924 case KeyInfo::isMain: {
1925 if (interval.par.substr(actual._dataStart, 2) == "% ")
1926 interval.addIntervall(actual._dataStart, actual._dataStart+2);
1927 if (actual.disabled) {
1929 if ((interval.par.substr(actual._dataStart, 3) == " \\[") ||
1930 (interval.par.substr(actual._dataStart, 8) == " \\begin{")) {
1931 // Discard also the space before math-equation
1932 interval.addIntervall(actual._dataStart, actual._dataStart+1);
1934 interval.resetOpenedP(actual._dataStart-1);
1937 if (actual._tokenstart == 0) {
1938 // for the first (and maybe dummy) language
1939 interval.setForDefaultLang(actual._tokenstart + actual._tokensize);
1941 interval.resetOpenedP(actual._dataStart-1);
1945 case KeyInfo::invalid:
1946 // This cannot happen, already handled
1949 // LYXERR0("Unhandled keytype");
1950 nextKeyIdx = getNextKey();
1957 int LatexInfo::process(ostringstream &os, KeyInfo &actual )
1959 int end = interval.nextNotIgnored(actual._dataEnd);
1960 int oldStart = actual._dataStart;
1961 int nextKeyIdx = getNextKey();
1963 if ((nextKeyIdx < 0) ||
1964 (entries[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
1965 (entries[nextKeyIdx].keytype == KeyInfo::invalid)) {
1966 if (oldStart <= end) {
1967 processRegion(oldStart, end);
1972 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
1974 if (nextKey.keytype == KeyInfo::isMain) {
1975 (void) dispatch(os, actual._dataStart, nextKey);
1976 end = nextKey._tokenstart;
1979 processRegion(oldStart, nextKey._tokenstart);
1980 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
1982 oldStart = nextKey._dataEnd+1;
1984 // now nextKey is either invalid or is outside of actual._dataEnd
1985 // output the remaining and discard myself
1986 if (oldStart <= end) {
1987 processRegion(oldStart, end);
1989 if (interval.par[end] == '}') {
1991 // This is the normal case.
1992 // But if using the firstlanguage, the closing may be missing
1994 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
1996 if (actual._dataEnd < end)
1997 output_end = interval.nextNotIgnored(actual._dataEnd);
1999 output_end = interval.nextNotIgnored(end);
2000 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2001 interval.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2003 if (interval.nextNotIgnored(actual._dataStart) < output_end)
2004 interval.output(os, output_end);
2005 interval.addIntervall(actual._tokenstart, end);
2009 string splitOnKnownMacros(string par, bool isPatternString) {
2011 LatexInfo li(par, isPatternString);
2012 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2014 DummyKey._tokensize = 0;
2015 DummyKey._tokenstart = 0;
2016 DummyKey._dataStart = 0;
2017 DummyKey._dataEnd = par.length();
2018 DummyKey.disabled = true;
2019 int firstkeyIdx = li.getFirstKey();
2021 if (firstkeyIdx >= 0) {
2022 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2024 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2025 // Use dummy firstKey
2026 firstKey = DummyKey;
2027 (void) li.setNextKey(firstkeyIdx);
2030 if (par.substr(firstKey._dataStart, 2) == "% ")
2031 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2033 nextkeyIdx = li.process(os, firstKey);
2034 while (nextkeyIdx >= 0) {
2035 // Check for a possible gap between the last
2036 // entry and this one
2037 int datastart = li.nextNotIgnored(firstKey._dataStart);
2038 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2039 if ((nextKey._tokenstart > datastart)) {
2041 firstKey._dataStart = datastart;
2042 firstKey._dataEnd = par.length();
2043 (void) li.setNextKey(nextkeyIdx);
2044 if (firstKey._tokensize > 0) {
2045 // Fake the last opened parenthesis
2046 li.setForDefaultLang(firstKey._tokensize);
2048 nextkeyIdx = li.process(os, firstKey);
2051 if (nextKey.keytype != KeyInfo::isMain) {
2052 firstKey._dataStart = datastart;
2053 firstKey._dataEnd = nextKey._dataEnd+1;
2054 (void) li.setNextKey(nextkeyIdx);
2055 if (firstKey._tokensize > 0)
2056 li.setForDefaultLang(firstKey._tokensize);
2057 nextkeyIdx = li.process(os, firstKey);
2060 nextkeyIdx = li.process(os, nextKey);
2064 // Handle the remaining
2065 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2066 firstKey._dataEnd = par.length();
2068 if ((firstKey._dataStart < firstKey._dataEnd) &&
2069 (par[firstKey._dataStart] != '}')) {
2070 if (firstKey._tokensize > 0)
2071 li.setForDefaultLang(firstKey._tokensize);
2072 (void) li.process(os, firstKey);
2076 // return string definitelly impossible to match
2077 s = "\\foreignlanguage{ignore}{ }";
2081 s = par; /* no known macros found */
2086 * Try to unify the language specs in the latexified text.
