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"
55 #include "support/textutils.h"
59 using namespace lyx::support;
64 // Helper class for deciding what should be ignored
69 : ignoreFamily_(false), ignoreSeries_(false),
70 ignoreShape_(false), ignoreUnderline_(false),
71 ignoreMarkUp_(false), ignoreStrikeOut_(false),
72 ignoreSectioning_(false), ignoreFrontMatter_(false),
73 ignoreColor_(false), ignoreLanguage_(false) {}
75 bool getFamily() { return ignoreFamily_; };
77 bool getSeries() { return ignoreSeries_; };
79 bool getShape() { return ignoreShape_; };
81 bool getUnderline() { return ignoreUnderline_; };
83 bool getMarkUp() { return ignoreMarkUp_; };
85 bool getStrikeOut() { return ignoreStrikeOut_; };
87 bool getSectioning() { return ignoreSectioning_; };
89 bool getFrontMatter() { return ignoreFrontMatter_; };
91 bool getColor() { return ignoreColor_; };
93 bool getLanguage() { return ignoreLanguage_; };
95 void setIgnoreFormat(string type, bool value);
105 bool ignoreUnderline_;
109 bool ignoreStrikeOut_;
111 bool ignoreSectioning_;
113 bool ignoreFrontMatter_;
117 bool ignoreLanguage_;
121 void IgnoreFormats::setIgnoreFormat(string type, bool value)
123 if (type == "color") {
124 ignoreColor_ = value;
126 else if (type == "language") {
127 ignoreLanguage_ = value;
129 else if (type == "sectioning") {
130 ignoreSectioning_ = value;
131 ignoreFrontMatter_ = value;
133 else if (type == "font") {
134 ignoreSeries_ = value;
135 ignoreShape_ = value;
136 ignoreFamily_ = value;
138 else if (type == "series") {
139 ignoreSeries_ = value;
141 else if (type == "shape") {
142 ignoreShape_ = value;
144 else if (type == "family") {
145 ignoreFamily_ = value;
147 else if (type == "markup") {
148 ignoreMarkUp_ = value;
150 else if (type == "underline") {
151 ignoreUnderline_ = value;
153 else if (type == "strike") {
154 ignoreStrikeOut_ = value;
158 // The global variable that can be changed from outside
159 IgnoreFormats ignoreFormats;
162 void setIgnoreFormat(string type, bool value)
164 ignoreFormats.setIgnoreFormat(type, value);
170 bool parse_bool(docstring & howto)
175 howto = split(howto, var, ' ');
180 class MatchString : public binary_function<Paragraph, pos_type, int>
183 MatchString(docstring const & str, bool cs, bool mw)
184 : str(str), case_sens(cs), whole_words(mw)
187 // returns true if the specified string is at the specified position
188 // del specifies whether deleted strings in ct mode will be considered
189 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
191 return par.find(str, case_sens, whole_words, pos, del);
199 // match whole words only
204 int findForward(DocIterator & cur, MatchString const & match,
205 bool find_del = true)
207 for (; cur; cur.forwardChar())
208 if (cur.inTexted()) {
209 int len = match(cur.paragraph(), cur.pos(), find_del);
217 int findBackwards(DocIterator & cur, MatchString const & match,
218 bool find_del = true)
222 if (cur.inTexted()) {
223 int len = match(cur.paragraph(), cur.pos(), find_del);
232 bool searchAllowed(docstring const & str)
235 frontend::Alert::error(_("Search error"), _("Search string is empty"));
242 bool findOne(BufferView * bv, docstring const & searchstr,
243 bool case_sens, bool whole, bool forward,
244 bool find_del = true, bool check_wrap = false)
246 if (!searchAllowed(searchstr))
249 DocIterator cur = forward
250 ? bv->cursor().selectionEnd()
251 : bv->cursor().selectionBegin();
253 MatchString const match(searchstr, case_sens, whole);
255 int match_len = forward
256 ? findForward(cur, match, find_del)
257 : findBackwards(cur, match, find_del);
260 bv->putSelectionAt(cur, match_len, !forward);
261 else if (check_wrap) {
262 DocIterator cur_orig(bv->cursor());
265 q = _("End of file reached while searching forward.\n"
266 "Continue searching from the beginning?");
268 q = _("Beginning of file reached while searching backward.\n"
269 "Continue searching from the end?");
270 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
271 q, 0, 1, _("&Yes"), _("&No"));
272 if (wrap_answer == 0) {
274 bv->cursor().clear();
275 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
277 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
278 bv->cursor().backwardPos();
280 bv->clearSelection();
281 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
284 bv->cursor().setCursor(cur_orig);
288 return match_len > 0;
292 int replaceAll(BufferView * bv,
293 docstring const & searchstr, docstring const & replacestr,
294 bool case_sens, bool whole)
296 Buffer & buf = bv->buffer();
298 if (!searchAllowed(searchstr) || buf.isReadonly())
301 DocIterator cur_orig(bv->cursor());
303 MatchString const match(searchstr, case_sens, whole);
306 int const rsize = replacestr.size();
307 int const ssize = searchstr.size();
310 cur.setCursor(doc_iterator_begin(&buf));
311 int match_len = findForward(cur, match, false);
312 while (match_len > 0) {
313 // Backup current cursor position and font.
314 pos_type const pos = cur.pos();
315 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
317 int striked = ssize -
318 cur.paragraph().eraseChars(pos, pos + match_len,
319 buf.params().track_changes);
320 cur.paragraph().insert(pos, replacestr, font,
321 Change(buf.params().track_changes
323 : Change::UNCHANGED));
324 for (int i = 0; i < rsize + striked; ++i)
327 match_len = findForward(cur, match, false);
330 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
332 cur_orig.fixIfBroken();
333 bv->setCursor(cur_orig);
339 // the idea here is that we are going to replace the string that
340 // is selected IF it is the search string.
341 // if there is a selection, but it is not the search string, then
342 // we basically ignore it. (FIXME We ought to replace only within
344 // if there is no selection, then:
345 // (i) if some search string has been provided, then we find it.
346 // (think of how the dialog works when you hit "replace" the
348 // (ii) if no search string has been provided, then we treat the
349 // word the cursor is in as the search string. (why? i have no
350 // idea.) but this only works in text?
352 // returns the number of replacements made (one, if any) and
353 // whether anything at all was done.
354 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
355 docstring const & replacestr, bool case_sens,
356 bool whole, bool forward, bool findnext)
358 Cursor & cur = bv->cursor();
360 if (!cur.selection()) {
361 // no selection, non-empty search string: find it
362 if (!searchstr.empty()) {
363 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
364 return make_pair(found, 0);
366 // empty search string
369 return make_pair(false, 0);
370 // select current word and treat it as the search string.
371 // This causes a minor bug as undo will restore this selection,
372 // which the user did not create (#8986).
373 cur.innerText()->selectWord(cur, WHOLE_WORD);
374 searchstr = cur.selectionAsString(false);
377 // if we still don't have a search string, report the error
379 if (!searchAllowed(searchstr))
380 return make_pair(false, 0);
382 bool have_selection = cur.selection();
383 docstring const selected = cur.selectionAsString(false);
386 ? searchstr == selected
387 : compare_no_case(searchstr, selected) == 0;
389 // no selection or current selection is not search word:
390 // just find the search word
391 if (!have_selection || !match) {
392 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
393 return make_pair(found, 0);
396 // we're now actually ready to replace. if the buffer is
397 // read-only, we can't, though.
398 if (bv->buffer().isReadonly())
399 return make_pair(false, 0);
401 cap::replaceSelectionWithString(cur, replacestr);
403 cur.pos() += replacestr.length();
404 LASSERT(cur.pos() <= cur.lastpos(),
405 cur.pos() = cur.lastpos());
408 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
410 return make_pair(true, 1);
416 docstring const find2string(docstring const & search,
417 bool casesensitive, bool matchword, bool forward)
421 << int(casesensitive) << ' '
422 << int(matchword) << ' '
428 docstring const replace2string(docstring const & replace,
429 docstring const & search,
430 bool casesensitive, bool matchword,
431 bool all, bool forward, bool findnext)
434 ss << replace << '\n'
436 << int(casesensitive) << ' '
437 << int(matchword) << ' '
439 << int(forward) << ' '
445 bool lyxfind(BufferView * bv, FuncRequest const & ev)
447 if (!bv || ev.action() != LFUN_WORD_FIND)
450 //lyxerr << "find called, cmd: " << ev << endl;
452 // data is of the form
454 // <casesensitive> <matchword> <forward>"
456 docstring howto = split(ev.argument(), search, '\n');
458 bool casesensitive = parse_bool(howto);
459 bool matchword = parse_bool(howto);
460 bool forward = parse_bool(howto);
462 return findOne(bv, search, casesensitive, matchword, forward, true, true);
466 bool lyxreplace(BufferView * bv,
467 FuncRequest const & ev, bool has_deleted)
469 if (!bv || ev.action() != LFUN_WORD_REPLACE)
472 // data is of the form
475 // <casesensitive> <matchword> <all> <forward> <findnext>"
478 docstring howto = split(ev.argument(), rplc, '\n');
479 howto = split(howto, search, '\n');
481 bool casesensitive = parse_bool(howto);
482 bool matchword = parse_bool(howto);
483 bool all = parse_bool(howto);
484 bool forward = parse_bool(howto);
485 bool findnext = howto.empty() ? true : parse_bool(howto);
490 int replace_count = 0;
492 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
493 update = replace_count > 0;
496 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
498 replace_count = rv.second;
501 Buffer const & buf = bv->buffer();
503 // emit message signal.
