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"
37 #include "frontends/Application.h"
38 #include "frontends/alert.h"
40 #include "mathed/InsetMath.h"
41 #include "mathed/InsetMathGrid.h"
42 #include "mathed/InsetMathHull.h"
43 #include "mathed/MathData.h"
44 #include "mathed/MathStream.h"
45 #include "mathed/MathSupport.h"
47 #include "support/convert.h"
48 #include "support/debug.h"
49 #include "support/docstream.h"
50 #include "support/FileName.h"
51 #include "support/gettext.h"
52 #include "support/lassert.h"
53 #include "support/lstrings.h"
55 #include "support/regex.h"
56 #include "support/textutils.h"
60 using namespace lyx::support;
65 // Helper class for deciding what should be ignored
70 : ignoreFamily_(false), ignoreSeries_(false),
71 ignoreShape_(false), ignoreUnderline_(false),
72 ignoreMarkUp_(false), ignoreStrikeOut_(false),
73 ignoreSectioning_(false), ignoreFrontMatter_(false),
74 ignoreColor_(false), ignoreLanguage_(false) {}
76 bool getFamily() { return ignoreFamily_; };
78 bool getSeries() { return ignoreSeries_; };
80 bool getShape() { return ignoreShape_; };
82 bool getUnderline() { return ignoreUnderline_; };
84 bool getMarkUp() { return ignoreMarkUp_; };
86 bool getStrikeOut() { return ignoreStrikeOut_; };
88 bool getSectioning() { return ignoreSectioning_; };
90 bool getFrontMatter() { return ignoreFrontMatter_; };
92 bool getColor() { return ignoreColor_; };
94 bool getLanguage() { return ignoreLanguage_; };
96 void setIgnoreFormat(string const & type, bool value);
106 bool ignoreUnderline_;
110 bool ignoreStrikeOut_;
112 bool ignoreSectioning_;
114 bool ignoreFrontMatter_;
118 bool ignoreLanguage_;
122 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
124 if (type == "color") {
125 ignoreColor_ = value;
127 else if (type == "language") {
128 ignoreLanguage_ = value;
130 else if (type == "sectioning") {
131 ignoreSectioning_ = value;
132 ignoreFrontMatter_ = value;
134 else if (type == "font") {
135 ignoreSeries_ = value;
136 ignoreShape_ = value;
137 ignoreFamily_ = value;
139 else if (type == "series") {
140 ignoreSeries_ = value;
142 else if (type == "shape") {
143 ignoreShape_ = value;
145 else if (type == "family") {
146 ignoreFamily_ = value;
148 else if (type == "markup") {
149 ignoreMarkUp_ = value;
151 else if (type == "underline") {
152 ignoreUnderline_ = value;
154 else if (type == "strike") {
155 ignoreStrikeOut_ = value;
159 // The global variable that can be changed from outside
160 IgnoreFormats ignoreFormats;
163 void setIgnoreFormat(string const & type, bool value)
165 ignoreFormats.setIgnoreFormat(type, value);
171 bool parse_bool(docstring & howto)
176 howto = split(howto, var, ' ');
181 class MatchString : public binary_function<Paragraph, pos_type, int>
184 MatchString(docstring const & s, bool cs, bool mw)
185 : str(s), case_sens(cs), whole_words(mw)
188 // returns true if the specified string is at the specified position
189 // del specifies whether deleted strings in ct mode will be considered
190 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
192 return par.find(str, case_sens, whole_words, pos, del);
200 // match whole words only
205 int findForward(DocIterator & cur, MatchString const & match,
206 bool find_del = true)
208 for (; cur; cur.forwardChar())
209 if (cur.inTexted()) {
210 int len = match(cur.paragraph(), cur.pos(), find_del);
218 int findBackwards(DocIterator & cur, MatchString const & match,
219 bool find_del = true)
223 if (cur.inTexted()) {
224 int len = match(cur.paragraph(), cur.pos(), find_del);
233 bool searchAllowed(docstring const & str)
236 frontend::Alert::error(_("Search error"), _("Search string is empty"));
243 bool findOne(BufferView * bv, docstring const & searchstr,
244 bool case_sens, bool whole, bool forward,
245 bool find_del = true, bool check_wrap = false)
247 if (!searchAllowed(searchstr))
250 DocIterator cur = forward
251 ? bv->cursor().selectionEnd()
252 : bv->cursor().selectionBegin();
254 MatchString const match(searchstr, case_sens, whole);
256 int match_len = forward
257 ? findForward(cur, match, find_del)
258 : findBackwards(cur, match, find_del);
261 bv->putSelectionAt(cur, match_len, !forward);
262 else if (check_wrap) {
263 DocIterator cur_orig(bv->cursor());
266 q = _("End of file reached while searching forward.\n"
267 "Continue searching from the beginning?");
269 q = _("Beginning of file reached while searching backward.\n"
270 "Continue searching from the end?");
271 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
272 q, 0, 1, _("&Yes"), _("&No"));
273 if (wrap_answer == 0) {
275 bv->cursor().clear();
276 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
278 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
279 bv->cursor().backwardPos();
281 bv->clearSelection();
282 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
285 bv->cursor().setCursor(cur_orig);
289 return match_len > 0;
293 int replaceAll(BufferView * bv,
294 docstring const & searchstr, docstring const & replacestr,
295 bool case_sens, bool whole)
297 Buffer & buf = bv->buffer();
299 if (!searchAllowed(searchstr) || buf.isReadonly())
302 DocIterator cur_orig(bv->cursor());
304 MatchString const match(searchstr, case_sens, whole);
307 int const rsize = replacestr.size();
308 int const ssize = searchstr.size();
311 cur.setCursor(doc_iterator_begin(&buf));
312 int match_len = findForward(cur, match, false);
313 while (match_len > 0) {
314 // Backup current cursor position and font.
315 pos_type const pos = cur.pos();
316 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
318 int striked = ssize -
319 cur.paragraph().eraseChars(pos, pos + match_len,
320 buf.params().track_changes);
321 cur.paragraph().insert(pos, replacestr, font,
322 Change(buf.params().track_changes
324 : Change::UNCHANGED));
325 for (int i = 0; i < rsize + striked; ++i)
328 match_len = findForward(cur, match, false);
331 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
333 cur_orig.fixIfBroken();
334 bv->setCursor(cur_orig);
340 // the idea here is that we are going to replace the string that
341 // is selected IF it is the search string.
342 // if there is a selection, but it is not the search string, then
343 // we basically ignore it. (FIXME We ought to replace only within
345 // if there is no selection, then:
346 // (i) if some search string has been provided, then we find it.
347 // (think of how the dialog works when you hit "replace" the
349 // (ii) if no search string has been provided, then we treat the
350 // word the cursor is in as the search string. (why? i have no
351 // idea.) but this only works in text?
353 // returns the number of replacements made (one, if any) and
354 // whether anything at all was done.
355 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
356 docstring const & replacestr, bool case_sens,
357 bool whole, bool forward, bool findnext)
359 Cursor & cur = bv->cursor();
361 if (!cur.selection()) {
362 // no selection, non-empty search string: find it
363 if (!searchstr.empty()) {
364 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
365 return make_pair(found, 0);
367 // empty search string
370 return make_pair(false, 0);
371 // select current word and treat it as the search string.
372 // This causes a minor bug as undo will restore this selection,
373 // which the user did not create (#8986).
374 cur.innerText()->selectWord(cur, WHOLE_WORD);
375 searchstr = cur.selectionAsString(false);
378 // if we still don't have a search string, report the error
380 if (!searchAllowed(searchstr))
381 return make_pair(false, 0);
383 bool have_selection = cur.selection();
384 docstring const selected = cur.selectionAsString(false);
387 ? searchstr == selected
388 : compare_no_case(searchstr, selected) == 0;
390 // no selection or current selection is not search word:
391 // just find the search word
392 if (!have_selection || !match) {
393 found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
394 return make_pair(found, 0);
397 // we're now actually ready to replace. if the buffer is
398 // read-only, we can't, though.
399 if (bv->buffer().isReadonly())
400 return make_pair(false, 0);
402 cap::replaceSelectionWithString(cur, replacestr);
404 cur.pos() += replacestr.length();
405 LASSERT(cur.pos() <= cur.lastpos(),
406 cur.pos() = cur.lastpos());
409 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
411 return make_pair(true, 1);
417 docstring const find2string(docstring const & search,
418 bool casesensitive, bool matchword, bool forward)
422 << int(casesensitive) << ' '
423 << int(matchword) << ' '
429 docstring const replace2string(docstring const & replace,
430 docstring const & search,
431 bool casesensitive, bool matchword,
432 bool all, bool forward, bool findnext)
435 ss << replace << '\n'
437 << int(casesensitive) << ' '
438 << int(matchword) << ' '
440 << int(forward) << ' '
446 bool lyxfind(BufferView * bv, FuncRequest const & ev)
448 if (!bv || ev.action() != LFUN_WORD_FIND)
451 //lyxerr << "find called, cmd: " << ev << endl;
453 // data is of the form
455 // <casesensitive> <matchword> <forward>"
457 docstring howto = split(ev.argument(), search, '\n');
459 bool casesensitive = parse_bool(howto);
460 bool matchword = parse_bool(howto);
461 bool forward = parse_bool(howto);
463 return findOne(bv, search, casesensitive, matchword, forward, true, true);
467 bool lyxreplace(BufferView * bv,
468 FuncRequest const & ev, bool has_deleted)
470 if (!bv || ev.action() != LFUN_WORD_REPLACE)
473 // data is of the form
476 // <casesensitive> <matchword> <all> <forward> <findnext>"
479 docstring howto = split(ev.argument(), rplc, '\n');
480 howto = split(howto, search, '\n');
482 bool casesensitive = parse_bool(howto);
483 bool matchword = parse_bool(howto);
484 bool all = parse_bool(howto);
485 bool forward = parse_bool(howto);
486 bool findnext = howto.empty() ? true : parse_bool(howto);
491 int replace_count = 0;
493 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
494 update = replace_count > 0;
497 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
499 replace_count = rv.second;
502 Buffer const & buf = bv->buffer();
504 // emit message signal.