2087 * Resulting modified string is set to "", if
2088 * the searched tex does not contain all the features in the search pattern
2090 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2092 static Features regex_f;
2093 static int missed = 0;
2094 static bool regex_with_format = false;
2096 int parlen = par.length();
2098 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2103 // Split the latex input into pieces which
2104 // can be digested by our search engine
2105 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2106 result = splitOnKnownMacros(par, isPatternString);
2107 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2110 result = par.substr(0, parlen);
2111 if (isPatternString) {
2114 regex_f = identifyFeatures(result);
2115 string features = "";
2116 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2117 string a = it->first;
2118 regex_with_format = true;
2119 features += " " + a;
2120 // LYXERR0("Identified regex format:" << a);
2122 LYXERR(Debug::FIND, "Identified Features" << features);
2125 } else if (regex_with_format) {
2126 Features info = identifyFeatures(result);
2127 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2128 string a = it->first;
2129 bool b = it->second;
2130 if (b && ! info[a]) {
2132 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2138 // LYXERR0("No regex formats");
2144 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2145 static int identifyClosing(string & t)
2147 int open_braces = 0;
2149 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2150 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2152 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]" REGEX_EOS, "$1"))
2154 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2156 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2166 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2167 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2169 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2170 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2171 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2172 // When using regexp, braces are hacked already by escape_for_regex()
2173 par_as_string = normalize(ds, !use_regexp);
2175 close_wildcards = 0;
2177 size_t lead_size = 0;
2178 // correct the language settings
2179 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2180 if (opt.ignoreformat) {
2182 // if par_as_string_nolead were emty,
2183 // the following call to findAux will always *find* the string
2184 // in the checked data, and thus always using the slow
2185 // examining of the current text part.