504 buf.message(_("String not found."));
506 if (replace_count == 0) {
507 buf.message(_("String found."));
508 } else if (replace_count == 1) {
509 buf.message(_("String has been replaced."));
511 docstring const str =
512 bformat(_("%1$d strings have been replaced."), replace_count);
516 } else if (findnext) {
517 // if we have deleted characters, we do not replace at all, but
518 // rather search for the next occurence
519 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
522 bv->message(_("String not found."));
528 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
530 for (; cur; cur.forwardPos())
531 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
535 DocIterator cur_orig(bv->cursor());
536 docstring q = _("End of file reached while searching forward.\n"
537 "Continue searching from the beginning?");
538 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
539 q, 0, 1, _("&Yes"), _("&No"));
540 if (wrap_answer == 0) {
541 bv->cursor().clear();
542 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
543 bv->clearSelection();
544 cur.setCursor(bv->cursor().selectionBegin());
545 if (findNextChange(bv, cur, false))
548 bv->cursor().setCursor(cur_orig);
555 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
557 for (cur.backwardPos(); cur; cur.backwardPos()) {
558 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
563 DocIterator cur_orig(bv->cursor());
564 docstring q = _("Beginning of file reached while searching backward.\n"
565 "Continue searching from the end?");
566 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
567 q, 0, 1, _("&Yes"), _("&No"));
568 if (wrap_answer == 0) {
569 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
570 bv->cursor().backwardPos();
571 bv->clearSelection();
572 cur.setCursor(bv->cursor().selectionBegin());
573 if (findPreviousChange(bv, cur, false))
576 bv->cursor().setCursor(cur_orig);
583 bool selectChange(Cursor & cur, bool forward)
585 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
587 Change ch = cur.paragraph().lookupChange(cur.pos());
589 CursorSlice tip1 = cur.top();
590 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
591 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
592 if (!ch2.isSimilarTo(ch))
595 CursorSlice tip2 = cur.top();
596 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
598 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
599 if (!ch2.isSimilarTo(ch)) {
600 // take a step forward to correctly set the selection
608 cur.bv().mouseSetCursor(cur, false);
610 cur.bv().mouseSetCursor(cur, true);
618 bool findChange(BufferView * bv, bool forward)
621 cur.setCursor(forward ? bv->cursor().selectionEnd()
622 : bv->cursor().selectionBegin());
623 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
624 return selectChange(cur, forward);
629 bool findNextChange(BufferView * bv)
631 return findChange(bv, true);
635 bool findPreviousChange(BufferView * bv)
637 return findChange(bv, false);
644 typedef vector<pair<string, string> > Escapes;
646 /// A map of symbols and their escaped equivalent needed within a regex.
647 /// @note Beware of order
648 Escapes const & get_regexp_escapes()
650 typedef std::pair<std::string, std::string> P;
652 static Escapes escape_map;
653 if (escape_map.empty()) {
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(".", "_x_."));
664 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
665 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
666 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
667 escape_map.push_back(P("_x_", "\\"));
672 /// A map of lyx escaped strings and their unescaped equivalent.
673 Escapes const & get_lyx_unescapes()
675 typedef std::pair<std::string, std::string> P;
677 static Escapes escape_map;
678 if (escape_map.empty()) {
679 escape_map.push_back(P("\\%", "%"));
680 escape_map.push_back(P("\\mathcircumflex ", "^"));
681 escape_map.push_back(P("\\mathcircumflex", "^"));
682 escape_map.push_back(P("\\backslash ", "\\"));
683 escape_map.push_back(P("\\backslash", "\\"));
684 escape_map.push_back(P("\\\\{", "_x_<"));
685 escape_map.push_back(P("\\\\}", "_x_>"));
686 escape_map.push_back(P("\\sim ", "~"));
687 escape_map.push_back(P("\\sim", "~"));
692 /// A map of escapes turning a regexp matching text to one matching latex.
693 Escapes const & get_regexp_latex_escapes()
695 typedef std::pair<std::string, std::string> P;
697 static Escapes escape_map;
698 if (escape_map.empty()) {
699 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
700 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
701 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
702 escape_map.push_back(P("\\[", "\\{\\[\\}"));
703 escape_map.push_back(P("\\]", "\\{\\]\\}"));
704 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
705 escape_map.push_back(P("%", "\\\\\\%"));
706 escape_map.push_back(P("#", "\\\\#"));
711 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
712 ** the found occurrence were escaped.
714 string apply_escapes(string s, Escapes const & escape_map)
716 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
717 Escapes::const_iterator it;
718 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
719 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
720 unsigned int pos = 0;
721 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
722 s.replace(pos, it->first.length(), it->second);
723 LYXERR(Debug::FIND, "After escape: " << s);
724 pos += it->second.length();
725 // LYXERR(Debug::FIND, "pos: " << pos);
728 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
733 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
734 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
735 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
736 string escape_for_regex(string s, bool match_latex)
739 while (pos < s.size()) {
740 size_t new_pos = s.find("\\regexp{", pos);
741 if (new_pos == string::npos)
745 LYXERR(Debug::FIND, "new_pos: " << new_pos);
746 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
747 LYXERR(Debug::FIND, "t [lyx]: " << t);
748 t = apply_escapes(t, get_regexp_escapes());
749 LYXERR(Debug::FIND, "t [rxp]: " << t);
750 s.replace(pos, new_pos - pos, t);
751 new_pos = pos + t.size();
752 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
753 LYXERR(Debug::FIND, "new_pos: " << new_pos);
754 if (new_pos == s.size())
757 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
758 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
759 LYXERR(Debug::FIND, "end_pos: " << end_pos);
760 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
761 LYXERR(Debug::FIND, "t in regexp : " << t);
762 t = apply_escapes(t, get_lyx_unescapes());
763 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
765 t = apply_escapes(t, get_regexp_latex_escapes());
766 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
768 if (end_pos == s.size()) {
769 s.replace(new_pos, end_pos - new_pos, t);
770 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
773 s.replace(new_pos, end_pos + 13 - new_pos, t);
774 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
775 pos = new_pos + t.size();
776 LYXERR(Debug::FIND, "pos: " << pos);
782 /// Wrapper for lyx::regex_replace with simpler interface
783 bool regex_replace(string const & s, string & t, string const & searchstr,
784 string const & replacestr)
786 lyx::regex e(searchstr, regex_constants::ECMAScript);
788 ostream_iterator<char, char> it(oss);
789 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
790 // tolerate t and s be references to the same variable
791 bool rv = (s != oss.str());
797 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
799 ** Verify that closed braces exactly match open braces. This avoids that, for example,
800 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
803 ** Number of open braces that must remain open at the end for the verification to succeed.
805 bool braces_match(string::const_iterator const & beg,
806 string::const_iterator const & end,
810 string::const_iterator it = beg;
811 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
812 for (; it != end; ++it) {
813 // Skip escaped braces in the count
818 } else if (*it == '{') {
820 } else if (*it == '}') {
821 if (open_pars == 0) {
822 LYXERR(Debug::FIND, "Found unmatched closed brace");
828 if (open_pars != unmatched) {
829 LYXERR(Debug::FIND, "Found " << open_pars
830 << " instead of " << unmatched
831 << " unmatched open braces at the end of count");
834 LYXERR(Debug::FIND, "Braces match as expected");
844 MatchResult(): match_len(0),match2end(0), pos(0) {};
847 /** The class performing a match between a position in the document and the FindAdvOptions.
849 class MatchStringAdv {
851 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
853 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
854 ** constructor as opt.search, under the opt.* options settings.
857 ** If set, then match is searched only against beginning of text starting at cur.
858 ** If unset, then match is searched anywhere in text starting at cur.
861 ** The length of the matching text, or zero if no match was found.
863 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
868 /// first buffer on which search was started
869 lyx::Buffer * const p_first_buf;
871 FindAndReplaceOptions const & opt;
874 /// Auxiliary find method (does not account for opt.matchword)
875 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
877 /** Normalize a stringified or latexified LyX paragraph.
881 ** <li>if search is not casesensitive, then lowercase the string;
882 ** <li>remove any newline at begin or end of the string;
883 ** <li>replace any newline in the middle of the string with a simple space;
884 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
887 ** @todo Normalization should also expand macros, if the corresponding
888 ** search option was checked.
890 string normalize(docstring const & s, bool hack_braces) const;
891 // normalized string to search
892 string par_as_string;
893 // regular expression to use for searching
895 // same as regexp, but prefixed with a ".*?"
897 // leading format material as string
898 string lead_as_string;
899 // par_as_string after removal of lead_as_string
900 string par_as_string_nolead;
901 // unmatched open braces in the search string/regexp
903 // number of (.*?) subexpressions added at end of search regexp for closing
904 // environments, math mode, styles, etc...
906 // Are we searching with regular expressions ?
911 static docstring buffer_to_latex(Buffer & buffer)
913 OutputParams runparams(&buffer.params().encoding());
914 odocstringstream ods;
916 runparams.nice = true;
917 runparams.flavor = OutputParams::XETEX;
918 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
919 // No side effect of file copying and image conversion
920 runparams.dryrun = true;
921 runparams.for_search = true;
922 pit_type const endpit = buffer.paragraphs().size();
923 for (pit_type pit = 0; pit != endpit; ++pit) {
924 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
925 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
931 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
934 if (!opt.ignoreformat) {
935 str = buffer_to_latex(buffer);
937 OutputParams runparams(&buffer.params().encoding());
938 runparams.nice = true;
939 runparams.flavor = OutputParams::XETEX;
940 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
941 runparams.dryrun = true;
942 runparams.for_search = true;
943 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
944 Paragraph const & par = buffer.paragraphs().at(pit);
945 LYXERR(Debug::FIND, "Adding to search string: '"
946 << par.asString(pos_type(0), par.size(),
947 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
950 str += par.asString(pos_type(0), par.size(),
951 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
959 /// Return separation pos between the leading material and the rest
960 static size_t identifyLeading(string const & s)
963 // @TODO Support \item[text]
964 // Kornel: Added textsl, textsf, textit, texttt and noun
965 // + allow to search for colored text too
966 while (regex_replace(t, t, REGEX_BOS "\\\\(((footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|emph|noun|minisec|text(bf|md|sl|sf|it|tt))|((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
967 || regex_replace(t, t, REGEX_BOS "\\$", "")
968 || regex_replace(t, t, REGEX_BOS "\\\\\\[ ", "")
969 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
970 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
972 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
977 * Given a latexified string, retrieve some handled features
978 * The features of the regex will later be compared with the features
979 * of the searched text. If the regex features are not a
980 * subset of the analized, then, in not format ignoring search
981 * we can early stop the search in the relevant inset.