505 buf.message(_("String not found."));
507 if (replace_count == 0) {
508 buf.message(_("String found."));
509 } else if (replace_count == 1) {
510 buf.message(_("String has been replaced."));
512 docstring const str =
513 bformat(_("%1$d strings have been replaced."), replace_count);
517 } else if (findnext) {
518 // if we have deleted characters, we do not replace at all, but
519 // rather search for the next occurence
520 if (findOne(bv, search, casesensitive, matchword, forward, true, findnext))
523 bv->message(_("String not found."));
529 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
531 for (; cur; cur.forwardPos())
532 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
536 DocIterator cur_orig(bv->cursor());
537 docstring q = _("End of file reached while searching forward.\n"
538 "Continue searching from the beginning?");
539 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
540 q, 0, 1, _("&Yes"), _("&No"));
541 if (wrap_answer == 0) {
542 bv->cursor().clear();
543 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
544 bv->clearSelection();
545 cur.setCursor(bv->cursor().selectionBegin());
546 if (findNextChange(bv, cur, false))
549 bv->cursor().setCursor(cur_orig);
556 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
558 for (cur.backwardPos(); cur; cur.backwardPos()) {
559 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
564 DocIterator cur_orig(bv->cursor());
565 docstring q = _("Beginning of file reached while searching backward.\n"
566 "Continue searching from the end?");
567 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
568 q, 0, 1, _("&Yes"), _("&No"));
569 if (wrap_answer == 0) {
570 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
571 bv->cursor().backwardPos();
572 bv->clearSelection();
573 cur.setCursor(bv->cursor().selectionBegin());
574 if (findPreviousChange(bv, cur, false))
577 bv->cursor().setCursor(cur_orig);
584 bool selectChange(Cursor & cur, bool forward)
586 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
588 Change ch = cur.paragraph().lookupChange(cur.pos());
590 CursorSlice tip1 = cur.top();
591 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
592 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
593 if (!ch2.isSimilarTo(ch))
596 CursorSlice tip2 = cur.top();
597 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
599 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
600 if (!ch2.isSimilarTo(ch)) {
601 // take a step forward to correctly set the selection
609 cur.bv().mouseSetCursor(cur, false);
611 cur.bv().mouseSetCursor(cur, true);
619 bool findChange(BufferView * bv, bool forward)
622 cur.setCursor(forward ? bv->cursor().selectionEnd()
623 : bv->cursor().selectionBegin());
624 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
625 return selectChange(cur, forward);
630 bool findNextChange(BufferView * bv)
632 return findChange(bv, true);
636 bool findPreviousChange(BufferView * bv)
638 return findChange(bv, false);
645 typedef vector<pair<string, string> > Escapes;
647 /// A map of symbols and their escaped equivalent needed within a regex.
648 /// @note Beware of order
649 Escapes const & get_regexp_escapes()
651 typedef std::pair<std::string, std::string> P;
653 static Escapes escape_map;
654 if (escape_map.empty()) {
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(".", "_x_."));
665 escape_map.push_back(P("\\", "(?:\\\\|\\\\backslash)"));
666 escape_map.push_back(P("~", "(?:\\\\textasciitilde|\\\\sim)"));
667 escape_map.push_back(P("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
668 escape_map.push_back(P("_x_", "\\"));
673 /// A map of lyx escaped strings and their unescaped equivalent.
674 Escapes const & get_lyx_unescapes()
676 typedef std::pair<std::string, std::string> P;
678 static Escapes escape_map;
679 if (escape_map.empty()) {
680 escape_map.push_back(P("\\%", "%"));
681 escape_map.push_back(P("\\mathcircumflex ", "^"));
682 escape_map.push_back(P("\\mathcircumflex", "^"));
683 escape_map.push_back(P("\\backslash ", "\\"));
684 escape_map.push_back(P("\\backslash", "\\"));
685 escape_map.push_back(P("\\\\{", "_x_<"));
686 escape_map.push_back(P("\\\\}", "_x_>"));
687 escape_map.push_back(P("\\sim ", "~"));
688 escape_map.push_back(P("\\sim", "~"));
693 /// A map of escapes turning a regexp matching text to one matching latex.
694 Escapes const & get_regexp_latex_escapes()
696 typedef std::pair<std::string, std::string> P;
698 static Escapes escape_map;
699 if (escape_map.empty()) {
700 escape_map.push_back(P("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\}|\\\\textbackslash)"));
701 escape_map.push_back(P("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
702 escape_map.push_back(P("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
703 escape_map.push_back(P("\\[", "\\{\\[\\}"));
704 escape_map.push_back(P("\\]", "\\{\\]\\}"));
705 escape_map.push_back(P("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\textasciicircum|\\\\mathcircumflex)"));
706 escape_map.push_back(P("%", "\\\\\\%"));
707 escape_map.push_back(P("#", "\\\\#"));
712 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
713 ** the found occurrence were escaped.
715 string apply_escapes(string s, Escapes const & escape_map)
717 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
718 Escapes::const_iterator it;
719 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
720 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
721 unsigned int pos = 0;
722 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
723 s.replace(pos, it->first.length(), it->second);
724 LYXERR(Debug::FIND, "After escape: " << s);
725 pos += it->second.length();
726 // LYXERR(Debug::FIND, "pos: " << pos);
729 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
734 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
735 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
736 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
737 string escape_for_regex(string s, bool match_latex)
740 while (pos < s.size()) {
741 size_t new_pos = s.find("\\regexp{", pos);
742 if (new_pos == string::npos)
746 LYXERR(Debug::FIND, "new_pos: " << new_pos);
747 t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
748 LYXERR(Debug::FIND, "t [lyx]: " << t);
749 t = apply_escapes(t, get_regexp_escapes());
750 LYXERR(Debug::FIND, "t [rxp]: " << t);
751 s.replace(pos, new_pos - pos, t);
752 new_pos = pos + t.size();
753 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
754 LYXERR(Debug::FIND, "new_pos: " << new_pos);
755 if (new_pos == s.size())
758 // Might fail if \\endregexp{} is preceeded by unexpected stuff (weird escapes)
759 size_t end_pos = s.find("\\endregexp{}}", new_pos + 8);
760 LYXERR(Debug::FIND, "end_pos: " << end_pos);
761 t = s.substr(new_pos + 8, end_pos - (new_pos + 8));
762 LYXERR(Debug::FIND, "t in regexp : " << t);
763 t = apply_escapes(t, get_lyx_unescapes());
764 LYXERR(Debug::FIND, "t in regexp [lyx]: " << t);
766 t = apply_escapes(t, get_regexp_latex_escapes());
767 LYXERR(Debug::FIND, "t in regexp [ltx]: " << t);
769 if (end_pos == s.size()) {
770 s.replace(new_pos, end_pos - new_pos, t);
771 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
774 s.replace(new_pos, end_pos + 13 - new_pos, t);
775 LYXERR(Debug::FIND, "Regexp after \\regexp{...\\endregexp{}} removal: " << s);
776 pos = new_pos + t.size();
777 LYXERR(Debug::FIND, "pos: " << pos);
783 /// Wrapper for lyx::regex_replace with simpler interface
784 bool regex_replace(string const & s, string & t, string const & searchstr,
785 string const & replacestr)
787 lyx::regex e(searchstr, regex_constants::ECMAScript);
789 ostream_iterator<char, char> it(oss);
790 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
791 // tolerate t and s be references to the same variable
792 bool rv = (s != oss.str());
798 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
800 ** Verify that closed braces exactly match open braces. This avoids that, for example,
801 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
804 ** Number of open braces that must remain open at the end for the verification to succeed.
806 bool braces_match(string::const_iterator const & beg,
807 string::const_iterator const & end,
811 string::const_iterator it = beg;
812 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
813 for (; it != end; ++it) {
814 // Skip escaped braces in the count
819 } else if (*it == '{') {
821 } else if (*it == '}') {
822 if (open_pars == 0) {
823 LYXERR(Debug::FIND, "Found unmatched closed brace");
829 if (open_pars != unmatched) {
830 LYXERR(Debug::FIND, "Found " << open_pars
831 << " instead of " << unmatched
832 << " unmatched open braces at the end of count");
835 LYXERR(Debug::FIND, "Braces match as expected");
845 MatchResult(): match_len(0),match2end(0), pos(0) {};
848 /** The class performing a match between a position in the document and the FindAdvOptions.
850 class MatchStringAdv {
852 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
854 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
855 ** constructor as opt.search, under the opt.* options settings.
858 ** If set, then match is searched only against beginning of text starting at cur.
859 ** If unset, then match is searched anywhere in text starting at cur.
862 ** The length of the matching text, or zero if no match was found.
864 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
869 /// first buffer on which search was started
870 lyx::Buffer * const p_first_buf;
872 FindAndReplaceOptions const & opt;
875 /// Auxiliary find method (does not account for opt.matchword)
876 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
878 /** Normalize a stringified or latexified LyX paragraph.
882 ** <li>if search is not casesensitive, then lowercase the string;
883 ** <li>remove any newline at begin or end of the string;
884 ** <li>replace any newline in the middle of the string with a simple space;
885 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
888 ** @todo Normalization should also expand macros, if the corresponding
889 ** search option was checked.
891 string normalize(docstring const & s, bool hack_braces) const;
892 // normalized string to search
893 string par_as_string;
894 // regular expression to use for searching
896 // same as regexp, but prefixed with a ".*?"
898 // leading format material as string
899 string lead_as_string;
900 // par_as_string after removal of lead_as_string
901 string par_as_string_nolead;
902 // unmatched open braces in the search string/regexp
904 // number of (.*?) subexpressions added at end of search regexp for closing
905 // environments, math mode, styles, etc...
907 // Are we searching with regular expressions ?
912 static docstring buffer_to_latex(Buffer & buffer)
914 //OutputParams runparams(&buffer.params().encoding());
915 OutputParams runparams(encodings.fromLyXName("utf8"));
916 odocstringstream ods;
918 runparams.nice = true;
919 runparams.flavor = OutputParams::XETEX;
920 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
921 // No side effect of file copying and image conversion
922 runparams.dryrun = true;
923 runparams.for_search = true;
924 pit_type const endpit = buffer.paragraphs().size();
925 for (pit_type pit = 0; pit != endpit; ++pit) {
926 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
927 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
933 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
936 if (!opt.ignoreformat) {
937 str = buffer_to_latex(buffer);
939 // OutputParams runparams(&buffer.params().encoding());
940 OutputParams runparams(encodings.fromLyXName("utf8"));
941 runparams.nice = true;
942 runparams.flavor = OutputParams::XETEX;
943 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
944 runparams.dryrun = true;
945 runparams.for_search = true;
946 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
947 Paragraph const & par = buffer.paragraphs().at(pit);
948 LYXERR(Debug::FIND, "Adding to search string: '"
949 << par.asString(pos_type(0), par.size(),
950 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
953 str += par.asString(pos_type(0), par.size(),
954 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
962 /// Return separation pos between the leading material and the rest
963 static size_t identifyLeading(string const & s)
966 // @TODO Support \item[text]
967 // Kornel: Added textsl, textsf, textit, texttt and noun
968 // + allow to search for colored text too
969 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)\\*?)\\{", "")
970 || regex_replace(t, t, REGEX_BOS "\\$", "")
971 || regex_replace(t, t, REGEX_BOS "\\\\\\[ ", "")
972 || regex_replace(t, t, REGEX_BOS " ?\\\\item\\{[a-z]+\\}", "")
973 || regex_replace(t, t, REGEX_BOS "\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
975 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
980 * Given a latexified string, retrieve some handled features
981 * The features of the regex will later be compared with the features
982 * of the searched text. If the regex features are not a
983 * subset of the analized, then, in not format ignoring search
984 * we can early stop the search in the relevant inset.