2186 par_as_string_nolead = par_as_string;
2189 lead_size = identifyLeading(par_as_string);
2190 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2191 lead_as_string = par_as_string.substr(0, lead_size);
2192 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2196 open_braces = identifyClosing(par_as_string);
2197 identifyClosing(par_as_string_nolead);
2198 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2199 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2201 string lead_as_regexp;
2202 if (lead_size > 0) {
2203 // @todo No need to search for \regexp{} insets in leading material
2204 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2205 par_as_string = par_as_string_nolead;
2206 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2207 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2209 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2210 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2211 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2213 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2214 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2215 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2216 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2217 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2218 || regex_replace(par_as_string, par_as_string,
2219 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2220 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2221 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2225 if (!opt.ignoreformat) {
2226 // Remove extra '\}' at end
2227 while ( regex_replace(par_as_string, par_as_string, "(.*)\\\\}$", "$1")) {
2232 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2233 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2234 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2235 // replace '[^...]' with '[^...\}\{\\]'
2236 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2237 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2238 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2240 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2243 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2244 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2245 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2246 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2248 // If entered regexp must match at begin of searched string buffer
2249 // Kornel: Added parentheses to use $1 for size of the leading string
2253 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2254 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2255 // so the convert has no effect in that case
2256 for (int i = 8; i > 0; --i) {
2257 string orig = "\\\\" + std::to_string(i);
2258 string dest = "\\" + std::to_string(i+1);
2259 while (regex_replace(par_as_string, par_as_string, orig, dest));
2261 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2262 regexp2_str = "(" + lead_as_regexp + ").*" + par_as_string;
2264 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2265 regexp = lyx::regex(regexp_str);
2267 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2268 regexp2 = lyx::regex(regexp2_str);
2273 // Count number of characters in string
2277 // \\[a-zA-Z]+ ==> 1
2278 static int computeSize(string s, int len)
2284 for (int i = 0; i < len; i += skip, count++) {
2287 if (isalpha(s[i+1])) {
2288 for (int j = 2; i+j < len; j++) {
2289 if (! isalpha(s[i+j])) {
2292 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2294 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2302 else if (s[i] == '{') {
2308 else if (s[i] == '-') {
2309 if (s[i+1] == '-') {
2325 int MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2328 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2331 docstring docstr = stringifyFromForSearch(opt, cur, len);
2332 string str = normalize(docstr, true);
2333 if (!opt.ignoreformat) {
2334 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2336 if (str.empty()) return(-1);
2337 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2338 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2341 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2342 regex const *p_regexp;
2343 regex_constants::match_flag_type flags;
2345 flags = regex_constants::match_continuous;
2348 flags = regex_constants::match_default;
2349 p_regexp = ®exp2;
2351 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2352 if (re_it == sregex_iterator())
2354 match_results<string::const_iterator> const & m = *re_it;
2356 if (0) { // Kornel Benko: DO NOT CHECKK
2357 // Check braces on the segment that matched the entire regexp expression,
2358 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2359 if (!braces_match(m[0].first, m[0].second, open_braces))
2363 // Check braces on segments that matched all (.*?) subexpressions,
2364 // except the last "padding" one inserted by lyx.
2365 for (size_t i = 1; i < m.size() - 1; ++i)
2366 if (!braces_match(m[i].first, m[i].second, open_braces))
2369 // Exclude from the returned match length any length
2370 // due to close wildcards added at end of regexp
2371 // and also the length of the leading (e.g. '\emph{')
2373 // Whole found string, including the leading: m[0].second - m[0].first
2374 // Size of the leading string: m[1].second - m[1].first
2375 int leadingsize = 0;
2377 leadingsize = m[1].second - m[1].first;