983 typedef map<string, bool> Features;
985 static Features identifyFeatures(string const & s)
987 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
988 static regex const valid("^("
990 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
991 "emph|noun|text(bf|md|sl|sf|it|tt)|"
992 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
993 "(u|uu)line|(s|x)out|uwave|"
994 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
995 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1000 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1003 if (sub.str(1).compare("regexp") == 0) {
1007 string token = sub.str(1);
1009 if (regex_match(token, sub2, valid)) {
1017 if (sub.str(1).compare("endregexp") == 0) {
1027 * defines values features of a key "\\[a-z]+{"
1032 /* Char type with content discarded
1033 * like \hspace{1cm} */
1035 /* Char, like \backslash */
1037 /* \part, \section*, ... */
1039 /* title, author etc */
1041 /* \foreignlanguage{ngerman}, ... */
1044 * to discard language in content */
1047 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1049 /* fonts, colors, markups, ... */
1051 /* footnotesize, ... large, ...
1052 * Ignore all of them */
1055 /* inputencoding, ...
1056 * Discard also content, because they do not help in search */
1059 * like remove, but also all arguments */
1061 /* item, listitem */
1063 /* tex, latex, ... like isChar */
1065 /* like \lettrine[lines=5]{}{} */
1067 /* End of arguments marker for lettrine,
1068 * so that they can be ignored */
1078 parenthesiscount(1),
1082 KeyInfo(KeyType type, int parcount, bool disable)
1088 parenthesiscount(parcount),
1097 int parenthesiscount;
1099 bool used; /* by pattern */
1104 Border(int l=0, int u=0) : low(l), upper(u) {};
1109 #define MAXOPENED 30
1111 bool isPatternString;
1113 explicit Intervall(bool isPattern) :
1114 isPatternString(isPattern),
1116 actualdeptindex(0) { depts[0] = 0; closes[0] = 0;};
1119 int depts[MAXOPENED];
1120 int closes[MAXOPENED];
1121 int actualdeptindex;
1122 Border borders[2*MAXOPENED];
1123 int previousNotIgnored(int);
1124 int nextNotIgnored(int);
1125 void handleOpenP(int i);
1126 void handleCloseP(int i, bool closingAllowed);
1127 void resetOpenedP(int openPos);
1128 void addIntervall(int upper);
1129 void addIntervall(int low, int upper); /* if explicit */
1130 void removeAccents();
1131 void setForDefaultLang(KeyInfo &defLang);
1132 int findclosing(int start, int end, char up, char down, int repeat);
1133 void handleParentheses(int lastpos, bool closingAllowed);
1135 int isOpeningPar(int pos);
1137 void output(ostringstream &os, int lastpos);
1138 // string show(int lastpos);
1141 int Intervall::isOpeningPar(int pos)
1143 if ((pos < 0) || (size_t(pos) >= par.size()))
1145 if (par[pos] != '{')
1147 if (size_t(pos) + 2 >= par.size())
1149 if (par[pos+2] != '}')
1151 if (par[pos+1] == '[' || par[pos+1] == ']')
1156 void Intervall::setForDefaultLang(KeyInfo &defLang)
1158 // Enable the use of first token again
1159 if (ignoreidx >= 0) {
1160 int value = defLang._tokenstart + defLang._tokensize;
1162 if (borders[0].low < value)
1163 borders[0].low = value;
1164 if (borders[0].upper < value)
1165 borders[0].upper = value;
1170 static void checkDepthIndex(int val)
1172 static int maxdepthidx = MAXOPENED-2;
1173 if (val > maxdepthidx) {
1175 LYXERR0("maxdepthidx now " << val);
1179 static void checkIgnoreIdx(int val)
1181 static int maxignoreidx = 2*MAXOPENED - 4;
1182 if (val > maxignoreidx) {
1184 LYXERR0("maxignoreidx now " << val);
1189 * Expand the region of ignored parts of the input latex string
1190 * The region is only relevant in output()
1192 void Intervall::addIntervall(int low, int upper)
1195 if (low == upper) return;
1196 for (idx = ignoreidx+1; idx > 0; --idx) {
1197 if (low > borders[idx-1].upper) {
1201 Border br(low, upper);
1202 if (idx > ignoreidx) {
1205 checkIgnoreIdx(ignoreidx);
1209 // Expand only if one of the new bound is inside the interwall
1210 // We know here that br.low > borders[idx-1].upper
1211 if (br.upper < borders[idx].low) {
1212 // We have to insert at this pos
1213 for (int i = ignoreidx+1; i > idx; --i) {
1214 borders[i] = borders[i-1];
1218 checkIgnoreIdx(ignoreidx);
1221 // Here we know, that we are overlapping
1222 if (br.low > borders[idx].low)
1223 br.low = borders[idx].low;
1224 // check what has to be concatenated
1226 for (int i = idx; i <= ignoreidx; i++) {
1227 if (br.upper >= borders[i].low) {
1229 if (br.upper < borders[i].upper)
1230 br.upper = borders[i].upper;
1236 // count should be >= 1 here
1239 for (int i = idx + count; i <= ignoreidx; i++) {
1240 borders[i-count+1] = borders[i];
1242 ignoreidx -= count - 1;
1248 typedef map<string, string> AccentsMap;
1249 static AccentsMap accents = map<string, string>();
1251 static void buildaccent(string n, string param, string values)
1255 const char delim = '|';
1256 while (getline(s, name, delim)) {
1258 for (size_t i = 0; i < param.size(); i++) {
1259 string key = name + "{" + param[i] + "}";
1260 // get the corresponding utf8-value
1261 if ((values[start] & 0xc0) != 0xc0) {
1262 // should not happen, utf8 encoding starts at least with 11xxxxxx
1266 for (int j = 1; ;j++) {
1267 if (start + j >= values.size())
1269 if ((values[start+j] & 0xc0) == 0xc0) {
1270 // This is the first byte of following utf8 char
1271 accents[key] = values.substr(start, j);
1280 static void buildAccentsMap()
1282 accents["imath"] = "ı";
1283 accents["ddot{\\imath}"] = "ï";
1284 accents["acute{\\imath}"] = "í";
1285 accents["tilde{\\imath}"] = "ĩ";
1286 accents["jmath"] = "ȷ";
1287 accents["hat{\\jmath}"] = "ĵ";
1288 accents["lyxmathsym{ß}"] = "ß";
1289 buildaccent("ddot", "aeouyAEOUY", "äëöüÿÄËÖÜŸ"); // umlaut
1290 buildaccent("dot", "aeoyzAEOYZ", "ȧėȯẏżȦĖȮẎŻ");
1291 buildaccent("acute", "aAcCeElLoOnNrRsSuUyYzZI", "áÁćĆéÉĺĹóÓńŃŕŔśŚúÚýÝźŹÍ");
1293 buildaccent("dacute", "oOuU", "őŐűŰ");
1294 buildaccent("H", "oOuU", "őŐűŰ"); // dacute in text
1296 buildaccent("mathring|r", "uU", "ůŮ");
1297 buildaccent("check", "cCdDeElLnNrRsSTzZ", "čČďĎěĚľĽňŇřŘšŠŤžŽ"); // caron
1298 buildaccent("hat", "cCgGhHJsSwWyYoOgG", "ĉĈĝĜĥĤĴŝŜŵŴŷŶôÔĝĜ"); // circ
1299 buildaccent("bar|=", "aAeEoOuU", "āĀēĒōŌūŪ"); // macron
1300 buildaccent("tilde", "I", "Ĩ"); // macron
1304 * Created accents in math or regexp environment
1305 * are macros, but we need the utf8 equivalent
1307 void Intervall::removeAccents()
1309 if (accents.empty())
1311 static regex const accre("\\\\((lyxmathsym|ddot|dot|acute|mathring|r|check|check|hat|bar|=)\\{[^\\{\\}]+\\}|imath|jmath)");
1313 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1315 string key = sub.str(1);
1316 if (accents.find(key) != accents.end()) {
1317 string val = accents[key];
1318 size_t pos = sub.position(0);
1319 for (size_t i = 0; i < val.size(); i++) {
1320 par[pos+i] = val[i];
1322 addIntervall(pos+val.size(), pos + sub.str(0).size());
1325 LYXERR0("Not added accent for \"" << key << "\"");
1330 void Intervall::handleOpenP(int i)
1333 depts[actualdeptindex] = i+1;
1334 closes[actualdeptindex] = -1;
1335 checkDepthIndex(actualdeptindex);
1338 void Intervall::handleCloseP(int i, bool closingAllowed)
1340 if (actualdeptindex <= 0) {
1341 if (! closingAllowed)
1342 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1343 // if we are at the very end
1344 addIntervall(i, i+1);
1347 closes[actualdeptindex] = i+1;
1352 void Intervall::resetOpenedP(int openPos)
1354 // Used as initializer for foreignlanguage entry
1355 actualdeptindex = 1;
1356 depts[1] = openPos+1;
1360 int Intervall::previousNotIgnored(int start)
1362 int idx = 0; /* int intervalls */
1363 for (idx = ignoreidx; idx >= 0; --idx) {
1364 if (start > borders[idx].upper)
1366 if (start >= borders[idx].low)
1367 start = borders[idx].low-1;
1372 int Intervall::nextNotIgnored(int start)
1374 int idx = 0; /* int intervalls */
1375 for (idx = 0; idx <= ignoreidx; idx++) {
1376 if (start < borders[idx].low)
1378 if (start < borders[idx].upper)
1379 start = borders[idx].upper;
1384 typedef map<string, KeyInfo> KeysMap;
1385 typedef vector< KeyInfo> Entries;
1386 static KeysMap keys = map<string, KeyInfo>();
1393 void buildKeys(bool);
1394 void buildEntries(bool);
1395 void makeKey(const string &, KeyInfo, bool isPatternString);
1396 void processRegion(int start, int region_end); /* remove {} parts */
1397 void removeHead(KeyInfo&, int count=0);
1400 LatexInfo(string par, bool isPatternString) : entidx(-1), interval(isPatternString) {
1402 interval.hasTitle = false;
1403 interval.titleValue = "";
1404 buildKeys(isPatternString);
1405 entries = vector<KeyInfo>();
1406 buildEntries(isPatternString);
1410 if (entries.empty()) {
1413 if (entries[0].keytype == KeyInfo::isTitle) {
1414 if (! entries[0].disabled) {
1415 interval.hasTitle = true;
1416 interval.titleValue = entries[0].head;
1419 interval.hasTitle = false;
1420 interval.titleValue = "";
1422 removeHead(entries[0]);
1423 if (entries.size() > 1)
1432 if (int(entries.size()) > entidx) {
1439 bool setNextKey(int idx) {
1440 if ((idx == entidx) && (entidx >= 0)) {
1447 int find(int start, KeyInfo::KeyType keytype) {
1451 while (tmpIdx < int(entries.size())) {
1452 if (entries[tmpIdx].keytype == keytype)
1458 int process(ostringstream &os, KeyInfo &actual);
1459 int dispatch(ostringstream &os, int previousStart, KeyInfo &actual);
1460 // string show(int lastpos) { return interval.show(lastpos);};
1461 int nextNotIgnored(int start) { return interval.nextNotIgnored(start);};
1462 KeyInfo &getKeyInfo(int keyinfo) {
1463 static KeyInfo invalidInfo = KeyInfo();
1464 if ((keyinfo < 0) || ( keyinfo >= int(entries.size())))
1467 return entries[keyinfo];
