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 type, bool value);
106 bool ignoreUnderline_;
110 bool ignoreStrikeOut_;
112 bool ignoreSectioning_;
114 bool ignoreFrontMatter_;
118 bool ignoreLanguage_;
122 void IgnoreFormats::setIgnoreFormat(string 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 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 & str, bool cs, bool mw)
185 : str(str), 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) :
1117 isPatternString(isPattern),
1119 actualdeptindex(0) { depts[0] = 0; closes[0] = 0;};
1122 int depts[MAXOPENED];
1123 int closes[MAXOPENED];
1124 int actualdeptindex;
1125 Border borders[2*MAXOPENED];
1126 int previousNotIgnored(int);
1127 int nextNotIgnored(int);
1128 void handleOpenP(int i);
1129 void handleCloseP(int i, bool closingAllowed);
1130 void resetOpenedP(int openPos);
1131 void addIntervall(int upper);
1132 void addIntervall(int low, int upper); /* if explicit */
1133 void removeAccents();
1134 void setForDefaultLang(KeyInfo &defLang);
1135 int findclosing(int start, int end, char up, char down, int repeat);
1136 void handleParentheses(int lastpos, bool closingAllowed);
1138 int isOpeningPar(int pos);
1140 void output(ostringstream &os, int lastpos);
1141 // string show(int lastpos);
1144 int Intervall::isOpeningPar(int pos)
1146 if ((pos < 0) || (size_t(pos) >= par.size()))
1148 if (par[pos] != '{')
1150 if (size_t(pos) + 2 >= par.size())
1152 if (par[pos+2] != '}')
1154 if (par[pos+1] == '[' || par[pos+1] == ']')
1159 void Intervall::setForDefaultLang(KeyInfo &defLang)
1161 // Enable the use of first token again
1162 if (ignoreidx >= 0) {
1163 int value = defLang._tokenstart + defLang._tokensize;
1165 if (borders[0].low < value)
1166 borders[0].low = value;
1167 if (borders[0].upper < value)
1168 borders[0].upper = value;
1173 static void checkDepthIndex(int val)
1175 static int maxdepthidx = MAXOPENED-2;
1176 if (val > maxdepthidx) {
1178 LYXERR0("maxdepthidx now " << val);
1182 static void checkIgnoreIdx(int val)
1184 static int maxignoreidx = 2*MAXOPENED - 4;
1185 if (val > maxignoreidx) {
1187 LYXERR0("maxignoreidx now " << val);
1192 * Expand the region of ignored parts of the input latex string
1193 * The region is only relevant in output()
1195 void Intervall::addIntervall(int low, int upper)
1198 if (low == upper) return;
1199 for (idx = ignoreidx+1; idx > 0; --idx) {
1200 if (low > borders[idx-1].upper) {
1204 Border br(low, upper);
1205 if (idx > ignoreidx) {
1208 checkIgnoreIdx(ignoreidx);
1212 // Expand only if one of the new bound is inside the interwall
1213 // We know here that br.low > borders[idx-1].upper
1214 if (br.upper < borders[idx].low) {
1215 // We have to insert at this pos
1216 for (int i = ignoreidx+1; i > idx; --i) {
1217 borders[i] = borders[i-1];
1221 checkIgnoreIdx(ignoreidx);
1224 // Here we know, that we are overlapping
1225 if (br.low > borders[idx].low)
1226 br.low = borders[idx].low;
1227 // check what has to be concatenated
1229 for (int i = idx; i <= ignoreidx; i++) {
1230 if (br.upper >= borders[i].low) {
1232 if (br.upper < borders[i].upper)
1233 br.upper = borders[i].upper;
1239 // count should be >= 1 here
1242 for (int i = idx + count; i <= ignoreidx; i++) {
1243 borders[i-count+1] = borders[i];
1245 ignoreidx -= count - 1;
1251 typedef map<string, string> AccentsMap;
1252 static AccentsMap accents = map<string, string>();
1254 static void buildaccent(string n, string param, string values)
1258 const char delim = '|';
1259 while (getline(s, name, delim)) {
1261 for (size_t i = 0; i < param.size(); i++) {
1262 string key = name + "{" + param[i] + "}";
1263 // get the corresponding utf8-value
1264 if ((values[start] & 0xc0) != 0xc0) {
1265 // should not happen, utf8 encoding starts at least with 11xxxxxx
1266 // but value for '\dot{i}' is 'i', which is ascii
1267 if ((values[start] & 0x80) == 0) {
1269 accents[key] = values.substr(start, 1);
1274 for (int j = 1; ;j++) {
1275 if (start + j >= values.size()) {
1276 accents[key] = values.substr(start, j);
1277 start = values.size() - 1;
1280 else if ((values[start+j] & 0xc0) != 0x80) {
1281 // This is the first byte of following utf8 char
1282 accents[key] = values.substr(start, j);
1291 static void buildAccentsMap()
1293 accents["imath"] = "ı";
1295 accents["jmath"] = "ȷ";
1296 accents["lyxmathsym{ß}"] = "ß";
1297 accents["text{ß}"] = "ß";
1298 accents["ddot{\\imath}"] = "ï";
1299 buildaccent("ddot", "aAeEiIioOuUyY",
1300 "äÄëËïÏïöÖüÜÿŸ"); // umlaut
1301 buildaccent("dot|.", "cCeEGgIizZaAoObBdDfFyY",
1302 "ċĊėĖĠġİİżŻȧȦȯȮḃḂḋḊḟḞẏẎ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1303 accents["acute{\\imath}"] = "í";
1304 buildaccent("acute", "aAcCeElLoOnNrRsSuUyYzZiI",
1305 "áÁćĆéÉĺĹóÓńŃŕŔśŚúÚýÝźŹíÍ");
1306 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1307 buildaccent("mathring|r", "aAuUwy",
1309 accents["check{\\imath}"] = "ǐ";
1310 accents["check{\\jmath}"] = "ǰ";
1311 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1312 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1313 accents["hat{\\imath}"] = "î";
1314 accents["hat{\\jmath}"] = "ĵ";
1315 buildaccent("hat|^", "aAeEiIcCgGhHjJsSwWyYzZoOuU",
1316 "âÂêÊîÎĉĈĝĜĥĤĵĴŝŜŵŴŷŶẑẐôÔûÛ"); // circ
1317 accents["bar{\\imath}"] = "ī";
1318 buildaccent("bar|=", "aAeEiIoOuUyY",
1319 "āĀēĒīĪōŌūŪȳȲ"); // macron
1320 accents["tilde{\\imath}"] = "ĩ";
1321 buildaccent("tilde", "aAnNoOiIuU",
1322 "ãÃñÑõÕĩĨũŨ"); // tilde
1323 accents["breve{\\imath}"] = "ĭ";
1324 buildaccent("breve|u", "aAeEgGiIoOuU",
1325 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1326 accents["grave{\\imath}"] = "ì";
1327 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1328 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1329 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1330 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1331 buildaccent("ogonek|k", "AaEeIiUuOo",
1332 "ĄąĘęĮįŲųǪǫ"); // ogonek
1333 buildaccent("cedilla|c", "CcGKkLlNnRrSsTtEeDdHh",
1334 "ÇçĢĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1335 buildaccent("subring|textsubring", "Aa",
1337 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1338 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1339 buildaccent("subtilde|textsubtilde", "EeIiUu",
1340 "ḚḛḬḭṴṵ"); // subtilde
1344 * Created accents in math or regexp environment
1345 * are macros, but we need the utf8 equivalent
1347 void Intervall::removeAccents()
1349 if (accents.empty())
1351 static regex const accre("\\\\((.|grave|breve|lyxmathsym|text|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde)\\{[^\\{\\}]+\\}|(i|imath|jmath)(?![a-zA-Z]))");
1353 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1355 string key = sub.str(1);
1356 if (accents.find(key) != accents.end()) {
1357 string val = accents[key];
1358 size_t pos = sub.position(size_t(0));
1359 for (size_t i = 0; i < val.size(); i++) {
1360 par[pos+i] = val[i];
1362 addIntervall(pos+val.size(), pos + sub.str(0).size());
1363 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1364 // remove traces of any remaining chars
1369 LYXERR0("Not added accent for \"" << key << "\"");
1374 void Intervall::handleOpenP(int i)
1377 depts[actualdeptindex] = i+1;
1378 closes[actualdeptindex] = -1;
1379 checkDepthIndex(actualdeptindex);
1382 void Intervall::handleCloseP(int i, bool closingAllowed)
1384 if (actualdeptindex <= 0) {
1385 if (! closingAllowed)
1386 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1387 // if we are at the very end
1388 addIntervall(i, i+1);
1391 closes[actualdeptindex] = i+1;
1396 void Intervall::resetOpenedP(int openPos)
1398 // Used as initializer for foreignlanguage entry
1399 actualdeptindex = 1;
1400 depts[1] = openPos+1;
1404 int Intervall::previousNotIgnored(int start)