2379 for (size_t i = 0; i < m.size(); i++) {
2380 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2382 if (close_wildcards == 0)
2383 result = m[0].second - m[0].first;
2386 result = m[m.size() - close_wildcards].first - m[0].first;
2388 size_t pos = m.position(0);
2389 // Ignore last closing characters
2390 while (result > 0) {
2391 if (str[pos+result-1] == '}')
2396 if (result > leadingsize)
2397 result -= leadingsize;
2400 return computeSize(str.substr(pos+leadingsize,result), result);
2403 // else !use_regexp: but all code paths above return
2404 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2405 << par_as_string << "', str='" << str << "'");
2406 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2407 << lead_as_string << "', par_as_string_nolead='"
2408 << par_as_string_nolead << "'");
2411 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2412 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2413 if (str.substr(0, par_as_string.size()) == par_as_string)
2414 return par_as_string.size();
2416 size_t pos = str.find(par_as_string_nolead);
2417 if (pos != string::npos)
2418 return par_as_string.size();
2424 int MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
2426 int res = findAux(cur, len, at_begin);
2428 "res=" << res << ", at_begin=" << at_begin
2429 << ", matchword=" << opt.matchword
2430 << ", inTexted=" << cur.inTexted());
2431 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
2433 Paragraph const & par = cur.paragraph();
2434 bool ws_left = (cur.pos() > 0)
2435 ? par.isWordSeparator(cur.pos() - 1)
2437 bool ws_right = (cur.pos() + res < par.size())
2438 ? par.isWordSeparator(cur.pos() + res)
2441 "cur.pos()=" << cur.pos() << ", res=" << res
2442 << ", separ: " << ws_left << ", " << ws_right
2444 if (ws_left && ws_right)
2450 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
2453 if (! opt.casesensitive)
2454 t = lyx::to_utf8(lowercase(s));
2456 t = lyx::to_utf8(s);
2457 // Remove \n at begin
2458 while (!t.empty() && t[0] == '\n')
2461 while (!t.empty() && t[t.size() - 1] == '\n')
2462 t = t.substr(0, t.size() - 1);
2464 // Replace all other \n with spaces
2465 while ((pos = t.find("\n")) != string::npos)
2466 t.replace(pos, 1, " ");
2467 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
2468 // Kornel: Added textsl, textsf, textit, texttt and noun
2469 // + allow to seach for colored text too
2470 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
2471 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
2472 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2473 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
2474 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2476 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
2477 // FIXME - check what preceeds the brace
2479 if (opt.ignoreformat)
2480 while (regex_replace(t, t, "\\{", "_x_<")
2481 || regex_replace(t, t, "\\}", "_x_>"))
2482 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2484 while (regex_replace(t, t, "\\\\\\{", "_x_<")
2485 || regex_replace(t, t, "\\\\\\}", "_x_>"))
2486 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2493 docstring stringifyFromCursor(DocIterator const & cur, int len)
2495 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
2496 if (cur.inTexted()) {
2497 Paragraph const & par = cur.paragraph();
2498 // TODO what about searching beyond/across paragraph breaks ?
2499 // TODO Try adding a AS_STR_INSERTS as last arg
2500 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
2501 int(par.size()) : cur.pos() + len;
2502 OutputParams runparams(&cur.buffer()->params().encoding());
2503 runparams.nice = true;
2504 runparams.flavor = OutputParams::LATEX;
2505 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
2506 // No side effect of file copying and image conversion
2507 runparams.dryrun = true;
2508 LYXERR(Debug::FIND, "Stringifying with cur: "
2509 << cur << ", from pos: " << cur.pos() << ", end: " << end);
2510 return par.asString(cur.pos(), end,
2511 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
2513 } else if (cur.inMathed()) {
2514 CursorSlice cs = cur.top();
2515 MathData md = cs.cell();
2516 MathData::const_iterator it_end =
2517 (( len == -1 || cs.pos() + len > int(md.size()))
2519 : md.begin() + cs.pos() + len );
2521 for (MathData::const_iterator it = md.begin() + cs.pos();
2524 docstring s = asString(md2);
2525 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
2528 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
2533 /** Computes the LaTeX export of buf starting from cur and ending len positions
2534 * after cur, if len is positive, or at the paragraph or innermost inset end
2537 docstring latexifyFromCursor(DocIterator const & cur, int len)
2539 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
2540 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
2541 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
2542 Buffer const & buf = *cur.buffer();
2544 odocstringstream ods;
2546 OutputParams runparams(&buf.params().encoding());
2547 runparams.nice = false;
2548 runparams.flavor = OutputParams::LATEX;
2549 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
2550 // No side effect of file copying and image conversion
2551 runparams.dryrun = true;
2552 runparams.for_search = true;
2554 if (cur.inTexted()) {
2555 // @TODO what about searching beyond/across paragraph breaks ?