986 typedef map<string, bool> Features;
988 static Features identifyFeatures(string const & s)
990 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
991 static regex const valid("^("
993 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
994 "emph|noun|text(bf|md|sl|sf|it|tt)|"
995 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
996 "(u|uu)line|(s|x)out|uwave|"
997 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
998 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1003 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1006 if (sub.str(1).compare("regexp") == 0) {
1010 string token = sub.str(1);
1012 if (regex_match(token, sub2, valid)) {
1020 if (sub.str(1).compare("endregexp") == 0) {
1030 * defines values features of a key "\\[a-z]+{"
1035 /* Char type with content discarded
1036 * like \hspace{1cm} */
1038 /* Char, like \backslash */
1040 /* \part, \section*, ... */
1042 /* title, author etc */
1044 /* \foreignlanguage{ngerman}, ... */
1047 * to discard language in content */
1050 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1052 /* fonts, colors, markups, ... */
1054 /* footnotesize, ... large, ...
1055 * Ignore all of them */
1058 /* inputencoding, ...
1059 * Discard also content, because they do not help in search */
1062 * like remove, but also all arguments */
1064 /* item, listitem */
1066 /* tex, latex, ... like isChar */
1068 /* like \lettrine[lines=5]{}{} */
1070 /* End of arguments marker for lettrine,
1071 * so that they can be ignored */
1081 parenthesiscount(1),
1085 KeyInfo(KeyType type, int parcount, bool disable)
1091 parenthesiscount(parcount),
1100 int parenthesiscount;
1102 bool used; /* by pattern */
1107 Border(int l=0, int u=0) : low(l), upper(u) {};
1112 #define MAXOPENED 30
1114 bool isPatternString_;
1116 explicit Intervall(bool isPattern, string const & p) :
1117 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1118 hasTitle(false), langcount(0)
1126 static vector<Border> borders;
1127 int depts[MAXOPENED];
1128 int closes[MAXOPENED];
1129 int actualdeptindex;
1130 int previousNotIgnored(int);
1131 int nextNotIgnored(int);
1132 void handleOpenP(int i);
1133 void handleCloseP(int i, bool closingAllowed);
1134 void resetOpenedP(int openPos);
1135 void addIntervall(int upper);
1136 void addIntervall(int low, int upper); /* if explicit */
1137 void removeAccents();
1138 void setForDefaultLang(KeyInfo &defLang);
1139 int findclosing(int start, int end, char up, char down, int repeat);
1140 void handleParentheses(int lastpos, bool closingAllowed);
1142 int langcount; // Number of disabled language specs up to current position in actual interval
1143 int isOpeningPar(int pos);
1145 void output(ostringstream &os, int lastpos);
1146 // string show(int lastpos);
1149 vector<Border> Intervall::borders = vector<Border>(30);
1151 int Intervall::isOpeningPar(int pos)
1153 if ((pos < 0) || (size_t(pos) >= par.size()))
1155 if (par[pos] != '{')
1157 if (size_t(pos) + 2 >= par.size())
1159 if (par[pos+2] != '}')
1161 if (par[pos+1] == '[' || par[pos+1] == ']')
1166 void Intervall::setForDefaultLang(KeyInfo &defLang)
1168 // Enable the use of first token again
1169 if (ignoreidx >= 0) {
1170 int value = defLang._tokenstart + defLang._tokensize;
1172 if (borders[0].low < value)
1173 borders[0].low = value;
1174 if (borders[0].upper < value)
1175 borders[0].upper = value;
1180 static void checkDepthIndex(int val)
1182 static int maxdepthidx = MAXOPENED-2;
1183 static int lastmaxdepth = 0;
1184 if (val > lastmaxdepth) {
1185 LYXERR0("Depth reached " << val);
1188 if (val > maxdepthidx) {
1190 LYXERR0("maxdepthidx now " << val);
1195 // Not needed, because borders are now dynamically expanded
1196 static void checkIgnoreIdx(int val)
1198 static int lastmaxignore = -1;
1199 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1200 LYXERR0("IgnoreIdx reached " << val);
1201 lastmaxignore = val;
1207 * Expand the region of ignored parts of the input latex string
1208 * The region is only relevant in output()
1210 void Intervall::addIntervall(int low, int upper)
1213 if (low == upper) return;
1214 for (idx = ignoreidx+1; idx > 0; --idx) {
1215 if (low > borders[idx-1].upper) {
1219 Border br(low, upper);
1220 if (idx > ignoreidx) {
1221 if (borders.size() <= size_t(idx)) {
1222 borders.push_back(br);
1228 // checkIgnoreIdx(ignoreidx);
1232 // Expand only if one of the new bound is inside the interwall
1233 // We know here that br.low > borders[idx-1].upper
1234 if (br.upper < borders[idx].low) {
1235 // We have to insert at this pos
1236 if (size_t(ignoreidx+1) >= borders.size()) {
1237 borders.push_back(borders[ignoreidx]);
1240 borders[ignoreidx+1] = borders[ignoreidx];
1242 for (int i = ignoreidx; i > idx; --i) {
1243 borders[i] = borders[i-1];
1247 // checkIgnoreIdx(ignoreidx);
1250 // Here we know, that we are overlapping
1251 if (br.low > borders[idx].low)
1252 br.low = borders[idx].low;
1253 // check what has to be concatenated
1255 for (int i = idx; i <= ignoreidx; i++) {
1256 if (br.upper >= borders[i].low) {
1258 if (br.upper < borders[i].upper)
1259 br.upper = borders[i].upper;
1265 // count should be >= 1 here
1268 for (int i = idx + count; i <= ignoreidx; i++) {
1269 borders[i-count+1] = borders[i];
1271 ignoreidx -= count - 1;
1277 typedef map<string, string> AccentsMap;
1278 static AccentsMap accents = map<string, string>();
1280 static void buildaccent(string n, string param, string values)
1284 const char delim = '|';
1285 while (getline(s, name, delim)) {
1287 for (size_t i = 0; i < param.size(); i++) {
1288 string key = name + "{" + param[i] + "}";
1289 // get the corresponding utf8-value
1290 if ((values[start] & 0xc0) != 0xc0) {
1291 // should not happen, utf8 encoding starts at least with 11xxxxxx
1292 // but value for '\dot{i}' is 'i', which is ascii
1293 if ((values[start] & 0x80) == 0) {
1295 accents[key] = values.substr(start, 1);
1296 // LYXERR0("" << key << "=" << accents[key]);
1301 for (int j = 1; ;j++) {
1302 if (start + j >= values.size()) {
1303 accents[key] = values.substr(start, j);
1304 start = values.size() - 1;
1307 else if ((values[start+j] & 0xc0) != 0x80) {
1308 // This is the first byte of following utf8 char
1309 accents[key] = values.substr(start, j);
1311 // LYXERR0("" << key << "=" << accents[key]);
1319 static void buildAccentsMap()
1321 accents["imath"] = "ı";
1323 accents["jmath"] = "ȷ";
1324 accents["lyxmathsym{ß}"] = "ß";
1325 accents["text{ß}"] = "ß";
1326 accents["ddot{\\imath}"] = "ï";
1327 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1328 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1329 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1330 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1331 accents["acute{\\imath}"] = "í";
1332 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1333 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1334 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1335 buildaccent("mathring|r", "aAuUwy",
1337 accents["check{\\imath}"] = "ǐ";
1338 accents["check{\\jmath}"] = "ǰ";
1339 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1340 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1341 accents["hat{\\imath}"] = "î";
1342 accents["hat{\\jmath}"] = "ĵ";
1343 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1344 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1345 accents["bar{\\imath}"] = "ī";
1346 buildaccent("bar|=", "aAeEiIoOuUyY",
1347 "āĀēĒīĪōŌūŪȳȲ"); // macron
1348 accents["tilde{\\imath}"] = "ĩ";
1349 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1350 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1351 accents["breve{\\imath}"] = "ĭ";
1352 buildaccent("breve|u", "aAeEgGiIoOuU",
1353 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1354 accents["grave{\\imath}"] = "ì";
1355 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1356 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1357 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1358 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1359 buildaccent("ogonek|k", "AaEeIiUuOo",
1360 "ĄąĘęĮįŲųǪǫ"); // ogonek
1361 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1362 "ÇçĢĢĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1363 buildaccent("subring|textsubring", "Aa",
1365 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1366 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1367 buildaccent("subtilde|textsubtilde", "EeIiUu",
1368 "ḚḛḬḭṴṵ"); // subtilde
1369 accents["dgrave{\\imath}"] = "ȉ";
1370 accents["textdoublegrave{\\i}"] = "ȉ";
1371 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1372 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1373 accents["rcap{\\imath}"] = "ȉ";
1374 accents["textroundcap{\\i}"] = "ȉ";
1375 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1376 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1377 buildaccent("slashed", "oO",
1382 * Created accents in math or regexp environment
1383 * are macros, but we need the utf8 equivalent
1385 void Intervall::removeAccents()
1387 if (accents.empty())
1389 static regex const accre("\\\\(([\\S]|grave|breve|lyxmathsym|text|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}|(i|imath|jmath)(?![a-zA-Z]))");
1391 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1393 string key = sub.str(1);
1394 if (accents.find(key) != accents.end()) {
1395 string val = accents[key];
1396 size_t pos = sub.position(size_t(0));
1397 for (size_t i = 0; i < val.size(); i++) {
1398 par[pos+i] = val[i];
1400 addIntervall(pos+val.size(), pos + sub.str(0).size());
1401 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1402 // remove traces of any remaining chars
1407 LYXERR0("Not added accent for \"" << key << "\"");
1412 void Intervall::handleOpenP(int i)
1415 depts[actualdeptindex] = i+1;
1416 closes[actualdeptindex] = -1;
1417 checkDepthIndex(actualdeptindex);
1420 void Intervall::handleCloseP(int i, bool closingAllowed)
1422 if (actualdeptindex <= 0) {
1423 if (! closingAllowed)
1424 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1425 // if we are at the very end
1426 addIntervall(i, i+1);
1429 closes[actualdeptindex] = i+1;
1434 void Intervall::resetOpenedP(int openPos)
1436 // Used as initializer for foreignlanguage entry
1437 actualdeptindex = 1;
1438 depts[1] = openPos+1;
1442 int Intervall::previousNotIgnored(int start)
1444 int idx = 0; /* int intervalls */
1445 for (idx = ignoreidx; idx >= 0; --idx) {
1446 if (start > borders[idx].upper)
1448 if (start >= borders[idx].low)
1449 start = borders[idx].low-1;
1454 int Intervall::nextNotIgnored(int start)
1456 int idx = 0; /* int intervalls */
1457 for (idx = 0; idx <= ignoreidx; idx++) {
1458 if (start < borders[idx].low)
1460 if (start < borders[idx].upper)
1461 start = borders[idx].upper;
1466 typedef map<string, KeyInfo> KeysMap;
1467 typedef vector< KeyInfo> Entries;
1468 static KeysMap keys = map<string, KeyInfo>();
1474 Intervall interval_;
1475 void buildKeys(bool);
1476 void buildEntries(bool);
1477 void makeKey(const string &, KeyInfo, bool isPatternString);