1406 int idx = 0; /* int intervalls */
1407 for (idx = ignoreidx; idx >= 0; --idx) {
1408 if (start > borders[idx].upper)
1410 if (start >= borders[idx].low)
1411 start = borders[idx].low-1;
1416 int Intervall::nextNotIgnored(int start)
1418 int idx = 0; /* int intervalls */
1419 for (idx = 0; idx <= ignoreidx; idx++) {
1420 if (start < borders[idx].low)
1422 if (start < borders[idx].upper)
1423 start = borders[idx].upper;
1428 typedef map<string, KeyInfo> KeysMap;
1429 typedef vector< KeyInfo> Entries;
1430 static KeysMap keys = map<string, KeyInfo>();
1437 void buildKeys(bool);
1438 void buildEntries(bool);
1439 void makeKey(const string &, KeyInfo, bool isPatternString);
1440 void processRegion(int start, int region_end); /* remove {} parts */
1441 void removeHead(KeyInfo&, int count=0);
1444 LatexInfo(string par, bool isPatternString) : entidx(-1), interval(isPatternString) {
1446 interval.hasTitle = false;
1447 interval.titleValue = "";
1448 buildKeys(isPatternString);
1449 entries = vector<KeyInfo>();
1450 buildEntries(isPatternString);
1454 if (entries.empty()) {
1457 if (entries[0].keytype == KeyInfo::isTitle) {
1458 if (! entries[0].disabled) {
1459 interval.hasTitle = true;
1460 interval.titleValue = entries[0].head;
1463 interval.hasTitle = false;
1464 interval.titleValue = "";
1466 removeHead(entries[0]);
1467 if (entries.size() > 1)
1476 if (int(entries.size()) > entidx) {
1483 bool setNextKey(int idx) {
1484 if ((idx == entidx) && (entidx >= 0)) {
1491 int find(int start, KeyInfo::KeyType keytype) {
1495 while (tmpIdx < int(entries.size())) {
1496 if (entries[tmpIdx].keytype == keytype)
1502 int process(ostringstream &os, KeyInfo &actual);
1503 int dispatch(ostringstream &os, int previousStart, KeyInfo &actual);
1504 // string show(int lastpos) { return interval.show(lastpos);};
1505 int nextNotIgnored(int start) { return interval.nextNotIgnored(start);};
1506 KeyInfo &getKeyInfo(int keyinfo) {
1507 static KeyInfo invalidInfo = KeyInfo();
1508 if ((keyinfo < 0) || ( keyinfo >= int(entries.size())))
1511 return entries[keyinfo];
1513 void setForDefaultLang(KeyInfo &defLang) {interval.setForDefaultLang(defLang);};
1514 void addIntervall(int low, int up) { interval.addIntervall(low, up); };
1518 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1523 for (int i = start; i < end; i += 1 + skip) {
1527 if (c == '\\') skip = 1;
1531 else if (c == down) {
1533 if ((repeat <= 0) || (par[i+1] != up))
1551 vector<MathEntry> entries;
1556 void insert(string wait, size_t start, size_t end) {
1557 MathEntry m = MathEntry();
1559 m.mathStart = start;
1561 m.mathSize = end - start;
1562 entries.push_back(m);
1564 bool empty() { return entries.empty(); };
1565 size_t getEndPos() {
1566 if (entries.empty() || (actualIdx >= entries.size())) {
1569 return entries[actualIdx].mathEnd;
1571 size_t getStartPos() {
1572 if (entries.empty() || (actualIdx >= entries.size())) {
1573 return 100000; /* definitely enough? */
1575 return entries[actualIdx].mathStart;
1577 size_t getFirstPos() {
1579 return getStartPos();
1582 if (entries.empty() || (actualIdx >= entries.size())) {
1585 return entries[actualIdx].mathSize;
1587 void incrEntry() { actualIdx++; };
1590 void LatexInfo::buildEntries(bool isPatternString)
1592 static regex const rmath("\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\}");
1593 static regex const rkeys("\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?))");
1594 static bool disableLanguageOverride = false;
1595 smatch sub, submath;
1596 bool evaluatingRegexp = false;
1598 bool evaluatingMath = false;
1599 bool evaluatingCode = false;
1601 bool evaluatingOptional = false;
1602 size_t optionalEnd = 0;
1605 bool math_end_waiting = false;
1606 size_t math_pos = 10000;
1609 interval.removeAccents();
1611 for (sregex_iterator itmath(interval.par.begin(), interval.par.end(), rmath), end; itmath != end; ++itmath) {
1613 if (math_end_waiting) {
1614 size_t pos = submath.position(size_t(0));
1615 if ((math_end == "$") &&
1616 (submath.str(0) == "$") &&
1617 (interval.par[pos-1] != '\\')) {
1618 mi.insert("$", math_pos, pos + 1);
1619 math_end_waiting = false;
1621 else if ((math_end == "\\]") &&
1622 (submath.str(0) == "\\]")) {
1623 mi.insert("\\]", math_pos, pos + 2);
1624 math_end_waiting = false;
1626 else if ((submath.str(1).compare("end") == 0) &&
1627 (submath.str(2).compare(math_end) == 0)) {
1628 mi.insert(math_end, math_pos, pos + submath.str(0).length());
1629 math_end_waiting = false;
1635 if (submath.str(1).compare("begin") == 0) {
1636 math_end_waiting = true;
1637 math_end = submath.str(2);
1638 math_pos = submath.position(size_t(0));
1640 else if (submath.str(0).compare("\\[") == 0) {
1641 math_end_waiting = true;
1643 math_pos = submath.position(size_t(0));
1645 else if (submath.str(0) == "$") {
1646 size_t pos = submath.position(size_t(0));
1647 if ((pos == 0) || (interval.par[pos-1] != '\\')) {
1648 math_end_waiting = true;
1655 // Ignore language if there is math somewhere in pattern-string
1656 if (isPatternString) {
1659 keys["foreignlanguage"].disabled = true;
1660 disableLanguageOverride = true;
1663 disableLanguageOverride = false;
1666 if (disableLanguageOverride) {
1667 keys["foreignlanguage"].disabled = true;
1670 math_pos = mi.getFirstPos();
1671 for (sregex_iterator it(interval.par.begin(), interval.par.end(), rkeys), end; it != end; ++it) {
1673 string key = sub.str(3);
1675 if (sub.str(0)[0] == '\\')
1676 key = sub.str(0)[1];
1680 size_t k_pos = sub.position(size_t(0));
1681 if ((k_pos > 0) && (interval.par[k_pos - 1] == '\\')) {
1682 // Escaped '$', ignoring
1688 if (evaluatingRegexp) {
1689 if (sub.str(1).compare("endregexp") == 0) {
1690 evaluatingRegexp = false;
1691 // found._tokenstart already set
1692 found._dataEnd = sub.position(size_t(0)) + 13;
1693 found._dataStart = found._dataEnd;
1694 found._tokensize = found._dataEnd - found._tokenstart;
1695 found.parenthesiscount = 0;
1696 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1703 if (evaluatingMath) {
1704 if (size_t(sub.position(size_t(0))) < mi.getEndPos())
1706 evaluatingMath = false;
1708 math_pos = mi.getStartPos();
1710 if (keys.find(key) == keys.end()) {
1711 found = KeyInfo(KeyInfo::isStandard, 0, true);
1712 if (isPatternString) {
1713 found.keytype = KeyInfo::isChar;
1714 found.disabled = false;
1721 if (key.compare("regexp") == 0) {
1722 evaluatingRegexp = true;
1723 found._tokenstart = sub.position(size_t(0));
1724 found._tokensize = 0;
1728 // Handle the other params of key
1729 if (found.keytype == KeyInfo::isIgnored)
1731 else if (found.keytype == KeyInfo::isMath) {
1732 if (size_t(sub.position(size_t(0))) == math_pos) {
1734 found._tokenstart = sub.position(size_t(0));
1735 found._tokensize = mi.getSize();
1736 found._dataEnd = found._tokenstart + found._tokensize;
1737 found._dataStart = found._dataEnd;
1738 found.parenthesiscount = 0;
1739 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1740 evaluatingMath = true;
1743 // begin|end of unknown env, discard
1744 // First handle tables
1745 // longtable|tabular
1746 bool discardComment;
1748 found.keytype = KeyInfo::doRemove;
1749 if ((sub.str(5).compare("longtable") == 0) ||
1750 (sub.str(5).compare("tabular") == 0)) {
1751 discardComment = true; /* '%' */
1754 discardComment = false;
1755 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1757 string token = sub.str(5);
1758 if (regex_match(token, sub2, removeArgs)) {
1759 found.keytype = KeyInfo::removeWithArg;
1762 // discard spaces before pos(0)
1763 int pos = sub.position(size_t(0));
1765 for (count = 0; pos - count > 0; count++) {
1766 char c = interval.par[pos-count-1];
1767 if (discardComment) {
1768 if ((c != ' ') && (c != '%'))