2556 pos_type endpos = cur.paragraph().size();
2557 if (len != -1 && endpos > cur.pos() + len)
2558 endpos = cur.pos() + len;
2559 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
2560 string(), cur.pos(), endpos);
2561 string s = lyx::to_utf8(ods.str());
2562 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
2563 return(lyx::from_utf8(s));
2564 } else if (cur.inMathed()) {
2565 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
2566 for (int s = cur.depth() - 1; s >= 0; --s) {
2567 CursorSlice const & cs = cur[s];
2568 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
2570 cs.asInsetMath()->asHullInset()->header_write(ws);
2575 CursorSlice const & cs = cur.top();
2576 MathData md = cs.cell();
2577 MathData::const_iterator it_end =
2578 ((len == -1 || cs.pos() + len > int(md.size()))
2580 : md.begin() + cs.pos() + len);
2582 for (MathData::const_iterator it = md.begin() + cs.pos();
2586 ods << asString(md2);
2587 // Retrieve the math environment type, and add '$' or '$]'
2588 // or others (\end{equation}) accordingly
2589 for (int s = cur.depth() - 1; s >= 0; --s) {
2590 CursorSlice const & cs2 = cur[s];
2591 InsetMath * inset = cs2.asInsetMath();
2592 if (inset && inset->asHullInset()) {
2594 inset->asHullInset()->footer_write(ws);
2598 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
2600 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
2606 /** Finalize an advanced find operation, advancing the cursor to the innermost
2607 ** position that matches, plus computing the length of the matching text to
2610 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
2612 // Search the foremost position that matches (avoids find of entire math
2613 // inset when match at start of it)
2615 DocIterator old_cur(cur.buffer());
2617 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
2621 } while (cur && cur.depth() > d && match(cur) > 0);
2623 if (match(cur) <= 0) return 0;
2624 LYXERR(Debug::FIND, "Ok");
2626 // Compute the match length
2628 if (cur.pos() + len > cur.lastpos())
2630 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
2631 while (cur.pos() + len <= cur.lastpos() && match(cur, len) <= 0) {
2633 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
2635 // Length of matched text (different from len param)
2636 int old_match = match(cur, len);
2639 int prev_old_match = old_match;
2643 if (match.opt.matchword)
2646 if (cur.pos() + len + step >= cur.lastpos()) {
2649 old_match = prev_old_match;
2652 new_match = match(cur, len + step);
2653 if (new_match > old_match) {
2654 prev_old_match = old_match;
2655 old_match = new_match;
2662 old_match = prev_old_match;
2666 // Greedy behaviour while matching regexps
2667 bool examining = true;
2670 // Kornel: The loop is needed, since it looks like
2671 // incrementing 'cur.pos()' does not always lead to the following
2672 // char which we could then match.
2674 if (match.opt.matchword)
2678 for (int count = 1; count < maxcnt; ++count) {
2679 if (cur.pos() + len + count > cur.lastpos()) {
2682 new_match = match(cur, len + count);
2683 if (new_match > old_match) {
2685 old_match = new_match;
2696 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
2700 while (!theApp()->longOperationCancelled() && cur) {
2701 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
2702 int match_len = match(cur, -1, false);
2703 LYXERR(Debug::FIND, "match_len: " << match_len);
2704 if (match_len > 0) {
2705 int match_len_zero_count = 0;
2706 for (; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
2707 LYXERR(Debug::FIND, "Advancing cur: " << cur);
2708 int match_len3 = match(cur, 1);
2711 int match_len2 = match(cur);
2712 LYXERR(Debug::FIND, "match_len2: " << match_len2);
2713 if (match_len2 > 0) {
2714 // Sometimes in finalize we understand it wasn't a match
2715 // and we need to continue the outest loop
2716 int len = findAdvFinalize(cur, match);
2721 if (match_len2 >= 0) {
2722 if (match_len2 == 0)
2723 match_len_zero_count++;
2725 match_len_zero_count = 0;
2728 if (++match_len_zero_count > 3) {
2729 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
2730 match_len_zero_count = 0;
2738 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
2739 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
2742 // This should exit nested insets, if any, or otherwise undefine the currsor.