1478 void processRegion(int start, int region_end); /* remove {} parts */
1479 void removeHead(KeyInfo&, int count=0);
1482 LatexInfo(string const & par, bool isPatternString)
1483 : entidx_(-1), interval_(isPatternString, par)
1485 buildKeys(isPatternString);
1486 entries_ = vector<KeyInfo>();
1487 buildEntries(isPatternString);
1491 if (entries_.empty()) {
1494 if (entries_[0].keytype == KeyInfo::isTitle) {
1495 if (! entries_[0].disabled) {
1496 interval_.hasTitle = true;
1497 interval_.titleValue = entries_[0].head;
1500 interval_.hasTitle = false;
1501 interval_.titleValue = "";
1503 removeHead(entries_[0]);
1504 if (entries_.size() > 1)
1513 if (int(entries_.size()) > entidx_) {
1520 bool setNextKey(int idx) {
1521 if ((idx == entidx_) && (entidx_ >= 0)) {
1528 int find(int start, KeyInfo::KeyType keytype) {
1532 while (tmpIdx < int(entries_.size())) {
1533 if (entries_[tmpIdx].keytype == keytype)
1539 int process(ostringstream &os, KeyInfo &actual);
1540 int dispatch(ostringstream &os, int previousStart, KeyInfo &actual);
1541 // string show(int lastpos) { return interval.show(lastpos);};
1542 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1543 KeyInfo &getKeyInfo(int keyinfo) {
1544 static KeyInfo invalidInfo = KeyInfo();
1545 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1548 return entries_[keyinfo];
1550 void setForDefaultLang(KeyInfo &defLang) {interval_.setForDefaultLang(defLang);};
1551 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1555 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1559 for (int i = start; i < end; i += 1 + skip) {
1563 if (c == '\\') skip = 1;
1567 else if (c == down) {
1570 if ((repeat <= 0) || (par[i+1] != up))
1588 vector<MathEntry> entries_;
1593 void insert(string const & wait, size_t start, size_t end) {
1594 MathEntry m = MathEntry();
1596 m.mathStart = start;
1598 m.mathSize = end - start;
1599 entries_.push_back(m);
1601 bool empty() { return entries_.empty(); };
1602 size_t getEndPos() {
1603 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1606 return entries_[actualIdx_].mathEnd;
1608 size_t getStartPos() {
1609 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1610 return 100000; /* definitely enough? */
1612 return entries_[actualIdx_].mathStart;
1614 size_t getFirstPos() {
1616 return getStartPos();
1619 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1622 return entries_[actualIdx_].mathSize;
1624 void incrEntry() { actualIdx_++; };
1627 void LatexInfo::buildEntries(bool isPatternString)
1629 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1630 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1631 static bool disableLanguageOverride = false;
1632 smatch sub, submath;
1633 bool evaluatingRegexp = false;
1635 bool evaluatingMath = false;
1636 bool evaluatingCode = false;
1638 bool evaluatingOptional = false;
1639 size_t optionalEnd = 0;
1642 bool math_end_waiting = false;
1643 size_t math_pos = 10000;
1646 interval_.removeAccents();
1648 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1650 if (math_end_waiting) {
1651 size_t pos = submath.position(size_t(0));
1652 if ((math_end == "$") &&
1653 (submath.str(0) == "$") &&
1654 (interval_.par[pos-1] != '\\')) {
1655 mi.insert("$", math_pos, pos + 1);
1656 math_end_waiting = false;
1658 else if ((math_end == "\\]") &&
1659 (submath.str(0) == "\\]")) {
1660 mi.insert("\\]", math_pos, pos + 2);
1661 math_end_waiting = false;
1663 else if ((submath.str(1).compare("end") == 0) &&
1664 (submath.str(2).compare(math_end) == 0)) {
1665 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1666 math_end_waiting = false;
1672 if (submath.str(1).compare("begin") == 0) {
1673 math_end_waiting = true;
1674 math_end = submath.str(2);
1675 math_pos = submath.position(size_t(0));
1677 else if (submath.str(0).compare("\\[") == 0) {
1678 math_end_waiting = true;
1680 math_pos = submath.position(size_t(0));
1682 else if (submath.str(0) == "$") {
1683 size_t pos = submath.position(size_t(0));
1684 if ((pos == 0) || (interval_.par[pos-1] != '\\')) {
1685 math_end_waiting = true;
1692 // Ignore language if there is math somewhere in pattern-string
1693 if (isPatternString) {
1696 keys["foreignlanguage"].disabled = true;
1697 disableLanguageOverride = true;
1700 disableLanguageOverride = false;
1703 if (disableLanguageOverride) {
1704 keys["foreignlanguage"].disabled = true;
1707 math_pos = mi.getFirstPos();
1708 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1710 string key = sub.str(3);
1712 if (sub.str(0)[0] == '\\')
1713 key = sub.str(0)[1];
1717 size_t k_pos = sub.position(size_t(0));
1718 if ((k_pos > 0) && (interval_.par[k_pos - 1] == '\\')) {
1719 // Escaped '$', ignoring
1725 if (evaluatingRegexp) {
1726 if (sub.str(1).compare("endregexp") == 0) {
1727 evaluatingRegexp = false;
1728 // found._tokenstart already set
1729 found._dataEnd = sub.position(size_t(0)) + 13;
1730 found._dataStart = found._dataEnd;
1731 found._tokensize = found._dataEnd - found._tokenstart;
1732 found.parenthesiscount = 0;
1733 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1740 if (evaluatingMath) {
1741 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1743 evaluatingMath = false;
1745 math_pos = mi.getStartPos();
1747 if (keys.find(key) == keys.end()) {
1748 found = KeyInfo(KeyInfo::isStandard, 0, true);
1749 if (isPatternString) {
1750 found.keytype = KeyInfo::isChar;
1751 found.disabled = false;
1758 if (key.compare("regexp") == 0) {
1759 evaluatingRegexp = true;
1760 found._tokenstart = sub.position(size_t(0));
1761 found._tokensize = 0;
1765 // Handle the other params of key
1766 if (found.keytype == KeyInfo::isIgnored)
1768 else if (found.keytype == KeyInfo::isMath) {
1769 if (size_t(sub.position(size_t(0))) == math_pos) {
1771 found._tokenstart = sub.position(size_t(0));
1772 found._tokensize = mi.getSize();
1773 found._dataEnd = found._tokenstart + found._tokensize;
1774 found._dataStart = found._dataEnd;
1775 found.parenthesiscount = 0;
1776 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1777 evaluatingMath = true;
1780 // begin|end of unknown env, discard
1781 // First handle tables
1782 // longtable|tabular
1783 bool discardComment;
1785 found.keytype = KeyInfo::doRemove;
1786 if ((sub.str(5).compare("longtable") == 0) ||
1787 (sub.str(5).compare("tabular") == 0)) {
1788 discardComment = true; /* '%' */
1791 discardComment = false;
1792 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1794 string token = sub.str(5);
1795 if (regex_match(token, sub2, removeArgs)) {
1796 found.keytype = KeyInfo::removeWithArg;
1799 // discard spaces before pos(0)
1800 int pos = sub.position(size_t(0));
1802 for (count = 0; pos - count > 0; count++) {
1803 char c = interval_.par[pos-count-1];
1804 if (discardComment) {
1805 if ((c != ' ') && (c != '%'))
1811 found._tokenstart = pos - count;
1812 if (sub.str(1).compare(0, 5, "begin") == 0) {
1813 size_t pos1 = pos + sub.str(0).length();
1814 if (sub.str(5).compare("cjk") == 0) {
1815 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1816 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1818 found.keytype = KeyInfo::isMain;
1819 found._dataStart = pos1;
1820 found._dataEnd = interval_.par.length();
1821 found.disabled = keys["foreignlanguage"].disabled;
1822 found.used = keys["foreignlanguage"].used;
1823 found._tokensize = pos1 - found._tokenstart;
1824 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1827 // Swallow possible optional params
1828 while (interval_.par[pos1] == '[') {
1829 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1831 // Swallow also the eventual parameter
1832 if (interval_.par[pos1] == '{') {
1833 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1836 found._dataEnd = pos1;
1838 found._dataStart = found._dataEnd;
1839 found._tokensize = count + found._dataEnd - pos;
1840 found.parenthesiscount = 0;
1841 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1842 found.disabled = true;
1846 // Handle "\end{...}"
1847 found._dataStart = pos + sub.str(0).length();
1848 found._dataEnd = found._dataStart;
1849 found._tokensize = count + found._dataEnd - pos;
1850 found.parenthesiscount = 0;
1851 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1852 found.disabled = true;
1856 else if (found.keytype != KeyInfo::isRegex) {
1857 found._tokenstart = sub.position(size_t(0));
1858 if (found.parenthesiscount == 0) {
1859 // Probably to be discarded
1860 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1861 char following = interval_.par[following_pos];
1862 if (following == ' ')
1863 found.head = "\\" + sub.str(3) + " ";
1864 else if (following == '=') {
1865 // like \uldepth=1000pt
1866 found.head = sub.str(0);
1869 found.head = "\\" + key;
1870 found._tokensize = found.head.length();
1871 found._dataEnd = found._tokenstart + found._tokensize;
1872 found._dataStart = found._dataEnd;
1875 int params = found._tokenstart + key.length() + 1;
1876 if (evaluatingOptional) {
1877 if (size_t(found._tokenstart) > optionalEnd) {
1878 evaluatingOptional = false;
1881 found.disabled = true;
1884 int optend = params;
1885 while (interval_.par[optend] == '[') {
1886 // discard optional parameters
1887 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
1889 if (optend > params) {
1890 key += interval_.par.substr(params, optend-params);
1891 evaluatingOptional = true;
1892 optionalEnd = optend;
1894 string token = sub.str(5);
1895 int closings = found.parenthesiscount;
1896 if (found.parenthesiscount == 1) {
1897 found.head = "\\" + key + "{";
1899 else if (found.parenthesiscount > 1) {
1901 found.head = sub.str(0) + "{";
1902 closings = found.parenthesiscount - 1;
1905 found.head = "\\" + key + "{";
1908 found._tokensize = found.head.length();
1909 found._dataStart = found._tokenstart + found.head.length();
1910 if (found.keytype == KeyInfo::doRemove) {
1911 int endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1912 found._dataStart = endpar;
1913 found._tokensize = found._dataStart - found._tokenstart;
1916 if (interval_.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1917 found._dataStart += 15;
1919 size_t endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1920 if (found.keytype == KeyInfo::isList) {
1921 // Check if it really is list env
1922 static regex const listre("^([a-z]+)$");
1924 if (!regex_match(token, sub2, listre)) {
1925 // Change the key of this entry. It is not in a list/item environment
1926 found.keytype = KeyInfo::endArguments;
1929 if (found.keytype == KeyInfo::noMain) {
1930 evaluatingCode = true;
1932 codeStart = found._dataStart;
1934 else if (evaluatingCode) {
1935 if (size_t(found._dataStart) > codeEnd)
1936 evaluatingCode = false;
1937 else if (found.keytype == KeyInfo::isMain) {
1938 // Disable this key, treate it as standard
1939 found.keytype = KeyInfo::isStandard;
1940 found.disabled = true;
1941 if ((codeEnd == interval_.par.length()) &&
1942 (found._tokenstart == codeStart)) {
1943 // trickery, because the code inset starts
1944 // with \selectlanguage ...