1774 found._tokenstart = pos - count;
1775 if (sub.str(1).compare(0, 5, "begin") == 0) {
1776 size_t pos1 = pos + sub.str(0).length();
1777 if (sub.str(5).compare("cjk") == 0) {
1778 pos1 = interval.findclosing(pos1+1, interval.par.length()) + 1;
1779 if ((interval.par[pos1] == '{') && (interval.par[pos1+1] == '}'))
1781 found.keytype = KeyInfo::isMain;
1782 found._dataStart = pos1;
1783 found._dataEnd = interval.par.length();
1784 found.disabled = keys["foreignlanguage"].disabled;
1785 found.used = keys["foreignlanguage"].used;
1786 found._tokensize = pos1 - found._tokenstart;
1787 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1790 // Swallow possible optional params
1791 while (interval.par[pos1] == '[') {
1792 pos1 = interval.findclosing(pos1+1, interval.par.length(), '[', ']')+1;
1794 // Swallow also the eventual parameter
1795 if (interval.par[pos1] == '{') {
1796 found._dataEnd = interval.findclosing(pos1+1, interval.par.length()) + 1;
1799 found._dataEnd = pos1;
1801 found._dataStart = found._dataEnd;
1802 found._tokensize = count + found._dataEnd - pos;
1803 found.parenthesiscount = 0;
1804 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1805 found.disabled = true;
1809 // Handle "\end{...}"
1810 found._dataStart = pos + sub.str(0).length();
1811 found._dataEnd = found._dataStart;
1812 found._tokensize = count + found._dataEnd - pos;
1813 found.parenthesiscount = 0;
1814 found.head = interval.par.substr(found._tokenstart, found._tokensize);
1815 found.disabled = true;
1819 else if (found.keytype != KeyInfo::isRegex) {
1820 found._tokenstart = sub.position(size_t(0));
1821 if (found.parenthesiscount == 0) {
1822 // Probably to be discarded
1823 size_t following_pos = sub.position(size_t(0)) + sub.str(3).length() + 1;
1824 char following = interval.par[following_pos];
1825 if (following == ' ')
1826 found.head = "\\" + sub.str(3) + " ";
1827 else if (following == '=') {
1828 // like \uldepth=1000pt
1829 found.head = sub.str(0);
1832 found.head = "\\" + key;
1833 found._tokensize = found.head.length();
1834 found._dataEnd = found._tokenstart + found._tokensize;
1835 found._dataStart = found._dataEnd;
1838 int params = found._tokenstart + key.length() + 1;
1839 if (evaluatingOptional) {
1840 if (size_t(found._tokenstart) > optionalEnd) {
1841 evaluatingOptional = false;
1844 found.disabled = true;
1847 int optend = params;
1848 while (interval.par[optend] == '[') {
1849 // discard optional parameters
1850 optend = interval.findclosing(optend+1, interval.par.length(), '[', ']') + 1;
1852 if (optend > params) {
1853 key += interval.par.substr(params, optend-params);
1854 evaluatingOptional = true;
1855 optionalEnd = optend;
1857 string token = sub.str(5);
1858 int closings = found.parenthesiscount;
1859 if (found.parenthesiscount == 1) {
1860 found.head = "\\" + key + "{";
1862 else if (found.parenthesiscount > 1) {
1864 found.head = sub.str(0) + "{";
1865 closings = found.parenthesiscount - 1;
1868 found.head = "\\" + key + "{";
1871 found._tokensize = found.head.length();
1872 found._dataStart = found._tokenstart + found.head.length();
1873 if (interval.par.substr(found._dataStart-1, 15).compare("\\endarguments{}") == 0) {
1874 found._dataStart += 15;
1876 size_t endpos = interval.findclosing(found._dataStart, interval.par.length(), '{', '}', closings);
1877 if (found.keytype == KeyInfo::isList) {
1878 // Check if it really is list env
1879 static regex const listre("^([a-z]+)$");
1881 if (!regex_match(token, sub2, listre)) {
1882 // Change the key of this entry. It is not in a list/item environment
1883 found.keytype = KeyInfo::endArguments;
1886 if (found.keytype == KeyInfo::noMain) {
1887 evaluatingCode = true;
1889 codeStart = found._dataStart;
1891 else if (evaluatingCode) {
1892 if (size_t(found._dataStart) > codeEnd)
1893 evaluatingCode = false;
1894 else if (found.keytype == KeyInfo::isMain) {
1895 // Disable this key, treate it as standard
1896 found.keytype = KeyInfo::isStandard;
1897 found.disabled = true;
1898 if ((codeEnd == interval.par.length()) &&
1899 (found._tokenstart == codeStart)) {
1900 // trickery, because the code inset starts
1901 // with \selectlanguage ...
1903 if (entries.size() > 1) {
1904 entries[entries.size()-1]._dataEnd = codeEnd;
1909 if ((endpos == interval.par.length()) &&
1910 (found.keytype == KeyInfo::doRemove)) {
1911 // Missing closing => error in latex-input?
1912 // therefore do not delete remaining data
1913 found._dataStart -= 1;
1914 found._dataEnd = found._dataStart;
1917 found._dataEnd = endpos;
1919 if (isPatternString) {
1920 keys[key].used = true;
1923 entries.push_back(found);
1927 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
1929 stringstream s(keysstring);
1931 const char delim = '|';
1932 while (getline(s, key, delim)) {
1933 KeyInfo keyII(keyI);
1934 if (isPatternString) {
1937 else if ( !keys[key].used)
1938 keyII.disabled = true;
1943 void LatexInfo::buildKeys(bool isPatternString)
1946 static bool keysBuilt = false;
1947 if (keysBuilt && !isPatternString) return;
1949 // Known standard keys with 1 parameter.
1950 // Split is done, if not at start of region
1951 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
1952 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
1953 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
1954 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
1955 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
1956 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
1958 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
1959 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1960 makeKey("section*|subsection*|subsubsection*|paragraph*",
1961 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1962 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
1963 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
1965 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
1967 // Split is done, if not at start of region
1968 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
1969 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
1971 // Split is done always.
1972 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
1976 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1977 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1978 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1979 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1981 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1982 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1983 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1985 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1986 // Custom space/skip, remove the content (== length value)
1987 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
1988 // Found in fr/UserGuide.lyx
1989 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1991 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1992 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
1993 // Known macros to remove (including their parameter)
1995 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
1996 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
1997 // handle like standard keys with 1 parameter.
1998 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2000 // Macros to remove, but let the parameter survive
2002 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2004 // Remove language spec from content of these insets
2005 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2007 // Same effect as previous, parameter will survive (because there is no one anyway)
2009 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2010 // Remove table decorations
2011 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2012 // Discard shape-header.