2743 cur.pos() = cur.lastpos();
2744 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
2752 /// Find the most backward consecutive match within same paragraph while searching backwards.
2753 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
2755 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
2756 DocIterator tmp_cur = cur;
2757 int len = findAdvFinalize(tmp_cur, match);
2758 Inset & inset = cur.inset();
2759 for (; cur != cur_begin; cur.backwardPos()) {
2760 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
2761 DocIterator new_cur = cur;
2762 new_cur.backwardPos();
2763 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur))
2765 int new_len = findAdvFinalize(new_cur, match);
2770 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
2776 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
2780 // Backup of original position
2781 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
2782 if (cur == cur_begin)
2785 DocIterator cur_orig(cur);
2786 bool pit_changed = false;
2789 bool found_match = match(cur, -1, false);
2793 cur.pos() = cur.lastpos();
2795 cur.pos() = cur_orig.pos();
2796 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
2797 DocIterator cur_prev_iter;
2799 found_match = match(cur);
2800 LYXERR(Debug::FIND, "findBackAdv3: found_match="
2801 << found_match << ", cur: " << cur);
2803 return findMostBackwards(cur, match);
2805 // Stop if begin of document reached
2806 if (cur == cur_begin)
2808 cur_prev_iter = cur;
2812 if (cur == cur_begin)
2819 } while (!theApp()->longOperationCancelled());
2827 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
2828 DocIterator const & cur, int len)
2830 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
2832 if (!opt.ignoreformat)
2833 return latexifyFromCursor(cur, len);
2835 return stringifyFromCursor(cur, len);
2839 FindAndReplaceOptions::FindAndReplaceOptions(
2840 docstring const & find_buf_name, bool casesensitive,
2841 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
2842 docstring const & repl_buf_name, bool keep_case,
2843 SearchScope scope, SearchRestriction restr)
2844 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
2845 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
2846 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope), restr(restr)
2854 /** Check if 'len' letters following cursor are all non-lowercase */
2855 static bool allNonLowercase(Cursor const & cur, int len)
2857 pos_type beg_pos = cur.selectionBegin().pos();
2858 pos_type end_pos = cur.selectionBegin().pos() + len;
2859 if (len > cur.lastpos() + 1 - beg_pos) {
2860 LYXERR(Debug::FIND, "This should not happen, more debug needed");
2861 len = cur.lastpos() + 1 - beg_pos;
2862 end_pos = beg_pos + len;
2864 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
2865 if (isLowerCase(cur.paragraph().getChar(pos)))
2871 /** Check if first letter is upper case and second one is lower case */
2872 static bool firstUppercase(Cursor const & cur)
2875 pos_type pos = cur.selectionBegin().pos();
2876 if (pos >= cur.lastpos() - 1) {
2877 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
2880 ch1 = cur.paragraph().getChar(pos);
2881 ch2 = cur.paragraph().getChar(pos + 1);
2882 bool result = isUpperCase(ch1) && isLowerCase(ch2);
2883 LYXERR(Debug::FIND, "firstUppercase(): "
2884 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
2885 << ch2 << "(" << char(ch2) << ")"
2886 << ", result=" << result << ", cur=" << cur);
2891 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
2893 ** \fixme What to do with possible further paragraphs in replace buffer ?