1946 if (entries_.size() > 1) {
1947 entries_[entries_.size()-1]._dataEnd = codeEnd;
1952 if ((endpos == interval_.par.length()) &&
1953 (found.keytype == KeyInfo::doRemove)) {
1954 // Missing closing => error in latex-input?
1955 // therefore do not delete remaining data
1956 found._dataStart -= 1;
1957 found._dataEnd = found._dataStart;
1960 found._dataEnd = endpos;
1962 if (isPatternString) {
1963 keys[key].used = true;
1966 entries_.push_back(found);
1970 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1972 stringstream s(keysstring);
1974 const char delim = '|';
1975 while (getline(s, key, delim)) {
1976 KeyInfo keyII(keyI);
1977 if (isPatternString) {
1980 else if ( !keys[key].used)
1981 keyII.disabled = true;
1986 void LatexInfo::buildKeys(bool isPatternString)
1989 static bool keysBuilt = false;
1990 if (keysBuilt && !isPatternString) return;
1992 // Known standard keys with 1 parameter.
1993 // Split is done, if not at start of region
1994 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1995 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1996 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1997 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1998 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1999 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2001 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2002 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2003 makeKey("section*|subsection*|subsubsection*|paragraph*",
2004 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2005 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2006 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2008 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2010 // Split is done, if not at start of region
2011 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2012 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2014 // Split is done always.
2015 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2019 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2020 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2021 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2022 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2024 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2025 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2026 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2028 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2029 // Custom space/skip, remove the content (== length value)
2030 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2031 // Found in fr/UserGuide.lyx
2032 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2034 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2035 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2036 // Known macros to remove (including their parameter)
2038 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2039 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2040 // handle like standard keys with 1 parameter.
2041 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2043 // Ignore deleted text
2044 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2045 // but preserve added text
2046 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2048 // Macros to remove, but let the parameter survive
2050 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2052 // Remove language spec from content of these insets
2053 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2055 // Same effect as previous, parameter will survive (because there is no one anyway)
2057 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2058 // Remove table decorations
2059 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2060 // Discard shape-header.
2061 // For footnote or shortcut too, because of lang settings
2062 // and wrong handling if used 'KeyInfo::noMain'
2063 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2064 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2065 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2066 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2067 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2068 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2069 // like ('tiny{}' or '\tiny ' ... )
2070 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2072 // Survives, like known character
2073 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2074 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2076 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2077 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2078 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2080 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2081 // Remove RTL/LTR marker
2082 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2083 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2084 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2085 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2086 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2087 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2088 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2089 if (isPatternString) {
2090 // Allow the first searched string to rebuild the keys too
2094 // no need to rebuild again
2100 * Keep the list of actual opened parentheses actual
2101 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2103 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2106 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2110 if (c == '\\') skip = 1;
2111 else if (c == '{') {
2114 else if (c == '}') {
2115 handleCloseP(i, closingAllowed);
2121 string Intervall::show(int lastpos)
2123 int idx = 0; /* int intervalls */
2126 for (idx = 0; idx <= ignoreidx; idx++) {
2127 while (i < lastpos) {
2129 if (i <= borders[idx].low) {
2130 if (borders[idx].low > lastpos)
2131 printsize = lastpos - i;
2133 printsize = borders[idx].low - i;
2134 s += par.substr(i, printsize);
2136 if (i >= borders[idx].low)
2137 i = borders[idx].upper;
2140 i = borders[idx].upper;
2146 s += par.substr(i, lastpos-i);
2152 void Intervall::output(ostringstream &os, int lastpos)
2154 // get number of chars to output
2155 int idx = 0; /* int intervalls */
2158 string startTitle = titleValue;
2159 for (idx = 0; idx <= ignoreidx; idx++) {
2161 if (i <= borders[idx].low) {
2163 if (borders[idx].low > lastpos)
2164 printsize = lastpos - i;
2166 printsize = borders[idx].low - i;
2167 if (printsize > 0) {
2168 os << startTitle << par.substr(i, printsize);
2170 printed += printsize;
2173 handleParentheses(i, false);
2174 if (i >= borders[idx].low)
2175 i = borders[idx].upper;
2178 i = borders[idx].upper;
2185 os << startTitle << par.substr(i, lastpos-i);
2186 printed += lastpos-i;
2188 handleParentheses(lastpos, false);
2190 if (keys["foreignlanguage"].disabled)
2191 startindex = actualdeptindex-langcount;
2193 startindex = actualdeptindex;
2194 for (int i = startindex; i > 0; --i) {
2197 if (hasTitle && (printed > 0))
2199 if (! isPatternString_)
2201 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2204 void LatexInfo::processRegion(int start, int region_end)
2206 while (start < region_end) { /* Let {[} and {]} survive */
2207 int cnt = interval_.isOpeningPar(start);
2209 // Closing is allowed past the region
2210 int closing = interval_.findclosing(start+1, interval_.par.length());
2211 interval_.addIntervall(start, start+1);
2212 interval_.addIntervall(closing, closing+1);
2216 start = interval_.nextNotIgnored(start+1);
2220 void LatexInfo::removeHead(KeyInfo &actual, int count)
2222 if (actual.parenthesiscount == 0) {
2223 // "{\tiny{} ...}" ==> "{{} ...}"
2224 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2227 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2228 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2229 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2233 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2236 switch (actual.keytype)
2238 case KeyInfo::isTitle: {
2240 nextKeyIdx = getNextKey();
2243 case KeyInfo::cleanToStart: {
2244 actual._dataEnd = actual._dataStart;
2245 nextKeyIdx = getNextKey();
2246 // Search for end of arguments
2247 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2249 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2250 entries_[i].disabled = true;
2252 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2254 while (interval_.par[actual._dataEnd] == ' ')
2256 interval_.addIntervall(0, actual._dataEnd+1);
2257 interval_.actualdeptindex = 0;
2258 interval_.depts[0] = actual._dataEnd+1;
2259 interval_.closes[0] = -1;
2262 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2263 if (actual.disabled)
2264 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2266 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2269 case KeyInfo::isChar: {
2270 nextKeyIdx = getNextKey();
2273 case KeyInfo::isSize: {
2274 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2275 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2276 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2277 nextKeyIdx = getNextKey();
2279 // Here _dataStart points to '{', so correct it
2280 actual._dataStart += 1;
2281 actual._tokensize += 1;
2282 actual.parenthesiscount = 1;
2283 if (interval_.par[actual._dataStart] == '}') {
2284 // Determine the end if used like '{\tiny{}...}'
2285 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2286 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2289 // Determine the end if used like '\tiny{...}'
2290 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2292 // Split on this key if not at start
2293 int start = interval_.nextNotIgnored(previousStart);
2294 if (start < actual._tokenstart) {
2295 interval_.output(os, actual._tokenstart);
2296 interval_.addIntervall(start, actual._tokenstart);
2298 // discard entry if at end of actual
2299 nextKeyIdx = process(os, actual);
2303 case KeyInfo::endArguments:
2304 // Remove trailing '{}' too
2305 actual._dataStart += 1;
2306 actual._dataEnd += 1;
2307 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2308 nextKeyIdx = getNextKey();
2310 case KeyInfo::noMain:
2312 case KeyInfo::isStandard: {
2313 if (actual.disabled) {
2315 processRegion(actual._dataStart, actual._dataStart+1);
2316 nextKeyIdx = getNextKey();
2318 // Split on this key if not at datastart of calling entry
2319 int start = interval_.nextNotIgnored(previousStart);
2320 if (start < actual._tokenstart) {
2321 interval_.output(os, actual._tokenstart);
2322 interval_.addIntervall(start, actual._tokenstart);
2324 // discard entry if at end of actual
2325 nextKeyIdx = process(os, actual);
2329 case KeyInfo::removeWithArg: {
2330 nextKeyIdx = getNextKey();
2331 // Search for end of arguments
2332 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2334 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2335 entries_[i].disabled = true;
2337 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2339 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2342 case KeyInfo::doRemove: {
2343 // Remove the key with all parameters and following spaces
2345 for (pos = actual._dataEnd+1; pos < interval_.par.length(); pos++) {
2346 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2349 // Remove also enclosing parentheses [] and {}
2352 while (actual._tokenstart > numpars) {
2353 if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2355 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2361 if (interval_.par[pos+numpars] == ' ')
2365 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2366 nextKeyIdx = getNextKey();
2369 case KeyInfo::isList: {
2370 // Discard space before _tokenstart
2372 for (count = 0; count < actual._tokenstart; count++) {
2373 if (interval_.par[actual._tokenstart-count-1] != ' ')
2376 nextKeyIdx = getNextKey();
2377 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2379 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2381 // How else can we catch this one?