2013 // For footnote or shortcut too, because of lang settings
2014 // and wrong handling if used 'KeyInfo::noMain'
2015 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2016 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2017 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2018 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2019 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2020 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2021 // like ('tiny{}' or '\tiny ' ... )
2022 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2024 // Survives, like known character
2025 makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2026 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2028 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2029 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2030 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2032 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2033 // Remove RTL/LTR marker
2034 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2035 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2036 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2037 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2038 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2039 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2040 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2041 if (isPatternString) {
2042 // Allow the first searched string to rebuild the keys too
2046 // no need to rebuild again
2052 * Keep the list of actual opened parentheses actual
2053 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2055 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2058 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2062 if (c == '\\') skip = 1;
2063 else if (c == '{') {
2066 else if (c == '}') {
2067 handleCloseP(i, closingAllowed);
2073 string Intervall::show(int lastpos)
2075 int idx = 0; /* int intervalls */
2078 for (idx = 0; idx <= ignoreidx; idx++) {
2079 while (i < lastpos) {
2081 if (i <= borders[idx].low) {
2082 if (borders[idx].low > lastpos)
2083 printsize = lastpos - i;
2085 printsize = borders[idx].low - i;
2086 s += par.substr(i, printsize);
2088 if (i >= borders[idx].low)
2089 i = borders[idx].upper;
2092 i = borders[idx].upper;
2098 s += par.substr(i, lastpos-i);
2104 void Intervall::output(ostringstream &os, int lastpos)
2106 // get number of chars to output
2107 int idx = 0; /* int intervalls */
2110 string startTitle = titleValue;
2111 for (idx = 0; idx <= ignoreidx; idx++) {
2113 if (i <= borders[idx].low) {
2115 if (borders[idx].low > lastpos)
2116 printsize = lastpos - i;
2118 printsize = borders[idx].low - i;
2119 if (printsize > 0) {
2120 os << startTitle << par.substr(i, printsize);
2122 printed += printsize;
2125 handleParentheses(i, false);
2126 if (i >= borders[idx].low)
2127 i = borders[idx].upper;
2130 i = borders[idx].upper;
2137 os << startTitle << par.substr(i, lastpos-i);
2138 printed += lastpos-i;
2140 handleParentheses(lastpos, false);
2141 for (int i = actualdeptindex; i > 0; --i) {
2144 if (hasTitle && (printed > 0))
2146 if (! isPatternString)
2148 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2151 void LatexInfo::processRegion(int start, int region_end)
2153 while (start < region_end) { /* Let {[} and {]} survive */
2154 int cnt = interval.isOpeningPar(start);
2156 // Closing is allowed past the region
2157 int closing = interval.findclosing(start+1, interval.par.length());
2158 interval.addIntervall(start, start+1);
2159 interval.addIntervall(closing, closing+1);
2163 start = interval.nextNotIgnored(start+1);
2167 void LatexInfo::removeHead(KeyInfo &actual, int count)
2169 if (actual.parenthesiscount == 0) {
2170 // "{\tiny{} ...}" ==> "{{} ...}"
2171 interval.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2174 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2175 interval.addIntervall(actual._tokenstart - count, actual._dataStart);
2176 interval.addIntervall(actual._dataEnd, actual._dataEnd+1);
2180 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2183 switch (actual.keytype)
2185 case KeyInfo::isTitle: {
2187 nextKeyIdx = getNextKey();
2190 case KeyInfo::cleanToStart: {
2191 actual._dataEnd = actual._dataStart;
2192 nextKeyIdx = getNextKey();
2193 // Search for end of arguments
2194 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2196 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2197 entries[i].disabled = true;
2199 actual._dataEnd = entries[tmpIdx]._dataEnd;
2201 while (interval.par[actual._dataEnd] == ' ')
2203 interval.addIntervall(0, actual._dataEnd+1);
2204 interval.actualdeptindex = 0;
2205 interval.depts[0] = actual._dataEnd+1;
2206 interval.closes[0] = -1;
2209 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2210 if (actual.disabled)
2211 interval.addIntervall(actual._tokenstart, actual._dataEnd);
2213 interval.addIntervall(actual._dataStart, actual._dataEnd);
2216 case KeyInfo::isChar: {
2217 nextKeyIdx = getNextKey();
2220 case KeyInfo::isSize: {
2221 if (actual.disabled || (interval.par[actual._dataStart] != '{') || (interval.par[actual._dataStart-1] == ' ')) {
2222 processRegion(actual._dataEnd, actual._dataEnd+1); /* remove possibly following {} */
2223 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2224 nextKeyIdx = getNextKey();
2226 // Here _dataStart points to '{', so correct it
2227 actual._dataStart += 1;
2228 actual._tokensize += 1;
2229 actual.parenthesiscount = 1;
2230 if (interval.par[actual._dataStart] == '}') {
2231 // Determine the end if used like '{\tiny{}...}'
2232 actual._dataEnd = interval.findclosing(actual._dataStart+1, interval.par.length()) + 1;
2233 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2236 // Determine the end if used like '\tiny{...}'
2237 actual._dataEnd = interval.findclosing(actual._dataStart, interval.par.length()) + 1;
2239 // Split on this key if not at start
2240 int start = interval.nextNotIgnored(previousStart);
2241 if (start < actual._tokenstart) {
2242 interval.output(os, actual._tokenstart);
2243 interval.addIntervall(start, actual._tokenstart);
2245 // discard entry if at end of actual
2246 nextKeyIdx = process(os, actual);
2250 case KeyInfo::endArguments:
2251 // Remove trailing '{}' too
2252 actual._dataStart += 1;
2253 actual._dataEnd += 1;
2254 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2255 nextKeyIdx = getNextKey();
2257 case KeyInfo::noMain:
2259 case KeyInfo::isStandard: {
2260 if (actual.disabled) {
2262 processRegion(actual._dataStart, actual._dataStart+1);
2263 nextKeyIdx = getNextKey();
2265 // Split on this key if not at datastart of calling entry
2266 int start = interval.nextNotIgnored(previousStart);
2267 if (start < actual._tokenstart) {
2268 interval.output(os, actual._tokenstart);
2269 interval.addIntervall(start, actual._tokenstart);
2271 // discard entry if at end of actual
2272 nextKeyIdx = process(os, actual);
2276 case KeyInfo::removeWithArg: {
2277 nextKeyIdx = getNextKey();
2278 // Search for end of arguments
2279 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2281 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2282 entries[i].disabled = true;
2284 actual._dataEnd = entries[tmpIdx]._dataEnd;
2286 interval.addIntervall(actual._tokenstart, actual._dataEnd+1);
2289 case KeyInfo::doRemove: {
2290 // Remove the key with all parameters and following spaces
2292 for (pos = actual._dataEnd+1; pos < interval.par.length(); pos++) {
2293 if ((interval.par[pos] != ' ') && (interval.par[pos] != '%'))
2296 interval.addIntervall(actual._tokenstart, pos);
2297 nextKeyIdx = getNextKey();
2300 case KeyInfo::isList: {
2301 // Discard space before _tokenstart
2303 for (count = 0; count < actual._tokenstart; count++) {
2304 if (interval.par[actual._tokenstart-count-1] != ' ')
2307 nextKeyIdx = getNextKey();
2308 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2310 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2312 // How else can we catch this one?