1469 void setForDefaultLang(KeyInfo &defLang) {interval.setForDefaultLang(defLang);};
1470 void addIntervall(int low, int up) { interval.addIntervall(low, up); };
1474 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1479 for (int i = start; i < end; i += 1 + skip) {
1483 if (c == '\\') skip = 1;
1487 else if (c == down) {
1489 if ((repeat <= 0) || (par[i+1] != up))
1507 vector<MathEntry> entries;
1512 void insert(string wait, size_t start, size_t end) {
1513 MathEntry m = MathEntry();
1515 m.mathStart = start;
1517 m.mathSize = end - start;
1518 entries.push_back(m);
1520 bool empty() { return entries.empty(); };
1521 size_t getEndPos() {
1522 if (entries.empty() || (actualIdx >= entries.size())) {
1525 return entries[actualIdx].mathEnd;
1527 size_t getStartPos() {
1528 if (entries.empty() || (actualIdx >= entries.size())) {
1529 return 100000; /* definitely enough? */
1531 return entries[actualIdx].mathStart;
1533 size_t getFirstPos() {
1535 return getStartPos();
1538 if (entries.empty() || (actualIdx >= entries.size())) {
1541 return entries[actualIdx].mathSize;
1543 void incrEntry() { actualIdx++; };
1546 void LatexInfo::buildEntries(bool isPatternString)
1548 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1549 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1550 static bool disableLanguageOverride = false;
1551 smatch sub, submath;
1552 bool evaluatingRegexp = false;
1554 bool evaluatingMath = false;
1555 bool evaluatingCode = false;
1557 bool evaluatingOptional = false;
1558 size_t optionalEnd = 0;
1561 bool math_end_waiting = false;
1562 size_t math_pos = 10000;
1565 interval.removeAccents();
1567 for (sregex_iterator itmath(interval.par.begin(), interval.par.end(), rmath), end; itmath != end; ++itmath) {
1569 if (math_end_waiting) {
1570 size_t pos = submath.position(size_t(0));
1571 if ((math_end == "$") &&
1572 (submath.str(0) == "$") &&
1573 (interval.par[pos-1] != '\\')) {
1574 mi.insert("$", math_pos, pos + 1);
1575 math_end_waiting = false;
1577 else if ((math_end == "\\]") &&
1578 (submath.str(0) == "\\]")) {
1579 mi.insert("\\]", math_pos, pos + 2);
1580 math_end_waiting = false;
1582 else if ((submath.str(1).compare("end") == 0) &&
1583 (submath.str(2).compare(math_end) == 0)) {
1584 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1585 math_end_waiting = false;
1591 if (submath.str(1).compare("begin") == 0) {
1592 math_end_waiting = true;
1593 math_end = submath.str(2);
1594 math_pos = submath.position(size_t(0));
1596 else if (submath.str(0).compare("\\[") == 0) {
1597 math_end_waiting = true;
1599 math_pos = submath.position(size_t(0));
1601 else if (submath.str(0) == "$") {
1602 size_t pos = submath.position(size_t(0));
1603 if ((pos == 0) || (interval.par[pos-1] != '\\')) {
1604 math_end_waiting = true;
1611 // Ignore language if there is math somewhere in pattern-string
1612 if (isPatternString) {
1615 keys["foreignlanguage"].disabled = true;
1616 disableLanguageOverride = true;
1619 disableLanguageOverride = false;
1622 if (disableLanguageOverride) {
1623 keys["foreignlanguage"].disabled = true;
1626 math_pos = mi.getFirstPos();
1627 for (sregex_iterator it(interval.par.begin(), interval.par.end(), rkeys), end; it != end; ++it) {
1629 string key = sub.str(3);
1631 if (sub.str(0)[0] == '\\')
1632 key = sub.str(0)[1];
1636 size_t k_pos = sub.position(size_t(0));
1637 if ((k_pos > 0) && (interval.par[k_pos - 1] == '\\')) {
1638 // Escaped '$', ignoring
1644 if (evaluatingRegexp) {
1645 if (sub.str(1).compare("endregexp") == 0) {
1646 evaluatingRegexp = false;
1647 // found._tokenstart already set
1648 found._dataEnd = sub.position(size_t(0)) + 13;
1649 found._dataStart = found._dataEnd;
1650 found._tokensize = found._dataEnd - found._tokenstart;
1651 found.parenthesiscount = 0;
1652 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1659 if (evaluatingMath) {
1660 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1662 evaluatingMath = false;
1664 math_pos = mi.getStartPos();
1666 if (keys.find(key) == keys.end()) {
1667 found = KeyInfo(KeyInfo::isStandard, 0, true);
1668 if (isPatternString) {
1669 found.keytype = KeyInfo::isChar;
1670 found.disabled = false;
1677 if (key.compare("regexp") == 0) {
1678 evaluatingRegexp = true;
1679 found._tokenstart = sub.position(size_t(0));
1680 found._tokensize = 0;
1684 // Handle the other params of key
1685 if (found.keytype == KeyInfo::isIgnored)
1687 else if (found.keytype == KeyInfo::isMath) {
1688 if (size_t(sub.position(size_t(0))) == math_pos) {
1690 found._tokenstart = sub.position(size_t(0));
1691 found._tokensize = mi.getSize();
1692 found._dataEnd = found._tokenstart + found._tokensize;
1693 found._dataStart = found._dataEnd;
1694 found.parenthesiscount = 0;
1695 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1696 evaluatingMath = true;
1699 // begin|end of unknown env, discard
1700 // First handle tables
1701 // longtable|tabular
1702 bool discardComment;
1704 found.keytype = KeyInfo::doRemove;
1705 if ((sub.str(5).compare("longtable") == 0) ||
1706 (sub.str(5).compare("tabular") == 0)) {
1707 discardComment = true; /* '%' */
1710 discardComment = false;
1711 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1713 string token = sub.str(5);
1714 if (regex_match(token, sub2, removeArgs)) {
1715 found.keytype = KeyInfo::removeWithArg;
1718 // discard spaces before pos(0)
1719 int pos = sub.position(size_t(0));
1721 for (count = 0; pos - count > 0; count++) {
1722 char c = interval.par[pos-count-1];
1723 if (discardComment) {
1724 if ((c != ' ') && (c != '%'))
1730 found._tokenstart = pos - count;
1731 if (sub.str(1).compare(0, 5, "begin") == 0) {
1732 size_t pos1 = pos + sub.str(0).length();
1733 if (sub.str(5).compare("cjk") == 0) {
1734 pos1 = interval.findclosing(pos1+1, interval.par.length()) + 1;
1735 if ((interval.par[pos1] == '{') && (interval.par[pos1+1] == '}'))
1737 found.keytype = KeyInfo::isMain;
1738 found._dataStart = pos1;
1739 found._dataEnd = interval.par.length();
1740 found.disabled = keys["foreignlanguage"].disabled;
1741 found.used = keys["foreignlanguage"].used;
1742 found._tokensize = pos1 - found._tokenstart;
1743 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1746 // Swallow possible optional params
1747 while (interval.par[pos1] == '[') {
1748 pos1 = interval.findclosing(pos1+1, interval.par.length(), '[', ']')+1;
1750 // Swallow also the eventual parameter
1751 if (interval.par[pos1] == '{') {
1752 found._dataEnd = interval.findclosing(pos1+1, interval.par.length()) + 1;
1755 found._dataEnd = pos1;
1757 found._dataStart = found._dataEnd;
1758 found._tokensize = count + found._dataEnd - pos;
1759 found.parenthesiscount = 0;
1760 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1761 found.disabled = true;
1765 // Handle "\end{...}"
1766 found._dataStart = pos + sub.str(0).length();
1767 found._dataEnd = found._dataStart;
1768 found._tokensize = count + found._dataEnd - pos;
1769 found.parenthesiscount = 0;
1770 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1771 found.disabled = true;
1775 else if (found.keytype != KeyInfo::isRegex) {
1776 found._tokenstart = sub.position(size_t(0));
1777 if (found.parenthesiscount == 0) {
1778 // Probably to be discarded
1779 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1780 char following = interval.par[following_pos];
1781 if (following == ' ')
1782 found.head = "\\" + sub.str(3) + " ";
1783 else if (following == '=') {
1784 // like \uldepth=1000pt
1785 found.head = sub.str(0);
1788 found.head = "\\" + key;
1789 found._tokensize = found.head.length();
1790 found._dataEnd = found._tokenstart + found._tokensize;
1791 found._dataStart = found._dataEnd;
1794 int params = found._tokenstart + key.length() + 1;
1795 if (evaluatingOptional) {
1796 if (size_t(found._tokenstart) > optionalEnd) {
1797 evaluatingOptional = false;
1800 found.disabled = true;
1803 int optend = params;
1804 while (interval.par[optend] == '[') {
1805 // discard optional parameters
1806 optend = interval.findclosing(optend+1, interval.par.length(), '[', ']') + 1;
1808 if (optend > params) {
1809 key += interval.par.substr(params, optend-params);
1810 evaluatingOptional = true;
1811 optionalEnd = optend;
1813 string token = sub.str(5);
1814 int closings = found.parenthesiscount;
1815 if (found.parenthesiscount == 1) {
1816 found.head = "\\" + key + "{";
1818 else if (found.parenthesiscount > 1) {
1820 found.head = sub.str(0) + "{";
1821 closings = found.parenthesiscount - 1;
1824 found.head = "\\" + key + "{";
1827 found._tokensize = found.head.length();
1828 found._dataStart = found._tokenstart + found.head.length();
1829 if (interval.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1830 found._dataStart += 15;
1832 size_t endpos = interval.findclosing(found._dataStart, interval.par.length(), '{', '}', closings);
1833 if (found.keytype == KeyInfo::isList) {
1834 // Check if it really is list env
1835 static regex const listre("^([a-z]+)$");
1837 if (!regex_match(token, sub2, listre)) {
1838 // Change the key of this entry. It is not in a list/item environment
1839 found.keytype = KeyInfo::endArguments;
1842 if (found.keytype == KeyInfo::noMain) {
1843 evaluatingCode = true;
1845 codeStart = found._dataStart;
1847 else if (evaluatingCode) {
1848 if (size_t(found._dataStart) > codeEnd)
1849 evaluatingCode = false;
1850 else if (found.keytype == KeyInfo::isMain) {
1851 // Disable this key, treate it as standard
1852 found.keytype = KeyInfo::isStandard;
1853 found.disabled = true;
1854 if ((codeEnd == interval.par.length()) &&
1855 (found._tokenstart == codeStart)) {
1856 // trickery, because the code inset starts
1857 // with \selectlanguage ...