2382 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2383 entries_[i].disabled = true;
2385 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2387 else if (nextKeyIdx > 0) {
2388 // Ignore any lang entries inside data region
2389 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2390 if (entries_[i].keytype == KeyInfo::isMain)
2391 entries_[i].disabled = true;
2394 if (actual.disabled) {
2395 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2398 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2400 if (interval_.par[actual._dataEnd+1] == '[') {
2401 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2402 if ((interval_.par[actual._dataEnd+2] == '{') &&
2403 (interval_.par[posdown-1] == '}')) {
2404 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2405 interval_.addIntervall(posdown-1, posdown+1);
2408 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2409 interval_.addIntervall(posdown, posdown+1);
2411 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2412 if (blk > posdown) {
2413 // Discard at most 1 space after empty item
2415 for (count = 0; count < 1; count++) {
2416 if (interval_.par[blk+count] != ' ')
2420 interval_.addIntervall(blk, blk+count);
2425 case KeyInfo::isSectioning: {
2426 // Discard spaces before _tokenstart
2428 int val = actual._tokenstart;
2429 for (count = 0; count < actual._tokenstart;) {
2430 val = interval_.previousNotIgnored(val-1);
2431 if (interval_.par[val] != ' ')
2434 count = actual._tokenstart - val;
2437 if (actual.disabled) {
2438 removeHead(actual, count);
2439 nextKeyIdx = getNextKey();
2441 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2442 nextKeyIdx = process(os, actual);
2446 case KeyInfo::isMath: {
2447 // Same as regex, use the content unchanged
2448 nextKeyIdx = getNextKey();
2451 case KeyInfo::isRegex: {
2452 // DO NOT SPLIT ON REGEX
2454 nextKeyIdx = getNextKey();
2457 case KeyInfo::isIgnored: {
2458 // Treat like a character for now
2459 nextKeyIdx = getNextKey();
2462 case KeyInfo::isMain: {
2463 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2464 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2465 if (actual._tokenstart > 0) {
2466 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2467 if ((prev >= 0) && interval_.par[prev] == '%')
2468 interval_.addIntervall(prev, prev+1);
2470 if (actual.disabled) {
2472 interval_.langcount++;
2473 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2474 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2475 // Discard also the space before math-equation
2476 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2478 nextKeyIdx = getNextKey();
2479 // interval.resetOpenedP(actual._dataStart-1);
2482 if (actual._tokenstart < 26) {
2483 // for the first (and maybe dummy) language
2484 interval_.setForDefaultLang(actual);
2486 interval_.resetOpenedP(actual._dataStart-1);
2490 case KeyInfo::invalid:
2491 // This cannot happen, already handled
2494 // LYXERR0("Unhandled keytype");
2495 nextKeyIdx = getNextKey();
2502 int LatexInfo::process(ostringstream &os, KeyInfo &actual )
2504 int end = interval_.nextNotIgnored(actual._dataEnd);
2505 int oldStart = actual._dataStart;
2506 int nextKeyIdx = getNextKey();
2508 if ((nextKeyIdx < 0) ||
2509 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2510 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2511 if (oldStart <= end) {
2512 processRegion(oldStart, end);
2517 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2519 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2520 (void) dispatch(os, actual._dataStart, nextKey);
2521 end = nextKey._tokenstart;
2524 processRegion(oldStart, nextKey._tokenstart);
2525 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2527 oldStart = nextKey._dataEnd+1;
2529 // now nextKey is either invalid or is outside of actual._dataEnd
2530 // output the remaining and discard myself
2531 if (oldStart <= end) {
2532 processRegion(oldStart, end);
2534 if (interval_.par[end] == '}') {
2536 // This is the normal case.
2537 // But if using the firstlanguage, the closing may be missing
2539 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2541 if (actual._dataEnd < end)
2542 output_end = interval_.nextNotIgnored(actual._dataEnd);
2544 output_end = interval_.nextNotIgnored(end);
2545 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2546 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2548 // Remove possible empty data
2549 int dstart = interval_.nextNotIgnored(actual._dataStart);
2550 while (interval_.isOpeningPar(dstart) == 1) {
2551 interval_.addIntervall(dstart, dstart+1);
2552 int dend = interval_.findclosing(dstart+1, output_end);
2553 interval_.addIntervall(dend, dend+1);
2554 dstart = interval_.nextNotIgnored(dstart+1);
2556 if (dstart < output_end)
2557 interval_.output(os, output_end);
2558 interval_.addIntervall(actual._tokenstart, end);
2562 string splitOnKnownMacros(string par, bool isPatternString)
2565 LatexInfo li(par, isPatternString);
2566 // LYXERR0("Berfore split: " << par);
2567 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2569 DummyKey._tokensize = 0;
2570 DummyKey._dataStart = 0;
2571 DummyKey._dataEnd = par.length();
2572 DummyKey.disabled = true;
2573 int firstkeyIdx = li.getFirstKey();
2575 if (firstkeyIdx >= 0) {
2576 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2577 DummyKey._tokenstart = firstKey._tokenstart;
2579 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2580 // Use dummy firstKey
2581 firstKey = DummyKey;
2582 (void) li.setNextKey(firstkeyIdx);
2585 if (par.substr(firstKey._dataStart, 2) == "% ")
2586 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2588 nextkeyIdx = li.process(os, firstKey);
2589 while (nextkeyIdx >= 0) {
2590 // Check for a possible gap between the last
2591 // entry and this one
2592 int datastart = li.nextNotIgnored(firstKey._dataStart);
2593 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2594 if ((nextKey._tokenstart > datastart)) {
2596 firstKey._dataStart = datastart;
2597 firstKey._dataEnd = par.length();
2598 (void) li.setNextKey(nextkeyIdx);
2599 // Fake the last opened parenthesis
2600 li.setForDefaultLang(firstKey);
2601 nextkeyIdx = li.process(os, firstKey);
2604 if (nextKey.keytype != KeyInfo::isMain) {
2605 firstKey._dataStart = datastart;
2606 firstKey._dataEnd = nextKey._dataEnd+1;
2607 (void) li.setNextKey(nextkeyIdx);
2608 li.setForDefaultLang(firstKey);
2609 nextkeyIdx = li.process(os, firstKey);
2612 nextkeyIdx = li.process(os, nextKey);
2616 // Handle the remaining
2617 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2618 firstKey._dataEnd = par.length();
2620 if ((firstKey._dataStart < firstKey._dataEnd) &&
2621 (par[firstKey._dataStart] != '}')) {
2622 li.setForDefaultLang(firstKey);
2623 (void) li.process(os, firstKey);
2627 // return string definitelly impossible to match
2628 s = "\\foreignlanguage{ignore}{ }";
2632 s = par; /* no known macros found */
2633 // LYXERR0("After split: " << s);
2638 * Try to unify the language specs in the latexified text.
2639 * Resulting modified string is set to "", if
2640 * the searched tex does not contain all the features in the search pattern
2642 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2644 static Features regex_f;
2645 static int missed = 0;
2646 static bool regex_with_format = false;
2648 int parlen = par.length();
2650 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2653 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2654 // Happens to be there in case of description or labeling environment
2659 // Split the latex input into pieces which
2660 // can be digested by our search engine
2661 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2662 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2663 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2666 result = par.substr(0, parlen);
2667 if (isPatternString) {
2670 regex_f = identifyFeatures(result);
2671 string features = "";
2672 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2673 string a = it->first;
2674 regex_with_format = true;
2675 features += " " + a;
2676 // LYXERR0("Identified regex format:" << a);
2678 LYXERR(Debug::FIND, "Identified Features" << features);
2681 } else if (regex_with_format) {
2682 Features info = identifyFeatures(result);
2683 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2684 string a = it->first;
2685 bool b = it->second;
2686 if (b && ! info[a]) {
2688 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2694 // LYXERR0("No regex formats");
2700 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2701 static int identifyClosing(string & t)
2703 int open_braces = 0;
2705 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2706 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2708 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]" REGEX_EOS, "$1"))
2710 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2712 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2721 static int num_replaced = 0;
2722 static bool previous_single_replace = true;
2724 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2725 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2727 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2728 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2729 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2730 if (opt.replace_all && previous_single_replace) {
2731 previous_single_replace = false;
2734 else if (!opt.replace_all) {
2735 num_replaced = 0; // count number of replaced strings
2736 previous_single_replace = true;
2738 // When using regexp, braces are hacked already by escape_for_regex()
2739 par_as_string = normalize(ds, !use_regexp);
2741 close_wildcards = 0;
2743 size_t lead_size = 0;
2744 // correct the language settings
2745 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2746 if (opt.ignoreformat) {
2748 // if par_as_string_nolead were emty,
2749 // the following call to findAux will always *find* the string
2750 // in the checked data, and thus always using the slow
2751 // examining of the current text part.