2313 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2314 entries[i].disabled = true;
2316 actual._dataEnd = entries[tmpIdx]._dataEnd;
2318 else if (nextKeyIdx > 0) {
2319 // Ignore any lang entries inside data region
2320 for (int i = nextKeyIdx; i < int(entries.size()) && entries[i]._tokenstart < actual._dataEnd; i++) {
2321 if (entries[i].keytype == KeyInfo::isMain)
2322 entries[i].disabled = true;
2325 if (actual.disabled) {
2326 interval.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2329 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2331 if (interval.par[actual._dataEnd+1] == '[') {
2332 int posdown = interval.findclosing(actual._dataEnd+2, interval.par.length(), '[', ']');
2333 if ((interval.par[actual._dataEnd+2] == '{') &&
2334 (interval.par[posdown-1] == '}')) {
2335 interval.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2336 interval.addIntervall(posdown-1, posdown+1);
2339 interval.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2340 interval.addIntervall(posdown, posdown+1);
2342 int blk = interval.nextNotIgnored(actual._dataEnd+1);
2343 if (blk > posdown) {
2344 // Discard at most 1 space after empty item
2346 for (count = 0; count < 1; count++) {
2347 if (interval.par[blk+count] != ' ')
2351 interval.addIntervall(blk, blk+count);
2356 case KeyInfo::isSectioning: {
2357 // Discard spaces before _tokenstart
2359 int val = actual._tokenstart;
2360 for (count = 0; count < actual._tokenstart;) {
2361 val = interval.previousNotIgnored(val-1);
2362 if (interval.par[val] != ' ')
2365 count = actual._tokenstart - val;
2368 if (actual.disabled) {
2369 removeHead(actual, count);
2370 nextKeyIdx = getNextKey();
2372 interval.addIntervall(actual._tokenstart-count, actual._tokenstart);
2373 nextKeyIdx = process(os, actual);
2377 case KeyInfo::isMath: {
2378 // Same as regex, use the content unchanged
2379 nextKeyIdx = getNextKey();
2382 case KeyInfo::isRegex: {
2383 // DO NOT SPLIT ON REGEX
2385 nextKeyIdx = getNextKey();
2388 case KeyInfo::isIgnored: {
2389 // Treat like a character for now
2390 nextKeyIdx = getNextKey();
2393 case KeyInfo::isMain: {
2394 if (interval.par.substr(actual._dataStart, 2) == "% ")
2395 interval.addIntervall(actual._dataStart, actual._dataStart+2);
2396 if (actual._tokenstart > 0) {
2397 int prev = interval.previousNotIgnored(actual._tokenstart - 1);
2398 if ((prev >= 0) && interval.par[prev] == '%')
2399 interval.addIntervall(prev, prev+1);
2401 if (actual.disabled) {
2403 if ((interval.par.substr(actual._dataStart, 3) == " \\[") ||
2404 (interval.par.substr(actual._dataStart, 8) == " \\begin{")) {
2405 // Discard also the space before math-equation
2406 interval.addIntervall(actual._dataStart, actual._dataStart+1);
2408 nextKeyIdx = getNextKey();
2409 // interval.resetOpenedP(actual._dataStart-1);
2412 if (actual._tokenstart < 26) {
2413 // for the first (and maybe dummy) language
2414 interval.setForDefaultLang(actual);
2416 interval.resetOpenedP(actual._dataStart-1);
2420 case KeyInfo::invalid:
2421 // This cannot happen, already handled
2424 // LYXERR0("Unhandled keytype");
2425 nextKeyIdx = getNextKey();
2432 int LatexInfo::process(ostringstream &os, KeyInfo &actual )
2434 int end = interval.nextNotIgnored(actual._dataEnd);
2435 int oldStart = actual._dataStart;
2436 int nextKeyIdx = getNextKey();
2438 if ((nextKeyIdx < 0) ||
2439 (entries[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2440 (entries[nextKeyIdx].keytype == KeyInfo::invalid)) {
2441 if (oldStart <= end) {
2442 processRegion(oldStart, end);
2447 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2449 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2450 (void) dispatch(os, actual._dataStart, nextKey);
2451 end = nextKey._tokenstart;
2454 processRegion(oldStart, nextKey._tokenstart);
2455 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2457 oldStart = nextKey._dataEnd+1;
2459 // now nextKey is either invalid or is outside of actual._dataEnd
2460 // output the remaining and discard myself
2461 if (oldStart <= end) {
2462 processRegion(oldStart, end);
2464 if (interval.par[end] == '}') {
2466 // This is the normal case.
2467 // But if using the firstlanguage, the closing may be missing
2469 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2471 if (actual._dataEnd < end)
2472 output_end = interval.nextNotIgnored(actual._dataEnd);
2474 output_end = interval.nextNotIgnored(end);
2475 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2476 interval.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2478 // Remove possible empty data
2479 int dstart = interval.nextNotIgnored(actual._dataStart);
2480 while (interval.isOpeningPar(dstart) == 1) {
2481 interval.addIntervall(dstart, dstart+1);
2482 int dend = interval.findclosing(dstart+1, output_end);
2483 interval.addIntervall(dend, dend+1);
2484 dstart = interval.nextNotIgnored(dstart+1);
2486 if (dstart < output_end)
2487 interval.output(os, output_end);
2488 interval.addIntervall(actual._tokenstart, end);
2492 string splitOnKnownMacros(string par, bool isPatternString)
2495 LatexInfo li(par, isPatternString);
2496 // LYXERR0("Berfore split: " << par);
2497 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2499 DummyKey._tokensize = 0;
2500 DummyKey._dataStart = 0;
2501 DummyKey._dataEnd = par.length();
2502 DummyKey.disabled = true;
2503 int firstkeyIdx = li.getFirstKey();
2505 if (firstkeyIdx >= 0) {
2506 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2507 DummyKey._tokenstart = firstKey._tokenstart;
2509 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2510 // Use dummy firstKey
2511 firstKey = DummyKey;
2512 (void) li.setNextKey(firstkeyIdx);
2515 if (par.substr(firstKey._dataStart, 2) == "% ")
2516 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2518 nextkeyIdx = li.process(os, firstKey);
2519 while (nextkeyIdx >= 0) {
2520 // Check for a possible gap between the last
2521 // entry and this one
2522 int datastart = li.nextNotIgnored(firstKey._dataStart);
2523 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2524 if ((nextKey._tokenstart > datastart)) {
2526 firstKey._dataStart = datastart;
2527 firstKey._dataEnd = par.length();
2528 (void) li.setNextKey(nextkeyIdx);
2529 // Fake the last opened parenthesis
2530 li.setForDefaultLang(firstKey);
2531 nextkeyIdx = li.process(os, firstKey);
2534 if (nextKey.keytype != KeyInfo::isMain) {
2535 firstKey._dataStart = datastart;
2536 firstKey._dataEnd = nextKey._dataEnd+1;
2537 (void) li.setNextKey(nextkeyIdx);
2538 li.setForDefaultLang(firstKey);
2539 nextkeyIdx = li.process(os, firstKey);
2542 nextkeyIdx = li.process(os, nextKey);
2546 // Handle the remaining
2547 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2548 firstKey._dataEnd = par.length();
2550 if ((firstKey._dataStart < firstKey._dataEnd) &&
2551 (par[firstKey._dataStart] != '}')) {
2552 li.setForDefaultLang(firstKey);
2553 (void) li.process(os, firstKey);
2557 // return string definitelly impossible to match
2558 s = "\\foreignlanguage{ignore}{ }";
2562 s = par; /* no known macros found */
2563 // LYXERR0("After split: " << s);
2568 * Try to unify the language specs in the latexified text.
2569 * Resulting modified string is set to "", if
2570 * the searched tex does not contain all the features in the search pattern
2572 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2574 static Features regex_f;
2575 static int missed = 0;
2576 static bool regex_with_format = false;
2578 int parlen = par.length();
2580 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2583 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2584 // Happens to be there in case of description or labeling environment
2589 // Split the latex input into pieces which
2590 // can be digested by our search engine
2591 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2592 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2593 LYXERR(Debug::FIND, "After split: \"" << result << "\"");
2596 result = par.substr(0, parlen);
2597 if (isPatternString) {
2600 regex_f = identifyFeatures(result);
2601 string features = "";
2602 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2603 string a = it->first;
2604 regex_with_format = true;
2605 features += " " + a;
2606 // LYXERR0("Identified regex format:" << a);
2608 LYXERR(Debug::FIND, "Identified Features" << features);
2611 } else if (regex_with_format) {
2612 Features info = identifyFeatures(result);
2613 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2614 string a = it->first;
2615 bool b = it->second;
2616 if (b && ! info[a]) {
2618 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2624 // LYXERR0("No regex formats");
2630 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2631 static int identifyClosing(string & t)
2633 int open_braces = 0;
2635 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2636 if (regex_replace(t, t, "(.*[^\\\\])\\$" REGEX_EOS, "$1"))
2638 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]" REGEX_EOS, "$1"))
2640 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}" REGEX_EOS, "$1"))
2642 if (regex_replace(t, t, "(.*[^\\\\])\\}" REGEX_EOS, "$1")) {
2651 static int num_replaced = 0;
2652 static bool previous_single_replace = true;
2654 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
2655 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2657 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2658 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2659 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2660 if (opt.replace_all && previous_single_replace) {
2661 previous_single_replace = false;
2664 else if (!opt.replace_all) {
2665 num_replaced = 0; // count number of replaced strings
2666 previous_single_replace = true;
2668 // When using regexp, braces are hacked already by escape_for_regex()
2669 par_as_string = normalize(ds, !use_regexp);
2671 close_wildcards = 0;
2673 size_t lead_size = 0;
2674 // correct the language settings
2675 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2676 if (opt.ignoreformat) {
2678 // if par_as_string_nolead were emty,
2679 // the following call to findAux will always *find* the string
2680 // in the checked data, and thus always using the slow
2681 // examining of the current text part.