1859 if (entries.size() > 1) {
1860 entries[entries.size()-1]._dataEnd = codeEnd;
1865 if ((endpos == interval.par.length()) &&
1866 (found.keytype == KeyInfo::doRemove)) {
1867 // Missing closing => error in latex-input?
1868 // therefore do not delete remaining data
1869 found._dataStart -= 1;
1870 found._dataEnd = found._dataStart;
1873 found._dataEnd = endpos;
1875 if (isPatternString) {
1876 keys[key].used = true;
1879 entries.push_back(found);
1883 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1885 stringstream s(keysstring);
1887 const char delim = '|';
1888 while (getline(s, key, delim)) {
1889 KeyInfo keyII(keyI);
1890 if (isPatternString) {
1893 else if ( !keys[key].used)
1894 keyII.disabled = true;
1899 void LatexInfo::buildKeys(bool isPatternString)
1902 static bool keysBuilt = false;
1903 if (keysBuilt && !isPatternString) return;
1905 // Known standard keys with 1 parameter.
1906 // Split is done, if not at start of region
1907 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1908 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1909 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1910 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1911 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1912 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
1914 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
1915 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1916 makeKey("section*|subsection*|subsubsection*|paragraph*",
1917 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1918 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1919 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
1921 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
1923 // Split is done, if not at start of region
1924 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
1925 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
1927 // Split is done always.
1928 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
1932 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1933 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1934 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1935 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1937 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1938 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1939 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1941 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1942 // Custom space/skip, remove the content (== length value)
1943 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
1944 // Found in fr/UserGuide.lyx
1945 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1947 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1948 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1949 // Known macros to remove (including their parameter)
1951 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
1952 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
1953 // handle like standard keys with 1 parameter.
1954 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
1956 // Macros to remove, but let the parameter survive
1958 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1960 // Remove language spec from content of these insets
1961 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
1963 // Same effect as previous, parameter will survive (because there is no one anyway)
1965 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1966 // Remove table decorations
1967 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
1968 // Discard shape-header.
1969 // For footnote or shortcut too, because of lang settings
1970 // and wrong handling if used 'KeyInfo::noMain'
1971 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1972 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1973 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1974 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1975 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
1976 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
1977 // like ('tiny{}' or '\tiny ' ... )
1978 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
1980 // Survives, like known character
1981 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1982 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
1984 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1985 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1986 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
1988 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1989 // Remove RTL/LTR marker
1990 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1991 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
1992 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
1993 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
1994 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
1995 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
1996 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
1997 if (isPatternString) {
1998 // Allow the first searched string to rebuild the keys too
2002 // no need to rebuild again
2008 * Keep the list of actual opened parentheses actual
2009 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2011 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2014 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2018 if (c == '\\') skip = 1;
2019 else if (c == '{') {
2022 else if (c == '}') {
2023 handleCloseP(i, closingAllowed);
2029 string Intervall::show(int lastpos)
2031 int idx = 0; /* int intervalls */
2034 for (idx = 0; idx <= ignoreidx; idx++) {
2035 while (i < lastpos) {
2037 if (i <= borders[idx].low) {
2038 if (borders[idx].low > lastpos)
2039 printsize = lastpos - i;
2041 printsize = borders[idx].low - i;
2042 s += par.substr(i, printsize);
2044 if (i >= borders[idx].low)
2045 i = borders[idx].upper;
2048 i = borders[idx].upper;
2054 s += par.substr(i, lastpos-i);
2060 void Intervall::output(ostringstream &os, int lastpos)
2062 // get number of chars to output
2063 int idx = 0; /* int intervalls */
2066 string startTitle = titleValue;
2067 for (idx = 0; idx <= ignoreidx; idx++) {
2069 if (i <= borders[idx].low) {
2071 if (borders[idx].low > lastpos)
2072 printsize = lastpos - i;
2074 printsize = borders[idx].low - i;
2075 if (printsize > 0) {
2076 os << startTitle << par.substr(i, printsize);
2078 printed += printsize;
2081 handleParentheses(i, false);
2082 if (i >= borders[idx].low)
2083 i = borders[idx].upper;
2086 i = borders[idx].upper;
2093 os << startTitle << par.substr(i, lastpos-i);
2094 printed += lastpos-i;
2096 handleParentheses(lastpos, false);
2097 for (int i = actualdeptindex; i > 0; --i) {
2100 if (hasTitle && (printed > 0))
2102 if (! isPatternString)
2104 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2107 void LatexInfo::processRegion(int start, int region_end)
2109 while (start < region_end) { /* Let {[} and {]} survive */
2110 int cnt = interval.isOpeningPar(start);
2112 // Closing is allowed past the region
2113 int closing = interval.findclosing(start+1, interval.par.length());
2114 interval.addIntervall(start, start+1);
2115 interval.addIntervall(closing, closing+1);
2119 start = interval.nextNotIgnored(start+1);
2123 void LatexInfo::removeHead(KeyInfo &actual, int count)
2125 if (actual.parenthesiscount == 0) {
2126 // "{\tiny{} ...}" ==> "{{} ...}"
2127 interval.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2130 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2131 interval.addIntervall(actual._tokenstart - count, actual._dataStart);
2132 interval.addIntervall(actual._dataEnd, actual._dataEnd+1);
2136 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2139 switch (actual.keytype)
2141 case KeyInfo::isTitle: {
2143 nextKeyIdx = getNextKey();
2146 case KeyInfo::cleanToStart: {
2147 actual._dataEnd = actual._dataStart;
2148 nextKeyIdx = getNextKey();
2149 // Search for end of arguments
2150 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2152 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2153 entries[i].disabled = true;
2155 actual._dataEnd = entries[tmpIdx]._dataEnd;
2157 while (interval.par[actual._dataEnd] == ' ')
2159 interval.addIntervall(0, actual._dataEnd+1);
2160 interval.actualdeptindex = 0;
2161 interval.depts[0] = actual._dataEnd+1;
2162 interval.closes[0] = -1;
2165 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2166 if (actual.disabled)
2167 interval.addIntervall(actual._tokenstart, actual._dataEnd);
2169 interval.addIntervall(actual._dataStart, actual._dataEnd);
2172 case KeyInfo::isChar: {
2173 nextKeyIdx = getNextKey();
2176 case KeyInfo::isSize: {
2177 if (actual.disabled || (interval.par[actual._dataStart] != '{') || (interval.par[actual._dataStart-1] == ' ')) {
2178 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2179 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2180 nextKeyIdx = getNextKey();
2182 // Here _dataStart points to '{', so correct it
2183 actual._dataStart += 1;
2184 actual._tokensize += 1;
2185 actual.parenthesiscount = 1;
2186 if (interval.par[actual._dataStart] == '}') {
2187 // Determine the end if used like '{\tiny{}...}'
2188 actual._dataEnd = interval.findclosing(actual._dataStart+1, interval.par.length()) + 1;
2189 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2192 // Determine the end if used like '\tiny{...}'
2193 actual._dataEnd = interval.findclosing(actual._dataStart, interval.par.length()) + 1;
2195 // Split on this key if not at start
2196 int start = interval.nextNotIgnored(previousStart);
2197 if (start < actual._tokenstart) {
2198 interval.output(os, actual._tokenstart);
2199 interval.addIntervall(start, actual._tokenstart);
2201 // discard entry if at end of actual
2202 nextKeyIdx = process(os, actual);
2206 case KeyInfo::endArguments:
2207 // Remove trailing '{}' too
2208 actual._dataStart += 1;
2209 actual._dataEnd += 1;
2210 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2211 nextKeyIdx = getNextKey();
2213 case KeyInfo::noMain:
2215 case KeyInfo::isStandard: {
2216 if (actual.disabled) {
2218 processRegion(actual._dataStart, actual._dataStart+1);
2219 nextKeyIdx = getNextKey();
2221 // Split on this key if not at datastart of calling entry
2222 int start = interval.nextNotIgnored(previousStart);
2223 if (start < actual._tokenstart) {
2224 interval.output(os, actual._tokenstart);
2225 interval.addIntervall(start, actual._tokenstart);
2227 // discard entry if at end of actual
2228 nextKeyIdx = process(os, actual);
2232 case KeyInfo::removeWithArg: {
2233 nextKeyIdx = getNextKey();
2234 // Search for end of arguments
2235 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2237 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2238 entries[i].disabled = true;
2240 actual._dataEnd = entries[tmpIdx]._dataEnd;
2242 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2245 case KeyInfo::doRemove: {
2246 // Remove the key with all parameters and following spaces
2248 for (pos = actual._dataEnd+1; pos < interval.par.length(); pos++) {
2249 if ((interval.par[pos] != ' ') && (interval.par[pos] != '%'))
2252 interval.addIntervall(actual._tokenstart, pos);
2253 nextKeyIdx = getNextKey();
2256 case KeyInfo::isList: {
2257 // Discard space before _tokenstart
2259 for (count = 0; count < actual._tokenstart; count++) {
2260 if (interval.par[actual._tokenstart-count-1] != ' ')
2263 nextKeyIdx = getNextKey();
2264 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2266 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2268 // How else can we catch this one?