2752 par_as_string_nolead = par_as_string;
2755 lead_size = identifyLeading(par_as_string);
2756 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2757 lead_as_string = par_as_string.substr(0, lead_size);
2758 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2762 open_braces = identifyClosing(par_as_string);
2763 identifyClosing(par_as_string_nolead);
2764 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2765 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2767 string lead_as_regexp;
2768 if (lead_size > 0) {
2769 // @todo No need to search for \regexp{} insets in leading material
2770 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2771 par_as_string = par_as_string_nolead;
2772 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2773 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2775 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2776 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2777 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2779 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2780 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2781 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2782 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2783 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2784 || regex_replace(par_as_string, par_as_string,
2785 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2786 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2787 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2791 if (!opt.ignoreformat) {
2792 // Remove extra '\}' at end if not part of \{\.\}
2793 size_t lng = par_as_string.size();
2795 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2797 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2806 if (lng < par_as_string.size())
2807 par_as_string = par_as_string.substr(0,lng);
2810 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2811 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2812 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2813 // replace '[^...]' with '[^...\}\{\\]'
2814 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2815 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2816 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2818 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2821 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2822 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2823 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2824 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2826 // If entered regexp must match at begin of searched string buffer
2827 // Kornel: Added parentheses to use $1 for size of the leading string
2831 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2832 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2833 // so the convert has no effect in that case
2834 for (int i = 8; i > 0; --i) {
2835 string orig = "\\\\" + std::to_string(i);
2836 string dest = "\\" + std::to_string(i+1);
2837 while (regex_replace(par_as_string, par_as_string, orig, dest));
2839 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2840 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2842 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2843 regexp = lyx::regex(regexp_str);
2845 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2846 regexp2 = lyx::regex(regexp2_str);
2851 // Count number of characters in string
2855 // \\[a-zA-Z]+ ==> 1
2856 static int computeSize(string s, int len)
2862 for (int i = 0; i < len; i += skip, count++) {
2865 if (isalpha(s[i+1])) {
2866 for (int j = 2; i+j < len; j++) {
2867 if (! isalpha(s[i+j])) {
2870 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2872 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2880 else if (s[i] == '{') {
2886 else if (s[i] == '-') {
2887 if (s[i+1] == '-') {
2903 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2908 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2911 docstring docstr = stringifyFromForSearch(opt, cur, len);
2912 string str = normalize(docstr, true);
2913 if (!opt.ignoreformat) {
2914 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2917 mres.match_len = -1;
2920 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2921 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2924 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2925 regex const *p_regexp;
2926 regex_constants::match_flag_type flags;
2928 flags = regex_constants::match_continuous;
2931 flags = regex_constants::match_default;
2932 p_regexp = ®exp2;
2934 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2935 if (re_it == sregex_iterator())
2937 match_results<string::const_iterator> const & m = *re_it;
2939 if (0) { // Kornel Benko: DO NOT CHECKK
2940 // Check braces on the segment that matched the entire regexp expression,
2941 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2942 if (!braces_match(m[0].first, m[0].second, open_braces))
2946 // Check braces on segments that matched all (.*?) subexpressions,
2947 // except the last "padding" one inserted by lyx.
2948 for (size_t i = 1; i < m.size() - 1; ++i)
2949 if (!braces_match(m[i].first, m[i].second, open_braces))
2952 // Exclude from the returned match length any length
2953 // due to close wildcards added at end of regexp
2954 // and also the length of the leading (e.g. '\emph{}')
2956 // Whole found string, including the leading: m[0].second - m[0].first
2957 // Size of the leading string: m[1].second - m[1].first
2958 int leadingsize = 0;
2960 leadingsize = m[1].second - m[1].first;
2962 for (size_t i = 0; i < m.size(); i++) {
2963 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2965 if (close_wildcards == 0)
2966 result = m[0].second - m[0].first;
2969 result = m[m.size() - close_wildcards].first - m[0].first;
2971 size_t pos = m.position(size_t(0));
2972 // Ignore last closing characters
2973 while (result > 0) {
2974 if (str[pos+result-1] == '}')
2979 if (result > leadingsize)
2980 result -= leadingsize;
2983 mres.match_len = computeSize(str.substr(pos+leadingsize,result), result);
2984 mres.match2end = str.size() - pos - leadingsize;
2985 mres.pos = pos+leadingsize;
2989 // else !use_regexp: but all code paths above return
2990 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2991 << par_as_string << "', str='" << str << "'");
2992 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2993 << lead_as_string << "', par_as_string_nolead='"
2994 << par_as_string_nolead << "'");
2997 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2998 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2999 if (str.substr(0, par_as_string.size()) == par_as_string) {
3000 mres.match_len = par_as_string.size();
3001 mres.match2end = str.size();
3006 // Start the search _after_ the leading part
3007 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
3008 if (pos != string::npos) {
3009 mres.match_len = par_as_string.size();
3010 mres.match2end = str.size() - pos;
3019 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3021 MatchResult mres = findAux(cur, len, at_begin);
3022 int res = mres.match_len;
3024 "res=" << res << ", at_begin=" << at_begin
3025 << ", matchword=" << opt.matchword
3026 << ", inTexted=" << cur.inTexted());
3027 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
3029 if ((len > 0) && (res < len)) {
3033 Paragraph const & par = cur.paragraph();
3034 bool ws_left = (cur.pos() > 0)
3035 ? par.isWordSeparator(cur.pos() - 1)
3037 bool ws_right = (cur.pos() + len < par.size())
3038 ? par.isWordSeparator(cur.pos() + len)
3041 "cur.pos()=" << cur.pos() << ", res=" << res
3042 << ", separ: " << ws_left << ", " << ws_right
3045 if (ws_left && ws_right) {
3046 // Check for word separators inside the found 'word'
3047 for (int i = 0; i < len; i++) {
3048 if (par.isWordSeparator(cur.pos() + i)) {
3060 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
3063 if (! opt.casesensitive)
3064 t = lyx::to_utf8(lowercase(s));
3066 t = lyx::to_utf8(s);
3067 // Remove \n at begin
3068 while (!t.empty() && t[0] == '\n')
3071 while (!t.empty() && t[t.size() - 1] == '\n')
3072 t = t.substr(0, t.size() - 1);
3074 // Handle all other '\n'
3075 while ((pos = t.find("\n")) != string::npos) {
3076 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3078 if (isAlnumASCII(t[pos+1])) {
3079 t.replace(pos-2, 3, " ");
3082 t.replace(pos-2, 3, "");
3085 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3086 // '\n' adjacent to non-alpha-numerics, discard
3087 t.replace(pos, 1, "");
3090 // Replace all other \n with spaces
3091 t.replace(pos, 1, " ");
3094 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3095 // Kornel: Added textsl, textsf, textit, texttt and noun
3096 // + allow to seach for colored text too
3097 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3098 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3099 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3100 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3101 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3103 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3104 // FIXME - check what preceeds the brace
3106 if (opt.ignoreformat)
3107 while (regex_replace(t, t, "\\{", "_x_<")
3108 || regex_replace(t, t, "\\}", "_x_>"))
3109 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3111 while (regex_replace(t, t, "\\\\\\{", "_x_<")
3112 || regex_replace(t, t, "\\\\\\}", "_x_>"))
3113 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3120 docstring stringifyFromCursor(DocIterator const & cur, int len)
3122 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3123 if (cur.inTexted()) {
3124 Paragraph const & par = cur.paragraph();
3125 // TODO what about searching beyond/across paragraph breaks ?
3126 // TODO Try adding a AS_STR_INSERTS as last arg
3127 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3128 int(par.size()) : cur.pos() + len;
3129 // OutputParams runparams(&cur.buffer()->params().encoding());
3130 OutputParams runparams(encodings.fromLyXName("utf8"));
3131 runparams.nice = true;
3132 runparams.flavor = OutputParams::XETEX;
3133 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3134 // No side effect of file copying and image conversion
3135 runparams.dryrun = true;
3136 runparams.for_search = true;
3137 LYXERR(Debug::FIND, "Stringifying with cur: "
3138 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3139 return par.asString(cur.pos(), end,
3140 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3142 } else if (cur.inMathed()) {
3143 CursorSlice cs = cur.top();
3144 MathData md = cs.cell();
3145 MathData::const_iterator it_end =
3146 (( len == -1 || cs.pos() + len > int(md.size()))
3148 : md.begin() + cs.pos() + len );
3150 for (MathData::const_iterator it = md.begin() + cs.pos();
3153 docstring s = asString(md2);
3154 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3157 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3162 /** Computes the LaTeX export of buf starting from cur and ending len positions
3163 * after cur, if len is positive, or at the paragraph or innermost inset end
3166 docstring latexifyFromCursor(DocIterator const & cur, int len)
3168 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3169 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3170 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3171 Buffer const & buf = *cur.buffer();
3173 odocstringstream ods;
3175 //OutputParams runparams(&buf.params().encoding());
3176 OutputParams runparams(encodings.fromLyXName("utf8"));
3177 runparams.nice = false;
3178 runparams.flavor = OutputParams::XETEX;
3179 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3180 // No side effect of file copying and image conversion
3181 runparams.dryrun = true;
3182 runparams.for_search = true;
3184 if (cur.inTexted()) {
3185 // @TODO what about searching beyond/across paragraph breaks ?
3186 pos_type endpos = cur.paragraph().size();
3187 if (len != -1 && endpos > cur.pos() + len)
3188 endpos = cur.pos() + len;
3189 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3190 string(), cur.pos(), endpos);
3191 string s = lyx::to_utf8(ods.str());
3192 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3193 return(lyx::from_utf8(s));
3194 } else if (cur.inMathed()) {
3195 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3196 for (int s = cur.depth() - 1; s >= 0; --s) {
3197 CursorSlice const & cs = cur[s];
3198 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3200 cs.asInsetMath()->asHullInset()->header_write(ws);
3205 CursorSlice const & cs = cur.top();
3206 MathData md = cs.cell();
3207 MathData::const_iterator it_end =
3208 ((len == -1 || cs.pos() + len > int(md.size()))
3210 : md.begin() + cs.pos() + len);
3212 for (MathData::const_iterator it = md.begin() + cs.pos();
3216 ods << asString(md2);
3217 // Retrieve the math environment type, and add '$' or '$]'
3218 // or others (\end{equation}) accordingly
3219 for (int s = cur.depth() - 1; s >= 0; --s) {
3220 CursorSlice const & cs2 = cur[s];
3221 InsetMath * inset = cs2.asInsetMath();
3222 if (inset && inset->asHullInset()) {
3224 inset->asHullInset()->footer_write(ws);
3228 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3230 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3236 /** Finalize an advanced find operation, advancing the cursor to the innermost
3237 ** position that matches, plus computing the length of the matching text to
3240 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
3242 // Search the foremost position that matches (avoids find of entire math
3243 // inset when match at start of it)
3245 DocIterator old_cur(cur.buffer());
3247 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3251 } while (cur && cur.depth() > d && match(cur).match_len > 0);
3253 int max_match = match(cur).match_len; /* match valid only if not searching whole words */
3254 if (max_match <= 0) return 0;
3255 LYXERR(Debug::FIND, "Ok");
3257 // Compute the match length
3259 if (cur.pos() + len > cur.lastpos())
3261 if (match.opt.matchword) {
3262 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3263 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3265 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3267 // Length of matched text (different from len param)
3268 int old_match = match(cur, len).match_len;
3272 // Greedy behaviour while matching regexps
3273 while ((new_match = match(cur, len + 1).match_len) > old_match) {
3275 old_match = new_match;
3276 LYXERR(Debug::FIND, "verifying match with len = " << len);
3283 int maxl = cur.lastpos() - cur.pos();
3284 // Greedy behaviour while matching regexps
3285 while (maxl > minl) {
3286 int actual_match = match(cur, len).match_len;
3287 if (actual_match >= max_match) {
3288 // actual_match > max_match _can_ happen,
3289 // if the search area splits
3290 // some following word so that the regex
3291 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3293 // This means, the len value is too big
3295 len = (int)((maxl + minl)/2);
3298 // (actual_match < max_match)
3300 len = (int)((maxl + minl)/2);
3304 // Search for real start of matched characters
3309 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3310 if (cur.depth() < old_cur.depth()) {
3312 LYXERR0("cur.depth() < old_cur.depth(), this should never happen");
3315 if (cur.pos() != old_cur.pos()) {
3316 // OK, forwarded 1 pos in actual inset
3317 actual_match = match(cur, len-1).match_len;
3318 if (actual_match == max_match) {
3319 // Ha, got it! The shorter selection has the same match length
3324 // OK, the shorter selection matches less chars, revert to previous value
3330 LYXERR0("cur.pos() == old_cur.pos(), this should never happen");
3331 actual_match = match(cur, len).match_len;
3332 if (actual_match == max_match)
3342 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3346 while (!theApp()->longOperationCancelled() && cur) {
3347 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3348 MatchResult mres = match(cur, -1, false);
3349 int match_len = mres.match_len;
3350 LYXERR(Debug::FIND, "match_len: " << match_len);
3351 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3352 LYXERR0("BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3355 if (match_len > 0) {
3356 // Try to find the begin of searched string
3357 int increment = mres.pos/2;
3358 while (mres.pos > 5 && (increment > 5)) {
3359 DocIterator old_cur = cur;
3360 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3362 if (! cur || (cur.pit() > old_cur.pit())) {
3363 // Are we outside of the paragraph?