2682 par_as_string_nolead = par_as_string;
2685 lead_size = identifyLeading(par_as_string);
2686 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2687 lead_as_string = par_as_string.substr(0, lead_size);
2688 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2692 open_braces = identifyClosing(par_as_string);
2693 identifyClosing(par_as_string_nolead);
2694 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2695 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
2697 string lead_as_regexp;
2698 if (lead_size > 0) {
2699 // @todo No need to search for \regexp{} insets in leading material
2700 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
2701 par_as_string = par_as_string_nolead;
2702 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2703 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2705 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2706 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2707 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2709 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
2710 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
2711 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
2712 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
2713 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
2714 || regex_replace(par_as_string, par_as_string,
2715 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
2716 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
2717 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
2721 if (!opt.ignoreformat) {
2722 // Remove extra '\}' at end if not part of \{\.\}
2723 size_t lng = par_as_string.size();
2725 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2727 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
2736 if (lng < par_as_string.size())
2737 par_as_string = par_as_string.substr(0,lng);
2740 regex_replace(par_as_string, par_as_string, "\\\\\\.", "_xxbdotxx_");
2741 // handle '.' -> '[^]', replace later as '[^\}\{\\]'
2742 regex_replace(par_as_string, par_as_string, "\\.", "[^]");
2743 // replace '[^...]' with '[^...\}\{\\]'
2744 regex_replace(par_as_string, par_as_string, "\\[\\^([^\\\\\\]]*)\\]", "_xxbrlxx_$1\\}\\{\\\\_xxbrrxx_");
2745 regex_replace(par_as_string, par_as_string, "_xxbrlxx_", "[^");
2746 regex_replace(par_as_string, par_as_string, "_xxbrrxx_", "]");
2748 regex_replace(par_as_string, par_as_string, "_xxbdotxx_", "\\.");
2751 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2752 LYXERR(Debug::FIND, "Open braces: " << open_braces);
2753 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
2754 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
2756 // If entered regexp must match at begin of searched string buffer
2757 // Kornel: Added parentheses to use $1 for size of the leading string
2761 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
2762 // Unfortunately is '\1', '\2', etc not working for strings with extra format
2763 // so the convert has no effect in that case
2764 for (int i = 8; i > 0; --i) {
2765 string orig = "\\\\" + std::to_string(i);
2766 string dest = "\\" + std::to_string(i+1);
2767 while (regex_replace(par_as_string, par_as_string, orig, dest));
2769 regexp_str = "(" + lead_as_regexp + ")" + par_as_string;
2770 regexp2_str = "(" + lead_as_regexp + ").*?" + par_as_string;
2772 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2773 regexp = lyx::regex(regexp_str);
2775 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2776 regexp2 = lyx::regex(regexp2_str);
2781 // Count number of characters in string
2785 // \\[a-zA-Z]+ ==> 1
2786 static int computeSize(string s, int len)
2792 for (int i = 0; i < len; i += skip, count++) {
2795 if (isalpha(s[i+1])) {
2796 for (int j = 2; i+j < len; j++) {
2797 if (! isalpha(s[i+j])) {
2800 else if ((s[i+j] == '{') && s[i+j+1] == '}')
2802 else if ((s[i+j] == '{') && (i + j + 1 >= len))
2810 else if (s[i] == '{') {
2816 else if (s[i] == '-') {
2817 if (s[i+1] == '-') {
2833 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
2838 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
2841 docstring docstr = stringifyFromForSearch(opt, cur, len);
2842 string str = normalize(docstr, true);
2843 if (!opt.ignoreformat) {
2844 str = correctlanguagesetting(str, false, !opt.ignoreformat);
2847 mres.match_len = -1;
2850 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
2851 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
2854 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
2855 regex const *p_regexp;
2856 regex_constants::match_flag_type flags;
2858 flags = regex_constants::match_continuous;
2861 flags = regex_constants::match_default;
2862 p_regexp = ®exp2;
2864 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
2865 if (re_it == sregex_iterator())
2867 match_results<string::const_iterator> const & m = *re_it;
2869 if (0) { // Kornel Benko: DO NOT CHECKK
2870 // Check braces on the segment that matched the entire regexp expression,
2871 // plus the last subexpression, if a (.*?) was inserted in the constructor.
2872 if (!braces_match(m[0].first, m[0].second, open_braces))
2876 // Check braces on segments that matched all (.*?) subexpressions,
2877 // except the last "padding" one inserted by lyx.
2878 for (size_t i = 1; i < m.size() - 1; ++i)
2879 if (!braces_match(m[i].first, m[i].second, open_braces))
2882 // Exclude from the returned match length any length
2883 // due to close wildcards added at end of regexp
2884 // and also the length of the leading (e.g. '\emph{}')
2886 // Whole found string, including the leading: m[0].second - m[0].first
2887 // Size of the leading string: m[1].second - m[1].first
2888 int leadingsize = 0;
2890 leadingsize = m[1].second - m[1].first;
2892 for (size_t i = 0; i < m.size(); i++) {
2893 LYXERR(Debug::FIND, "Match " << i << " is " << m[i].second - m[i].first << " long");
2895 if (close_wildcards == 0)
2896 result = m[0].second - m[0].first;
2899 result = m[m.size() - close_wildcards].first - m[0].first;
2901 size_t pos = m.position(size_t(0));
2902 // Ignore last closing characters
2903 while (result > 0) {
2904 if (str[pos+result-1] == '}')
2909 if (result > leadingsize)
2910 result -= leadingsize;
2913 mres.match_len = computeSize(str.substr(pos+leadingsize,result), result);
2914 mres.match2end = str.size() - pos - leadingsize;
2915 mres.pos = pos+leadingsize;
2919 // else !use_regexp: but all code paths above return
2920 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='"
2921 << par_as_string << "', str='" << str << "'");
2922 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='"
2923 << lead_as_string << "', par_as_string_nolead='"
2924 << par_as_string_nolead << "'");
2927 LYXERR(Debug::FIND, "size=" << par_as_string.size()
2928 << ", substr='" << str.substr(0, par_as_string.size()) << "'");
2929 if (str.substr(0, par_as_string.size()) == par_as_string) {
2930 mres.match_len = par_as_string.size();
2931 mres.match2end = str.size();
2936 // Start the search _after_ the leading part
2937 size_t pos = str.find(par_as_string_nolead, lead_as_string.size());
2938 if (pos != string::npos) {
2939 mres.match_len = par_as_string.size();
2940 mres.match2end = str.size() - pos;
2949 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
2951 MatchResult mres = findAux(cur, len, at_begin);
2952 int res = mres.match_len;
2954 "res=" << res << ", at_begin=" << at_begin
2955 << ", matchword=" << opt.matchword
2956 << ", inTexted=" << cur.inTexted());
2957 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
2959 if ((len > 0) && (res < len)) {
2963 Paragraph const & par = cur.paragraph();
2964 bool ws_left = (cur.pos() > 0)
2965 ? par.isWordSeparator(cur.pos() - 1)
2967 bool ws_right = (cur.pos() + len < par.size())
2968 ? par.isWordSeparator(cur.pos() + len)
2971 "cur.pos()=" << cur.pos() << ", res=" << res
2972 << ", separ: " << ws_left << ", " << ws_right
2975 if (ws_left && ws_right) {
2976 // Check for word separators inside the found 'word'
2977 for (int i = 0; i < len; i++) {
2978 if (par.isWordSeparator(cur.pos() + i)) {
2990 string MatchStringAdv::normalize(docstring const & s, bool hack_braces) const
2993 if (! opt.casesensitive)
2994 t = lyx::to_utf8(lowercase(s));
2996 t = lyx::to_utf8(s);
2997 // Remove \n at begin
2998 while (!t.empty() && t[0] == '\n')
3001 while (!t.empty() && t[t.size() - 1] == '\n')
3002 t = t.substr(0, t.size() - 1);
3004 // Handle all other '\n'
3005 while ((pos = t.find("\n")) != string::npos) {
3006 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3008 if (isAlnumASCII(t[pos+1])) {
3009 t.replace(pos-2, 3, " ");
3012 t.replace(pos-2, 3, "");
3015 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3016 // '\n' adjacent to non-alpha-numerics, discard
3017 t.replace(pos, 1, "");
3020 // Replace all other \n with spaces
3021 t.replace(pos, 1, " ");
3024 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3025 // Kornel: Added textsl, textsf, textit, texttt and noun
3026 // + allow to seach for colored text too
3027 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
3028 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3029 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3030 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3031 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3033 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3034 // FIXME - check what preceeds the brace
3036 if (opt.ignoreformat)
3037 while (regex_replace(t, t, "\\{", "_x_<")
3038 || regex_replace(t, t, "\\}", "_x_>"))
3039 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3041 while (regex_replace(t, t, "\\\\\\{", "_x_<")
3042 || regex_replace(t, t, "\\\\\\}", "_x_>"))
3043 LYXERR(Debug::FIND, "After {} replacement: '" << t << "'");
3050 docstring stringifyFromCursor(DocIterator const & cur, int len)
3052 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3053 if (cur.inTexted()) {
3054 Paragraph const & par = cur.paragraph();
3055 // TODO what about searching beyond/across paragraph breaks ?