2269 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2270 entries[i].disabled = true;
2272 actual._dataEnd = entries[tmpIdx]._dataEnd;
2274 else if (nextKeyIdx > 0) {
2275 // Ignore any lang entries inside data region
2276 for (int i = nextKeyIdx; i < int(entries.size()) && entries[i]._tokenstart < actual._dataEnd; i++) {
2277 if (entries[i].keytype == KeyInfo::isMain)
2278 entries[i].disabled = true;
2281 if (actual.disabled) {
2282 interval.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2285 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2287 if (interval.par[actual._dataEnd+1] == '[') {
2288 int posdown = interval.findclosing(actual._dataEnd+2, interval.par.length(), '[', ']');
2289 if ((interval.par[actual._dataEnd+2] == '{') &&
2290 (interval.par[posdown-1] == '}')) {
2291 interval.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2292 interval.addIntervall(posdown-1, posdown+1);
2295 interval.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2296 interval.addIntervall(posdown, posdown+1);
2298 int blk = interval.nextNotIgnored(actual._dataEnd+1);
2299 if (blk > posdown) {
2300 // Discard at most 1 space after empty item
2302 for (count = 0; count < 1; count++) {
2303 if (interval.par[blk+count] != ' ')
2307 interval.addIntervall(blk, blk+count);
2312 case KeyInfo::isSectioning: {
2313 // Discard spaces before _tokenstart
2315 int val = actual._tokenstart;
2316 for (count = 0; count < actual._tokenstart;) {
2317 val = interval.previousNotIgnored(val-1);
2318 if (interval.par[val] != ' ')
2321 count = actual._tokenstart - val;
2324 if (actual.disabled) {
2325 removeHead(actual, count);
2326 nextKeyIdx = getNextKey();
2328 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2329 nextKeyIdx = process(os, actual);
2333 case KeyInfo::isMath: {
2334 // Same as regex, use the content unchanged
2335 nextKeyIdx = getNextKey();
2338 case KeyInfo::isRegex: {
2339 // DO NOT SPLIT ON REGEX
2341 nextKeyIdx = getNextKey();
2344 case KeyInfo::isIgnored: {
2345 // Treat like a character for now
2346 nextKeyIdx = getNextKey();
2349 case KeyInfo::isMain: {
2350 if (interval.par.substr(actual._dataStart, 2) == "% ")
2351 interval.addIntervall(actual._dataStart, actual._dataStart+2);
2352 if (actual._tokenstart > 0) {
2353 int prev = interval.previousNotIgnored(actual._tokenstart - 1);
2354 if ((prev >= 0) && interval.par[prev] == '%')
2355 interval.addIntervall(prev, prev+1);
2357 if (actual.disabled) {
2359 if ((interval.par.substr(actual._dataStart, 3) == " \\[") ||
2360 (interval.par.substr(actual._dataStart, 8) == " \\begin{")) {
2361 // Discard also the space before math-equation
2362 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2364 nextKeyIdx = getNextKey();
2365 // interval.resetOpenedP(actual._dataStart-1);
2368 if (actual._tokenstart < 26) {
2369 // for the first (and maybe dummy) language
2370 interval.setForDefaultLang(actual);
2372 interval.resetOpenedP(actual._dataStart-1);
2376 case KeyInfo::invalid:
2377 // This cannot happen, already handled
2380 // LYXERR0("Unhandled keytype");
2381 nextKeyIdx = getNextKey();
2388 int LatexInfo::process(ostringstream &os, KeyInfo &actual )
2390 int end = interval.nextNotIgnored(actual._dataEnd);
2391 int oldStart = actual._dataStart;
2392 int nextKeyIdx = getNextKey();
2394 if ((nextKeyIdx < 0) ||
2395 (entries[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2396 (entries[nextKeyIdx].keytype == KeyInfo::invalid)) {
2397 if (oldStart <= end) {
2398 processRegion(oldStart, end);
2403 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2405 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2406 (void) dispatch(os, actual._dataStart, nextKey);
2407 end = nextKey._tokenstart;
2410 processRegion(oldStart, nextKey._tokenstart);
2411 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2413 oldStart = nextKey._dataEnd+1;
2415 // now nextKey is either invalid or is outside of actual._dataEnd
2416 // output the remaining and discard myself
2417 if (oldStart <= end) {
2418 processRegion(oldStart, end);
2420 if (interval.par[end] == '}') {
2422 // This is the normal case.
2423 // But if using the firstlanguage, the closing may be missing
2425 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2427 if (actual._dataEnd < end)
2428 output_end = interval.nextNotIgnored(actual._dataEnd);
2430 output_end = interval.nextNotIgnored(end);
2431 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2432 interval.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2434 // Remove possible empty data
2435 int dstart = interval.nextNotIgnored(actual._dataStart);
2436 while (interval.isOpeningPar(dstart) == 1) {
2437 interval.addIntervall(dstart, dstart+1);
2438 int dend = interval.findclosing(dstart+1, output_end);
2439 interval.addIntervall(dend, dend+1);
2440 dstart = interval.nextNotIgnored(dstart+1);
2442 if (dstart < output_end)
2443 interval.output(os, output_end);
2444 interval.addIntervall(actual._tokenstart, end);
2448 string splitOnKnownMacros(string par, bool isPatternString)
2451 LatexInfo li(par, isPatternString);
2452 // LYXERR0("Berfore split: " << par);
2453 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2455 DummyKey._tokensize = 0;
2456 DummyKey._dataStart = 0;
2457 DummyKey._dataEnd = par.length();
2458 DummyKey.disabled = true;
2459 int firstkeyIdx = li.getFirstKey();
2461 if (firstkeyIdx >= 0) {
2462 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2463 DummyKey._tokenstart = firstKey._tokenstart;
2465 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2466 // Use dummy firstKey
2467 firstKey = DummyKey;
2468 (void) li.setNextKey(firstkeyIdx);
2471 if (par.substr(firstKey._dataStart, 2) == "% ")
2472 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2474 nextkeyIdx = li.process(os, firstKey);
2475 while (nextkeyIdx >= 0) {
2476 // Check for a possible gap between the last
2477 // entry and this one
2478 int datastart = li.nextNotIgnored(firstKey._dataStart);
2479 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2480 if ((nextKey._tokenstart > datastart)) {
2482 firstKey._dataStart = datastart;
2483 firstKey._dataEnd = par.length();
2484 (void) li.setNextKey(nextkeyIdx);
2485 // Fake the last opened parenthesis
2486 li.setForDefaultLang(firstKey);
2487 nextkeyIdx = li.process(os, firstKey);
2490 if (nextKey.keytype != KeyInfo::isMain) {
2491 firstKey._dataStart = datastart;
2492 firstKey._dataEnd = nextKey._dataEnd+1;
2493 (void) li.setNextKey(nextkeyIdx);
2494 li.setForDefaultLang(firstKey);
2495 nextkeyIdx = li.process(os, firstKey);
2498 nextkeyIdx = li.process(os, nextKey);
2502 // Handle the remaining
2503 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2504 firstKey._dataEnd = par.length();
2506 if ((firstKey._dataStart < firstKey._dataEnd) &&
2507 (par[firstKey._dataStart] != '}')) {
2508 li.setForDefaultLang(firstKey);
2509 (void) li.process(os, firstKey);
2513 // return string definitelly impossible to match
2514 s = "\\foreignlanguage{ignore}{ }";
2518 s = par; /* no known macros found */
2519 // LYXERR0("After split: " << s);
2524 * Try to unify the language specs in the latexified text.
2525 * Resulting modified string is set to "", if
2526 * the searched tex does not contain all the features in the search pattern
2528 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2530 static Features regex_f;
2531 static int missed = 0;
2532 static bool regex_with_format = false;
2534 int parlen = par.length();
2536 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2539 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2540 // Happens to be there in case of description or labeling environment
2545 // Split the latex input into pieces which
2546 // can be digested by our search engine
2547 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2548 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2549 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2552 result = par.substr(0, parlen);
2553 if (isPatternString) {
2556 regex_f = identifyFeatures(result);
2557 string features = "";
2558 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2559 string a = it->first;
2560 regex_with_format = true;
2561 features += " " + a;
2562 // LYXERR0("Identified regex format:" << a);
2564 LYXERR(Debug::FIND, "Identified Features" << features);
2567 } else if (regex_with_format) {
2568 Features info = identifyFeatures(result);
2569 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2570 string a = it->first;
2571 bool b = it->second;
2572 if (b && ! info[a]) {
2574 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2580 // LYXERR0("No regex formats");
2586 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2587 static int identifyClosing(string & t)
2589 int open_braces = 0;
2591 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2592 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2594 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]" REGEX_EOS, "$1"))
2596 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2598 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2608 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2609 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2611 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2612 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2613 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2614 // When using regexp, braces are hacked already by escape_for_regex()
2615 par_as_string = normalize(ds, !use_regexp);
2617 close_wildcards = 0;
2619 size_t lead_size = 0;
2620 // correct the language settings
2621 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2622 if (opt.ignoreformat) {
2624 // if par_as_string_nolead were emty,
2625 // the following call to findAux will always *find* the string
2626 // in the checked data, and thus always using the slow
2627 // examining of the current text part.