2895 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
2897 ParagraphList::iterator pit = buffer.paragraphs().begin();
2898 LASSERT(pit->size() >= 1, /**/);
2899 pos_type right = pos_type(1);
2900 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
2901 right = pit->size();
2902 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
2908 static void findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
2910 Cursor & cur = bv->cursor();
2911 if (opt.repl_buf_name == docstring()
2912 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
2913 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
2916 DocIterator sel_beg = cur.selectionBegin();
2917 DocIterator sel_end = cur.selectionEnd();
2918 if (&sel_beg.inset() != &sel_end.inset()
2919 || sel_beg.pit() != sel_end.pit()
2920 || sel_beg.idx() != sel_end.idx())
2922 int sel_len = sel_end.pos() - sel_beg.pos();
2923 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
2924 << ", sel_len: " << sel_len << endl);
2927 LASSERT(sel_len > 0, return);
2929 if (!matchAdv(sel_beg, sel_len))
2932 // Build a copy of the replace buffer, adapted to the KeepCase option
2933 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
2935 repl_buffer_orig.write(oss);
2936 string lyx = oss.str();
2937 Buffer repl_buffer("", false);
2938 repl_buffer.setUnnamed(true);
2939 LASSERT(repl_buffer.readString(lyx), return);
2940 if (opt.keep_case && sel_len >= 2) {
2941 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
2942 if (cur.inTexted()) {
2943 if (firstUppercase(cur))
2944 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
2945 else if (allNonLowercase(cur, sel_len))
2946 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
2949 cap::cutSelection(cur, false);
2950 if (cur.inTexted()) {
2951 repl_buffer.changeLanguage(
2952 repl_buffer.language(),
2953 cur.getFont().language());
2954 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
2955 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
2956 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
2957 repl_buffer.params().documentClassPtr(),
2958 bv->buffer().errorList("Paste"));
2959 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
2960 sel_len = repl_buffer.paragraphs().begin()->size();
2961 } else if (cur.inMathed()) {
2962 odocstringstream ods;
2964 OutputParams runparams(&repl_buffer.params().encoding());
2965 runparams.nice = false;
2966 runparams.flavor = OutputParams::LATEX;
2967 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
2968 runparams.dryrun = true;
2969 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
2970 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
2971 docstring repl_latex = ods.str();
2972 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
2974 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
2975 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
2976 repl_latex = from_utf8(s);
2977 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
2978 MathData ar(cur.buffer());
2979 asArray(repl_latex, ar, Parse::NORMAL);
2981 sel_len = ar.size();
2982 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
2984 if (cur.pos() >= sel_len)
2985 cur.pos() -= sel_len;
2988 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
2989 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
2990 bv->processUpdateFlags(Update::Force);
2994 /// Perform a FindAdv operation.
2995 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3000 // e.g., when invoking word-findadv from mini-buffer wither with
3001 // wrong options syntax or before ever opening advanced F&R pane
3002 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3006 MatchStringAdv matchAdv(bv->buffer(), opt);
3007 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3009 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3010 findAdvReplace(bv, opt, matchAdv);
3013 match_len = findForwardAdv(cur, matchAdv);
3015 match_len = findBackwardsAdv(cur, matchAdv);
3017 // This may only be raised by lyx::regex()
3018 bv->message(_("Invalid regular expression!"));
3022 if (match_len == 0) {
3023 bv->message(_("Match not found!"));
3027 bv->message(_("Match found!"));
3029 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3030 bv->putSelectionAt(cur, match_len, !opt.forward);
3036 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3038 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3039 << opt.casesensitive << ' '
3040 << opt.matchword << ' '
3041 << opt.forward << ' '
3042 << opt.expandmacros << ' '
3043 << opt.ignoreformat << ' '
3044 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3045 << opt.keep_case << ' '
3046 << int(opt.scope) << ' '
3049 LYXERR(Debug::FIND, "built: " << os.str());
3055 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3057 LYXERR(Debug::FIND, "parsing");
3061 while (line != "EOSS") {
3065 if (is.eof()) // Tolerate malformed request
3069 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3070 opt.find_buf_name = from_utf8(s);
3071 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat;
3072 is.get(); // Waste space before replace string
3075 while (line != "EOSS") {
3079 if (is.eof()) // Tolerate malformed request
3083 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3084 opt.repl_buf_name = from_utf8(s);
3085 is >> opt.keep_case;
3088 opt.scope = FindAndReplaceOptions::SearchScope(i);
3090 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3092 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3093 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3094 << opt.scope << ' ' << opt.restr);