3364 // This can happen if moving past some UTF8-encoded chars
3369 MatchResult mres2 = match(cur, -1, false);
3370 if ((mres2.match2end < mres.match2end) ||
3371 (mres2.match_len < mres.match_len)) {
3378 if (increment > mres.pos/2)
3379 increment = mres.pos/2;
3383 int match_len_zero_count = 0;
3384 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3386 int remaining_len = match(cur, -1, false).match_len;
3387 if (remaining_len <= 0) {
3388 // Apparently the searched string is not in the remaining part
3395 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3396 int match_len3 = match(cur, 1).match_len;
3399 int match_len2 = match(cur).match_len;
3400 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3401 if (match_len2 > 0) {
3402 // Sometimes in finalize we understand it wasn't a match
3403 // and we need to continue the outest loop
3404 int len = findAdvFinalize(cur, match);
3409 if (match_len2 >= 0) {
3410 if (match_len2 == 0)
3411 match_len_zero_count++;
3413 match_len_zero_count = 0;
3416 if (++match_len_zero_count > 3) {
3417 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3418 match_len_zero_count = 0;
3426 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3427 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3430 // This should exit nested insets, if any, or otherwise undefine the currsor.
3431 cur.pos() = cur.lastpos();
3432 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3440 /// Find the most backward consecutive match within same paragraph while searching backwards.
3441 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3443 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3444 DocIterator tmp_cur = cur;
3445 int len = findAdvFinalize(tmp_cur, match);
3446 Inset & inset = cur.inset();
3447 for (; cur != cur_begin; cur.backwardPos()) {
3448 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3449 DocIterator new_cur = cur;
3450 new_cur.backwardPos();
3451 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3453 int new_len = findAdvFinalize(new_cur, match);
3458 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3464 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3468 // Backup of original position
3469 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3470 if (cur == cur_begin)
3473 DocIterator cur_orig(cur);
3474 bool pit_changed = false;
3477 bool found_match = (match(cur, -1, false).match_len > 0);
3481 cur.pos() = cur.lastpos();
3483 cur.pos() = cur_orig.pos();
3484 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3485 DocIterator cur_prev_iter;
3487 found_match = (match(cur).match_len > 0);
3488 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3489 << found_match << ", cur: " << cur);
3491 return findMostBackwards(cur, match);
3493 // Stop if begin of document reached
3494 if (cur == cur_begin)
3496 cur_prev_iter = cur;
3500 if (cur == cur_begin)
3507 } while (!theApp()->longOperationCancelled());
3515 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3516 DocIterator const & cur, int len)
3518 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3520 if (!opt.ignoreformat)
3521 return latexifyFromCursor(cur, len);
3523 return stringifyFromCursor(cur, len);
3527 FindAndReplaceOptions::FindAndReplaceOptions(
3528 docstring const & _find_buf_name, bool _casesensitive,
3529 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3530 docstring const & _repl_buf_name, bool _keep_case,
3531 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3532 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3533 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3534 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3542 /** Check if 'len' letters following cursor are all non-lowercase */
3543 static bool allNonLowercase(Cursor const & cur, int len)
3545 pos_type beg_pos = cur.selectionBegin().pos();
3546 pos_type end_pos = cur.selectionBegin().pos() + len;
3547 if (len > cur.lastpos() + 1 - beg_pos) {
3548 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3549 len = cur.lastpos() + 1 - beg_pos;
3550 end_pos = beg_pos + len;
3552 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3553 if (isLowerCase(cur.paragraph().getChar(pos)))
3559 /** Check if first letter is upper case and second one is lower case */
3560 static bool firstUppercase(Cursor const & cur)
3563 pos_type pos = cur.selectionBegin().pos();
3564 if (pos >= cur.lastpos() - 1) {
3565 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3568 ch1 = cur.paragraph().getChar(pos);
3569 ch2 = cur.paragraph().getChar(pos + 1);
3570 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3571 LYXERR(Debug::FIND, "firstUppercase(): "
3572 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3573 << ch2 << "(" << char(ch2) << ")"
3574 << ", result=" << result << ", cur=" << cur);
3579 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3581 ** \fixme What to do with possible further paragraphs in replace buffer ?
3583 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3585 ParagraphList::iterator pit = buffer.paragraphs().begin();
3586 LASSERT(pit->size() >= 1, /**/);
3587 pos_type right = pos_type(1);
3588 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3589 right = pit->size();
3590 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3596 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3598 Cursor & cur = bv->cursor();
3599 if (opt.repl_buf_name == docstring()
3600 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3601 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3604 DocIterator sel_beg = cur.selectionBegin();
3605 DocIterator sel_end = cur.selectionEnd();
3606 if (&sel_beg.inset() != &sel_end.inset()
3607 || sel_beg.pit() != sel_end.pit()
3608 || sel_beg.idx() != sel_end.idx())
3610 int sel_len = sel_end.pos() - sel_beg.pos();
3611 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3612 << ", sel_len: " << sel_len << endl);
3615 LASSERT(sel_len > 0, return 0);
3617 if (!matchAdv(sel_beg, sel_len).match_len)
3620 // Build a copy of the replace buffer, adapted to the KeepCase option
3621 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3623 repl_buffer_orig.write(oss);
3624 string lyx = oss.str();
3625 Buffer repl_buffer("", false);
3626 repl_buffer.setUnnamed(true);
3627 LASSERT(repl_buffer.readString(lyx), return 0);
3628 if (opt.keep_case && sel_len >= 2) {
3629 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3630 if (cur.inTexted()) {
3631 if (firstUppercase(cur))
3632 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3633 else if (allNonLowercase(cur, sel_len))
3634 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3637 cap::cutSelection(cur, false);
3638 if (cur.inTexted()) {
3639 repl_buffer.changeLanguage(
3640 repl_buffer.language(),
3641 cur.getFont().language());
3642 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3643 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3644 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3645 repl_buffer.params().documentClassPtr(),
3646 bv->buffer().errorList("Paste"));
3647 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3648 sel_len = repl_buffer.paragraphs().begin()->size();
3649 } else if (cur.inMathed()) {
3650 odocstringstream ods;
3652 // OutputParams runparams(&repl_buffer.params().encoding());
3653 OutputParams runparams(encodings.fromLyXName("utf8"));
3654 runparams.nice = false;
3655 runparams.flavor = OutputParams::XETEX;
3656 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3657 runparams.dryrun = true;
3658 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3659 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3660 docstring repl_latex = ods.str();
3661 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3663 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3664 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3665 repl_latex = from_utf8(s);
3666 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3667 MathData ar(cur.buffer());
3668 asArray(repl_latex, ar, Parse::NORMAL);
3670 sel_len = ar.size();
3671 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3673 if (cur.pos() >= sel_len)
3674 cur.pos() -= sel_len;
3677 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3678 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3679 bv->processUpdateFlags(Update::Force);
3684 /// Perform a FindAdv operation.
3685 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3690 // e.g., when invoking word-findadv from mini-buffer wither with
3691 // wrong options syntax or before ever opening advanced F&R pane
3692 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3696 MatchStringAdv matchAdv(bv->buffer(), opt);
3697 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3699 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3700 num_replaced += findAdvReplace(bv, opt, matchAdv);
3703 match_len = findForwardAdv(cur, matchAdv);
3705 match_len = findBackwardsAdv(cur, matchAdv);
3707 // This may only be raised by lyx::regex()
3708 bv->message(_("Invalid regular expression!"));
3712 if (match_len == 0) {
3713 if (num_replaced > 0) {
3714 switch (num_replaced)
3717 bv->message(_("One match has been replaced."));
3720 bv->message(_("Two matches have been replaced."));
3723 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
3729 bv->message(_("Match not found."));
3734 if (num_replaced > 0)
3735 bv->message(_("Match has been replaced."));
3737 bv->message(_("Match found."));
3739 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3740 bv->putSelectionAt(cur, match_len, !opt.forward);
3746 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3748 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3749 << opt.casesensitive << ' '
3750 << opt.matchword << ' '
3751 << opt.forward << ' '
3752 << opt.expandmacros << ' '
3753 << opt.ignoreformat << ' '
3754 << opt.replace_all << ' '
3755 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3756 << opt.keep_case << ' '
3757 << int(opt.scope) << ' '
3760 LYXERR(Debug::FIND, "built: " << os.str());
3766 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3768 LYXERR(Debug::FIND, "parsing");
3772 while (line != "EOSS") {
3776 if (is.eof()) // Tolerate malformed request
3780 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3781 opt.find_buf_name = from_utf8(s);
3782 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
3783 is.get(); // Waste space before replace string
3786 while (line != "EOSS") {
3790 if (is.eof()) // Tolerate malformed request
3794 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3795 opt.repl_buf_name = from_utf8(s);
3796 is >> opt.keep_case;
3799 opt.scope = FindAndReplaceOptions::SearchScope(i);
3801 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3803 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3804 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3805 << opt.scope << ' ' << opt.restr);