3056 // TODO Try adding a AS_STR_INSERTS as last arg
3057 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3058 int(par.size()) : cur.pos() + len;
3059 // OutputParams runparams(&cur.buffer()->params().encoding());
3060 OutputParams runparams(encodings.fromLyXName("utf8"));
3061 runparams.nice = true;
3062 runparams.flavor = OutputParams::XETEX;
3063 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3064 // No side effect of file copying and image conversion
3065 runparams.dryrun = true;
3066 runparams.for_search = true;
3067 LYXERR(Debug::FIND, "Stringifying with cur: "
3068 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3069 return par.asString(cur.pos(), end,
3070 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3072 } else if (cur.inMathed()) {
3073 CursorSlice cs = cur.top();
3074 MathData md = cs.cell();
3075 MathData::const_iterator it_end =
3076 (( len == -1 || cs.pos() + len > int(md.size()))
3078 : md.begin() + cs.pos() + len );
3080 for (MathData::const_iterator it = md.begin() + cs.pos();
3083 docstring s = asString(md2);
3084 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3087 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3092 /** Computes the LaTeX export of buf starting from cur and ending len positions
3093 * after cur, if len is positive, or at the paragraph or innermost inset end
3096 docstring latexifyFromCursor(DocIterator const & cur, int len)
3098 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3099 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3100 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3101 Buffer const & buf = *cur.buffer();
3103 odocstringstream ods;
3105 //OutputParams runparams(&buf.params().encoding());
3106 OutputParams runparams(encodings.fromLyXName("utf8"));
3107 runparams.nice = false;
3108 runparams.flavor = OutputParams::XETEX;
3109 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3110 // No side effect of file copying and image conversion
3111 runparams.dryrun = true;
3112 runparams.for_search = true;
3114 if (cur.inTexted()) {
3115 // @TODO what about searching beyond/across paragraph breaks ?
3116 pos_type endpos = cur.paragraph().size();
3117 if (len != -1 && endpos > cur.pos() + len)
3118 endpos = cur.pos() + len;
3119 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3120 string(), cur.pos(), endpos);
3121 string s = lyx::to_utf8(ods.str());
3122 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3123 return(lyx::from_utf8(s));
3124 } else if (cur.inMathed()) {
3125 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3126 for (int s = cur.depth() - 1; s >= 0; --s) {
3127 CursorSlice const & cs = cur[s];
3128 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3130 cs.asInsetMath()->asHullInset()->header_write(ws);
3135 CursorSlice const & cs = cur.top();
3136 MathData md = cs.cell();
3137 MathData::const_iterator it_end =
3138 ((len == -1 || cs.pos() + len > int(md.size()))
3140 : md.begin() + cs.pos() + len);
3142 for (MathData::const_iterator it = md.begin() + cs.pos();
3146 ods << asString(md2);
3147 // Retrieve the math environment type, and add '$' or '$]'
3148 // or others (\end{equation}) accordingly
3149 for (int s = cur.depth() - 1; s >= 0; --s) {
3150 CursorSlice const & cs2 = cur[s];
3151 InsetMath * inset = cs2.asInsetMath();
3152 if (inset && inset->asHullInset()) {
3154 inset->asHullInset()->footer_write(ws);
3158 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3160 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3166 /** Finalize an advanced find operation, advancing the cursor to the innermost
3167 ** position that matches, plus computing the length of the matching text to
3170 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
3172 // Search the foremost position that matches (avoids find of entire math
3173 // inset when match at start of it)
3175 DocIterator old_cur(cur.buffer());
3177 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
3181 } while (cur && cur.depth() > d && match(cur).match_len > 0);
3183 int max_match = match(cur).match_len; /* match valid only if not searching whole words */
3184 if (max_match <= 0) return 0;
3185 LYXERR(Debug::FIND, "Ok");
3187 // Compute the match length
3189 if (cur.pos() + len > cur.lastpos())
3191 if (match.opt.matchword) {
3192 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3193 while (cur.pos() + len <= cur.lastpos() && match(cur, len).match_len <= 0) {
3195 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
3197 // Length of matched text (different from len param)
3198 int old_match = match(cur, len).match_len;
3202 // Greedy behaviour while matching regexps
3203 while ((new_match = match(cur, len + 1).match_len) > old_match) {
3205 old_match = new_match;
3206 LYXERR(Debug::FIND, "verifying match with len = " << len);
3213 int maxl = cur.lastpos() - cur.pos();
3214 // Greedy behaviour while matching regexps
3215 while (maxl > minl) {
3216 int actual_match = match(cur, len).match_len;
3217 if (actual_match >= max_match) {
3218 // actual_match > max_match _can_ happen,
3219 // if the search area splits
3220 // some following word so that the regex
3221 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3223 // This means, the len value is too big
3225 len = (int)((maxl + minl)/2);
3228 // (actual_match < max_match)
3230 len = (int)((maxl + minl)/2);
3234 // Search for real start of matched characters
3239 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3240 if (cur.depth() < old_cur.depth()) {
3242 LYXERR0("cur.depth() < old_cur.depth(), this should never happen");
3245 if (cur.pos() != old_cur.pos()) {
3246 // OK, forwarded 1 pos in actual inset
3247 actual_match = match(cur, len-1).match_len;
3248 if (actual_match == max_match) {
3249 // Ha, got it! The shorter selection has the same match length
3254 // OK, the shorter selection matches less chars, revert to previous value
3260 LYXERR0("cur.pos() == old_cur.pos(), this should never happen");
3261 actual_match = match(cur, len).match_len;
3262 if (actual_match == max_match)
3272 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3276 while (!theApp()->longOperationCancelled() && cur) {
3277 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3278 MatchResult mres = match(cur, -1, false);
3279 int match_len = mres.match_len;
3280 LYXERR(Debug::FIND, "match_len: " << match_len);
3281 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3282 LYXERR0("BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3285 if (match_len > 0) {
3286 // Try to find the begin of searched string
3287 int increment = mres.pos/2;
3288 while (mres.pos > 5 && (increment > 5)) {
3289 DocIterator old_cur = cur;
3290 for (int i = 0; i < increment && cur; cur.forwardPos(), i++) {
3292 if (! cur || (cur.pit() > old_cur.pit())) {
3293 // Are we outside of the paragraph?
3294 // This can happen if moving past some UTF8-encoded chars
3299 MatchResult mres2 = match(cur, -1, false);
3300 if ((mres2.match2end < mres.match2end) ||
3301 (mres2.match_len < mres.match_len)) {
3308 if (increment > mres.pos/2)
3309 increment = mres.pos/2;
3313 int match_len_zero_count = 0;
3314 for (int i = 0; !theApp()->longOperationCancelled() && cur; cur.forwardPos()) {
3316 int remaining_len = match(cur, -1, false).match_len;
3317 if (remaining_len <= 0) {
3318 // Apparently the searched string is not in the remaining part
3325 LYXERR(Debug::FIND, "Advancing cur: " << cur);
3326 int match_len3 = match(cur, 1).match_len;
3329 int match_len2 = match(cur).match_len;
3330 LYXERR(Debug::FIND, "match_len2: " << match_len2);
3331 if (match_len2 > 0) {
3332 // Sometimes in finalize we understand it wasn't a match
3333 // and we need to continue the outest loop
3334 int len = findAdvFinalize(cur, match);
3339 if (match_len2 >= 0) {
3340 if (match_len2 == 0)
3341 match_len_zero_count++;
3343 match_len_zero_count = 0;
3346 if (++match_len_zero_count > 3) {
3347 LYXERR(Debug::FIND, "match_len2_zero_count: " << match_len_zero_count << ", match_len was " << match_len);
3348 match_len_zero_count = 0;
3356 if (match_len >= 0 && cur.pit() < cur.lastpit()) {
3357 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
3360 // This should exit nested insets, if any, or otherwise undefine the currsor.