2628 par_as_string_nolead = par_as_string;
2631 lead_size = identifyLeading(par_as_string);
2632 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2633 lead_as_string = par_as_string.substr(0, lead_size);
2634 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2638 open_braces = identifyClosing(par_as_string);
2639 identifyClosing(par_as_string_nolead);
2640 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2641 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2643 string lead_as_regexp;
2644 if (lead_size > 0) {
2645 // @todo No need to search for \regexp{} insets in leading material
2646 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2647 par_as_string = par_as_string_nolead;
2648 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2649 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2651 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2652 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2653 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2655 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2656 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2657 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2658 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2659 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2660 || regex_replace(par_as_string, par_as_string,
2661 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2662 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2663 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2667 if (!opt.ignoreformat) {
2668 // Remove extra '\}' at end if not part of \{\.\}
2669 size_t lng = par_as_string.size();
2671 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2673 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2682 if (lng < par_as_string.size())
2683 par_as_string = par_as_string.substr(0,lng);
2686 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2687 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2688 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2689 // replace '[^...]' with '[^...\}\{\\]'
2690 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2691 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2692 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2694 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2697 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2698 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2699 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2700 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2702 // If entered regexp must match at begin of searched string buffer
2703 // Kornel: Added parentheses to use $1 for size of the leading string
2707 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2708 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2709 // so the convert has no effect in that case
2710 for (int i = 8; i > 0; --i) {
2711 string orig = "\\\\" + std::to_string(i);
2712 string dest = "\\" + std::to_string(i+1);
2713 while (regex_replace(par_as_string, par_as_string, orig, dest));
2715 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2716 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2718 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2719 regexp = lyx::regex(regexp_str);
2721 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2722 regexp2 = lyx::regex(regexp2_str);
2727 // Count number of characters in string
2731 // \\[a-zA-Z]+ ==> 1
2732 static int computeSize(string s, int len)
2738 for (int i = 0; i < len; i += skip, count++) {
2741 if (isalpha(s[i+1])) {
2742 for (int j = 2; i+j < len; j++) {
2743 if (! isalpha(s[i+j])) {
2746 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2748 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2756 else if (s[i] == '{') {
2762 else if (s[i] == '-') {
2763 if (s[i+1] == '-') {
2779 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2784 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2787 docstring docstr = stringifyFromForSearch(opt, cur, len);
2788 string str = normalize(docstr, true);
2789 if (!opt.ignoreformat) {
2790 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2793 mres.match_len = -1;
2796 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2797 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2800 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2801 regex const *p_regexp;
2802 regex_constants::match_flag_type flags;
2804 flags = regex_constants::match_continuous;
2807 flags = regex_constants::match_default;
2808 p_regexp = ®exp2;
2810 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2811 if (re_it == sregex_iterator())
2813 match_results<string::const_iterator> const & m = *re_it;
2815 if (0) { // Kornel Benko: DO NOT CHECKK
2816 // Check braces on the segment that matched the entire regexp expression,
2817 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2818 if (!braces_match(m[0].first, m[0].second, open_braces))
2822 // Check braces on segments that matched all (.*?) subexpressions,
2823 // except the last "padding" one inserted by lyx.
2824 for (size_t i = 1; i < m.size() - 1; ++i)
2825 if (!braces_match(m[i].first, m[i].second, open_braces))
2828 // Exclude from the returned match length any length
2829 // due to close wildcards added at end of regexp
2830 // and also the length of the leading (e.g. '\emph{')
2832 // Whole found string, including the leading: m[0].second - m[0].first
2833 // Size of the leading string: m[1].second - m[1].first
2834 int leadingsize = 0;
2836 leadingsize = m[1].second - m[1].first;
2838 for (size_t i = 0; i < m.size(); i++) {
2839 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2841 if (close_wildcards == 0)
2842 result = m[0].second - m[0].first;
2845 result = m[m.size() - close_wildcards].first - m[0].first;
2847 size_t pos = m.position(size_t(0));
2848 // Ignore last closing characters
2849 while (result > 0) {
2850 if (str[pos+result-1] == '}')
2855 if (result > leadingsize)
2856 result -= leadingsize;
2859 mres.match_len = computeSize(str.substr(pos+leadingsize,result), result);
2860 mres.match2end = str.size() - pos - leadingsize;
2861 mres.pos = pos+leadingsize;
2865 // else !use_regexp: but all code paths above return
2866 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2867 << par_as_string << "', str='" << str << "'");
2868 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2869 << lead_as_string << "', par_as_string_nolead='"
2870 << par_as_string_nolead << "'");
2873 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2874 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2875 if (str.substr(0, par_as_string.size()) == par_as_string) {
2876 mres.match_len = par_as_string.size();
2877 mres.match2end = str.size();
2882 size_t pos = str.find(par_as_string_nolead);
2883 if (pos != string::npos) {
2884 mres.match_len = par_as_string.size();
2885 mres.match2end = str.size() - pos;
2894 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
2896 MatchResult mres = findAux(cur, len, at_begin);
2897 int res = mres.match_len;
2899 "res=" << res << ", at_begin=" << at_begin
2900 << ", matchword=" << opt.matchword
2901 << ", inTexted=" << cur.inTexted());
2902 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
2904 if ((len > 0) && (res < len)) {
2908 Paragraph const & par = cur.paragraph();
2909 bool ws_left = (cur.pos() > 0)
2910 ? par.isWordSeparator(cur.pos() - 1)
2912 bool ws_right = (cur.pos() + len < par.size())
2913 ? par.isWordSeparator(cur.pos() + len)
2916 "cur.pos()=" << cur.pos() << ", res=" << res
2917 << ", separ: " << ws_left << ", " << ws_right
2920 if (ws_left && ws_right) {
2921 // Check for word separators inside the found 'word'
2922 for (int i = 0; i < len; i++) {
2923 if (par.isWordSeparator(cur.pos() + i)) {
2935 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
2938 if (! opt.casesensitive)
2939 t = lyx::to_utf8(lowercase(s));
2941 t = lyx::to_utf8(s);
2942 // Remove \n at begin
2943 while (!t.empty() && t[0] == '\n')
2946 while (!t.empty() && t[t.size() - 1] == '\n')
2947 t = t.substr(0, t.size() - 1);
2949 // Handle all other '\n'
2950 while ((pos = t.find("\n")) != string::npos) {
2951 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
2953 if (isAlnumASCII(t[pos+1])) {
2954 t.replace(pos-2, 3, " ");
2957 t.replace(pos-2, 3, "");
2960 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
2961 // '\n' adjacent to non-alpha-numerics, discard
2962 t.replace(pos, 1, "");
2965 // Replace all other \n with spaces
2966 t.replace(pos, 1, " ");
2969 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
2970 // Kornel: Added textsl, textsf, textit, texttt and noun
2971 // + allow to seach for colored text too
2972 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
2973 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
2974 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2975 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
2976 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
2978 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
2979 // FIXME - check what preceeds the brace
2981 if (opt.ignoreformat)
2982 while (regex_replace(t, t, "\\{", "_x_<")
2983 || regex_replace(t, t, "\\}", "_x_>"))
2984 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2986 while (regex_replace(t, t, "\\\\\\{", "_x_<")
2987 || regex_replace(t, t, "\\\\\\}", "_x_>"))
2988 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
2995 docstring stringifyFromCursor(DocIterator const & cur, int len)
2997 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
2998 if (cur.inTexted()) {
2999 Paragraph const & par = cur.paragraph();
3000 // TODO what about searching beyond/across paragraph breaks ?
3001 // TODO Try adding a AS_STR_INSERTS as last arg
3002 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3003 int(par.size()) : cur.pos() + len;
3004 OutputParams runparams(&cur.buffer()->params().encoding());
3005 runparams.nice = true;
3006 runparams.flavor = OutputParams::XETEX;
3007 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3008 // No side effect of file copying and image conversion
3009 runparams.dryrun = true;
3010 runparams.for_search = true;
3011 LYXERR(Debug::FIND, "Stringifying with cur: "
3012 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3013 return par.asString(cur.pos(), end,
3014 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3016 } else if (cur.inMathed()) {
3017 CursorSlice cs = cur.top();
3018 MathData md = cs.cell();
3019 MathData::const_iterator it_end =
3020 (( len == -1 || cs.pos() + len > int(md.size()))
3022 : md.begin() + cs.pos() + len );
3024 for (MathData::const_iterator it = md.begin() + cs.pos();
3027 docstring s = asString(md2);
3028 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3031 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3036 /** Computes the LaTeX export of buf starting from cur and ending len positions
3037 * after cur, if len is positive, or at the paragraph or innermost inset end
3040 docstring latexifyFromCursor(DocIterator const & cur, int len)
3042 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3043 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3044 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3045 Buffer const & buf = *cur.buffer();
3047 odocstringstream ods;
3049 OutputParams runparams(&buf.params().encoding());
3050 runparams.nice = false;
3051 runparams.flavor = OutputParams::XETEX;
3052 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3053 // No side effect of file copying and image conversion
3054 runparams.dryrun = true;
3055 runparams.for_search = true;
3057 if (cur.inTexted()) {
3058 // @TODO what about searching beyond/across paragraph breaks ?
3059 pos_type endpos = cur.paragraph().size();
3060 if (len != -1 && endpos > cur.pos() + len)
3061 endpos = cur.pos() + len;
3062 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3063 string(), cur.pos(), endpos);
3064 string s = lyx::to_utf8(ods.str());
3065 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3066 return(lyx::from_utf8(s));
3067 } else if (cur.inMathed()) {
3068 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3069 for (int s = cur.depth() - 1; s >= 0; --s) {
3070 CursorSlice const & cs = cur[s];
3071 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3073 cs.asInsetMath()->asHullInset()->header_write(ws);
3078 CursorSlice const & cs = cur.top();
3079 MathData md = cs.cell();
3080 MathData::const_iterator it_end =
3081 ((len == -1 || cs.pos() + len > int(md.size()))
3083 : md.begin() + cs.pos() + len);
3085 for (MathData::const_iterator it = md.begin() + cs.pos();
3089 ods << asString(md2);
3090 // Retrieve the math environment type, and add '$' or '$]'
3091 // or others (\end{equation}) accordingly
3092 for (int s = cur.depth() - 1; s >= 0; --s) {
3093 CursorSlice const & cs2 = cur[s];
3094 InsetMath * inset = cs2.asInsetMath();
3095 if (inset && inset->asHullInset()) {
3097 inset->asHullInset()->footer_write(ws);
3101 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3103 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3109 /** Finalize an advanced find operation, advancing the cursor to the innermost
3110 ** position that matches, plus computing the length of the matching text to
3113 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
3115 // Search the foremost position that matches (avoids find of entire math
3116 // inset when match at start of it)
3118 DocIterator old_cur(cur.buffer());
3120 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3124 } while (cur && cur.depth() > d && match(cur).match_len > 0);
3126 int max_match = match(cur).match_len; /* match valid only if not searching whole words */
3127 if (max_match <= 0) return 0;
3128 LYXERR(Debug::FIND, "Ok");
3130 // Compute the match length
3132 if (cur.pos() + len > cur.lastpos())
3134 if (match.opt.matchword) {
3135 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3136 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3138 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3140 // Length of matched text (different from len param)
3141 int old_match = match(cur, len).match_len;
3145 // Greedy behaviour while matching regexps
3146 while ((new_match = match(cur, len + 1).match_len) > old_match) {
3148 old_match = new_match;
3149 LYXERR(Debug::FIND, "verifying match with len = " << len);
3156 int maxl = cur.lastpos() - cur.pos();
3157 // Greedy behaviour while matching regexps
3158 while (maxl > minl) {
3159 int actual_match = match(cur, len).match_len;
3160 if (actual_match >= max_match) {
3161 // actual_match > max_match _can_ happen,
3162 // if the search area splits
3163 // some following word so that the regex
3164 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3166 // This means, the len value is too big
3168 len = (int)((maxl + minl)/2);
3171 // (actual_match < max_match)
3173 len = (int)((maxl + minl)/2);
3177 // Search for real start of matched characters
3182 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3183 if (cur.depth() < old_cur.depth()) {
3185 LYXERR0("cur.depth() < old_cur.depth(), this should never happen");
3188 if (cur.pos() != old_cur.pos()) {
3189 // OK, forwarded 1 pos in actual inset
3190 actual_match = match(cur, len-1).match_len;
3191 if (actual_match == max_match) {
3192 // Ha, got it! The shorter selection has the same match length
3197 // OK, the shorter selection matches less chars, revert to previous value
3203 LYXERR0("cur.pos() == old_cur.pos(), this should never happen");
3204 actual_match = match(cur, len).match_len;
3205 if (actual_match == max_match)
3215 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3219 while (!theApp()->longOperationCancelled() && cur) {
3220 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3221 MatchResult mres = match(cur, -1, false);
3222 int match_len = mres.match_len;
3223 LYXERR(Debug::FIND, "match_len: " << match_len);
3224 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3225 LYXERR0("BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3228 if (match_len > 0) {
3229 // Try to find the begin of searched string
3230 int increment = mres.pos/2;
3231 while (mres.pos > 5 && (increment > 5)) {
3232 DocIterator old_cur = cur;
3233 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3240 MatchResult mres2 = match(cur, -1, false);
3241 if ((mres2.match2end < mres.match2end) ||
3242 (mres2.match_len < mres.match_len)) {
3249 if (increment > mres.pos/2)
3250 increment = mres.pos/2;
3254 int match_len_zero_count = 0;
3255 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3257 int remaining_len = match(cur, -1, false).match_len;
3258 if (remaining_len <= 0) {
3259 // Apparently the searched string is not in the remaining part
3266 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3267 int match_len3 = match(cur, 1).match_len;
3270 int match_len2 = match(cur).match_len;
3271 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3272 if (match_len2 > 0) {
3273 // Sometimes in finalize we understand it wasn't a match
3274 // and we need to continue the outest loop
3275 int len = findAdvFinalize(cur, match);
3280 if (match_len2 >= 0) {
3281 if (match_len2 == 0)
3282 match_len_zero_count++;
3284 match_len_zero_count = 0;
3287 if (++match_len_zero_count > 3) {
3288 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3289 match_len_zero_count = 0;
3297 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3298 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3301 // This should exit nested insets, if any, or otherwise undefine the currsor.
3302 cur.pos() = cur.lastpos();
3303 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3311 /// Find the most backward consecutive match within same paragraph while searching backwards.
3312 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3314 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3315 DocIterator tmp_cur = cur;
3316 int len = findAdvFinalize(tmp_cur, match);
3317 Inset & inset = cur.inset();
3318 for (; cur != cur_begin; cur.backwardPos()) {
3319 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3320 DocIterator new_cur = cur;
3321 new_cur.backwardPos();
3322 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3324 int new_len = findAdvFinalize(new_cur, match);
3329 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3335 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3339 // Backup of original position
3340 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3341 if (cur == cur_begin)
3344 DocIterator cur_orig(cur);
3345 bool pit_changed = false;
3348 bool found_match = (match(cur, -1, false).match_len > 0);
3352 cur.pos() = cur.lastpos();
3354 cur.pos() = cur_orig.pos();
3355 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3356 DocIterator cur_prev_iter;
3358 found_match = (match(cur).match_len > 0);
3359 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3360 << found_match << ", cur: " << cur);
3362 return findMostBackwards(cur, match);
3364 // Stop if begin of document reached
3365 if (cur == cur_begin)
3367 cur_prev_iter = cur;
3371 if (cur == cur_begin)
3378 } while (!theApp()->longOperationCancelled());
3386 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3387 DocIterator const & cur, int len)
3389 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3391 if (!opt.ignoreformat)
3392 return latexifyFromCursor(cur, len);
3394 return stringifyFromCursor(cur, len);
3398 FindAndReplaceOptions::FindAndReplaceOptions(
3399 docstring const & find_buf_name, bool casesensitive,
3400 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
3401 docstring const & repl_buf_name, bool keep_case,
3402 SearchScope scope, SearchRestriction restr)
3403 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
3404 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
3405 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope), restr(restr)
3413 /** Check if 'len' letters following cursor are all non-lowercase */
3414 static bool allNonLowercase(Cursor const & cur, int len)
3416 pos_type beg_pos = cur.selectionBegin().pos();
3417 pos_type end_pos = cur.selectionBegin().pos() + len;
3418 if (len > cur.lastpos() + 1 - beg_pos) {
3419 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3420 len = cur.lastpos() + 1 - beg_pos;
3421 end_pos = beg_pos + len;
3423 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3424 if (isLowerCase(cur.paragraph().getChar(pos)))
3430 /** Check if first letter is upper case and second one is lower case */
3431 static bool firstUppercase(Cursor const & cur)
3434 pos_type pos = cur.selectionBegin().pos();
3435 if (pos >= cur.lastpos() - 1) {
3436 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3439 ch1 = cur.paragraph().getChar(pos);
3440 ch2 = cur.paragraph().getChar(pos + 1);
3441 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3442 LYXERR(Debug::FIND, "firstUppercase(): "
3443 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3444 << ch2 << "(" << char(ch2) << ")"
3445 << ", result=" << result << ", cur=" << cur);
3450 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3452 ** \fixme What to do with possible further paragraphs in replace buffer ?
3454 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3456 ParagraphList::iterator pit = buffer.paragraphs().begin();
3457 LASSERT(pit->size() >= 1, /**/);
3458 pos_type right = pos_type(1);
3459 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3460 right = pit->size();
3461 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3467 static void findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3469 Cursor & cur = bv->cursor();
3470 if (opt.repl_buf_name == docstring()
3471 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3472 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3475 DocIterator sel_beg = cur.selectionBegin();
3476 DocIterator sel_end = cur.selectionEnd();
3477 if (&sel_beg.inset() != &sel_end.inset()
3478 || sel_beg.pit() != sel_end.pit()
3479 || sel_beg.idx() != sel_end.idx())
3481 int sel_len = sel_end.pos() - sel_beg.pos();
3482 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3483 << ", sel_len: " << sel_len << endl);
3486 LASSERT(sel_len > 0, return);
3488 if (!matchAdv(sel_beg, sel_len).match_len)
3491 // Build a copy of the replace buffer, adapted to the KeepCase option
3492 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3494 repl_buffer_orig.write(oss);
3495 string lyx = oss.str();
3496 Buffer repl_buffer("", false);
3497 repl_buffer.setUnnamed(true);
3498 LASSERT(repl_buffer.readString(lyx), return);
3499 if (opt.keep_case && sel_len >= 2) {
3500 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3501 if (cur.inTexted()) {
3502 if (firstUppercase(cur))
3503 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3504 else if (allNonLowercase(cur, sel_len))
3505 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3508 cap::cutSelection(cur, false);
3509 if (cur.inTexted()) {
3510 repl_buffer.changeLanguage(
3511 repl_buffer.language(),
3512 cur.getFont().language());
3513 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3514 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3515 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3516 repl_buffer.params().documentClassPtr(),
3517 bv->buffer().errorList("Paste"));
3518 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3519 sel_len = repl_buffer.paragraphs().begin()->size();
3520 } else if (cur.inMathed()) {
3521 odocstringstream ods;
3523 OutputParams runparams(&repl_buffer.params().encoding());
3524 runparams.nice = false;
3525 runparams.flavor = OutputParams::XETEX;
3526 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3527 runparams.dryrun = true;
3528 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3529 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3530 docstring repl_latex = ods.str();
3531 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3533 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3534 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3535 repl_latex = from_utf8(s);
3536 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3537 MathData ar(cur.buffer());
3538 asArray(repl_latex, ar, Parse::NORMAL);
3540 sel_len = ar.size();
3541 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3543 if (cur.pos() >= sel_len)
3544 cur.pos() -= sel_len;
3547 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3548 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3549 bv->processUpdateFlags(Update::Force);
3553 /// Perform a FindAdv operation.
3554 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3559 // e.g., when invoking word-findadv from mini-buffer wither with
3560 // wrong options syntax or before ever opening advanced F&R pane
3561 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3565 MatchStringAdv matchAdv(bv->buffer(), opt);
3566 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3568 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3569 findAdvReplace(bv, opt, matchAdv);
3572 match_len = findForwardAdv(cur, matchAdv);
3574 match_len = findBackwardsAdv(cur, matchAdv);
3576 // This may only be raised by lyx::regex()
3577 bv->message(_("Invalid regular expression!"));
3581 if (match_len == 0) {
3582 bv->message(_("Match not found!"));
3586 bv->message(_("Match found!"));
3588 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3589 bv->putSelectionAt(cur, match_len, !opt.forward);
3595 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3597 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3598 << opt.casesensitive << ' '
3599 << opt.matchword << ' '
3600 << opt.forward << ' '
3601 << opt.expandmacros << ' '
3602 << opt.ignoreformat << ' '
3603 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3604 << opt.keep_case << ' '
3605 << int(opt.scope) << ' '
3608 LYXERR(Debug::FIND, "built: " << os.str());
3614 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3616 LYXERR(Debug::FIND, "parsing");
3620 while (line != "EOSS") {
3624 if (is.eof()) // Tolerate malformed request
3628 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3629 opt.find_buf_name = from_utf8(s);
3630 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat;
3631 is.get(); // Waste space before replace string
3634 while (line != "EOSS") {
3638 if (is.eof()) // Tolerate malformed request
3642 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3643 opt.repl_buf_name = from_utf8(s);
3644 is >> opt.keep_case;
3647 opt.scope = FindAndReplaceOptions::SearchScope(i);
3649 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3651 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3652 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3653 << opt.scope << ' ' << opt.restr);