3361 cur.pos() = cur.lastpos();
3362 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3370 /// Find the most backward consecutive match within same paragraph while searching backwards.
3371 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3373 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3374 DocIterator tmp_cur = cur;
3375 int len = findAdvFinalize(tmp_cur, match);
3376 Inset & inset = cur.inset();
3377 for (; cur != cur_begin; cur.backwardPos()) {
3378 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3379 DocIterator new_cur = cur;
3380 new_cur.backwardPos();
3381 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3383 int new_len = findAdvFinalize(new_cur, match);
3388 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3394 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3398 // Backup of original position
3399 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3400 if (cur == cur_begin)
3403 DocIterator cur_orig(cur);
3404 bool pit_changed = false;
3407 bool found_match = (match(cur, -1, false).match_len > 0);
3411 cur.pos() = cur.lastpos();
3413 cur.pos() = cur_orig.pos();
3414 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3415 DocIterator cur_prev_iter;
3417 found_match = (match(cur).match_len > 0);
3418 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3419 << found_match << ", cur: " << cur);
3421 return findMostBackwards(cur, match);
3423 // Stop if begin of document reached
3424 if (cur == cur_begin)
3426 cur_prev_iter = cur;
3430 if (cur == cur_begin)
3437 } while (!theApp()->longOperationCancelled());
3445 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3446 DocIterator const & cur, int len)
3448 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3450 if (!opt.ignoreformat)
3451 return latexifyFromCursor(cur, len);
3453 return stringifyFromCursor(cur, len);
3457 FindAndReplaceOptions::FindAndReplaceOptions(
3458 docstring const & find_buf_name, bool casesensitive,
3459 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
3460 docstring const & repl_buf_name, bool keep_case,
3461 SearchScope scope, SearchRestriction restr, bool replace_all)
3462 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
3463 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
3464 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope), restr(restr), replace_all(replace_all)
3472 /** Check if 'len' letters following cursor are all non-lowercase */
3473 static bool allNonLowercase(Cursor const & cur, int len)
3475 pos_type beg_pos = cur.selectionBegin().pos();
3476 pos_type end_pos = cur.selectionBegin().pos() + len;
3477 if (len > cur.lastpos() + 1 - beg_pos) {
3478 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3479 len = cur.lastpos() + 1 - beg_pos;
3480 end_pos = beg_pos + len;
3482 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3483 if (isLowerCase(cur.paragraph().getChar(pos)))
3489 /** Check if first letter is upper case and second one is lower case */
3490 static bool firstUppercase(Cursor const & cur)
3493 pos_type pos = cur.selectionBegin().pos();
3494 if (pos >= cur.lastpos() - 1) {
3495 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3498 ch1 = cur.paragraph().getChar(pos);
3499 ch2 = cur.paragraph().getChar(pos + 1);
3500 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3501 LYXERR(Debug::FIND, "firstUppercase(): "
3502 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3503 << ch2 << "(" << char(ch2) << ")"
3504 << ", result=" << result << ", cur=" << cur);
3509 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3511 ** \fixme What to do with possible further paragraphs in replace buffer ?
3513 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3515 ParagraphList::iterator pit = buffer.paragraphs().begin();
3516 LASSERT(pit->size() >= 1, /**/);
3517 pos_type right = pos_type(1);
3518 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3519 right = pit->size();
3520 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3526 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3528 Cursor & cur = bv->cursor();
3529 if (opt.repl_buf_name == docstring()
3530 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3531 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3534 DocIterator sel_beg = cur.selectionBegin();
3535 DocIterator sel_end = cur.selectionEnd();
3536 if (&sel_beg.inset() != &sel_end.inset()
3537 || sel_beg.pit() != sel_end.pit()
3538 || sel_beg.idx() != sel_end.idx())
3540 int sel_len = sel_end.pos() - sel_beg.pos();
3541 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3542 << ", sel_len: " << sel_len << endl);
3545 LASSERT(sel_len > 0, return 0);
3547 if (!matchAdv(sel_beg, sel_len).match_len)
3550 // Build a copy of the replace buffer, adapted to the KeepCase option
3551 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3553 repl_buffer_orig.write(oss);
3554 string lyx = oss.str();
3555 Buffer repl_buffer("", false);
3556 repl_buffer.setUnnamed(true);
3557 LASSERT(repl_buffer.readString(lyx), return 0);
3558 if (opt.keep_case && sel_len >= 2) {
3559 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3560 if (cur.inTexted()) {
3561 if (firstUppercase(cur))
3562 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3563 else if (allNonLowercase(cur, sel_len))
3564 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3567 cap::cutSelection(cur, false);
3568 if (cur.inTexted()) {
3569 repl_buffer.changeLanguage(
3570 repl_buffer.language(),
3571 cur.getFont().language());
3572 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3573 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3574 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3575 repl_buffer.params().documentClassPtr(),
3576 bv->buffer().errorList("Paste"));
3577 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3578 sel_len = repl_buffer.paragraphs().begin()->size();
3579 } else if (cur.inMathed()) {
3580 odocstringstream ods;
3582 // OutputParams runparams(&repl_buffer.params().encoding());
3583 OutputParams runparams(encodings.fromLyXName("utf8"));
3584 runparams.nice = false;
3585 runparams.flavor = OutputParams::XETEX;
3586 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3587 runparams.dryrun = true;
3588 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3589 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3590 docstring repl_latex = ods.str();
3591 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3593 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3594 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3595 repl_latex = from_utf8(s);
3596 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3597 MathData ar(cur.buffer());
3598 asArray(repl_latex, ar, Parse::NORMAL);
3600 sel_len = ar.size();
3601 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3603 if (cur.pos() >= sel_len)
3604 cur.pos() -= sel_len;
3607 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3608 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3609 bv->processUpdateFlags(Update::Force);
3614 /// Perform a FindAdv operation.
3615 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
3620 // e.g., when invoking word-findadv from mini-buffer wither with
3621 // wrong options syntax or before ever opening advanced F&R pane
3622 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3626 MatchStringAdv matchAdv(bv->buffer(), opt);
3627 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
3629 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
3630 num_replaced += findAdvReplace(bv, opt, matchAdv);
3633 match_len = findForwardAdv(cur, matchAdv);
3635 match_len = findBackwardsAdv(cur, matchAdv);
3637 // This may only be raised by lyx::regex()
3638 bv->message(_("Invalid regular expression!"));
3642 if (match_len == 0) {
3643 if (num_replaced > 0) {
3644 switch (num_replaced)
3647 bv->message(_("One match has been replaced."));
3650 bv->message(_("Two matches have been replaced."));
3653 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
3659 bv->message(_("Match not found."));
3664 if (num_replaced > 0)
3665 bv->message(_("Match has been replaced."));
3667 bv->message(_("Match found."));
3669 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
3670 bv->putSelectionAt(cur, match_len, !opt.forward);
3676 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
3678 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
3679 << opt.casesensitive << ' '
3680 << opt.matchword << ' '
3681 << opt.forward << ' '
3682 << opt.expandmacros << ' '
3683 << opt.ignoreformat << ' '
3684 << opt.replace_all << ' '
3685 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
3686 << opt.keep_case << ' '
3687 << int(opt.scope) << ' '
3690 LYXERR(Debug::FIND, "built: " << os.str());
3696 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
3698 LYXERR(Debug::FIND, "parsing");
3702 while (line != "EOSS") {
3706 if (is.eof()) // Tolerate malformed request
3710 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
3711 opt.find_buf_name = from_utf8(s);
3712 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
3713 is.get(); // Waste space before replace string
3716 while (line != "EOSS") {
3720 if (is.eof()) // Tolerate malformed request
3724 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
3725 opt.repl_buf_name = from_utf8(s);
3726 is >> opt.keep_case;
3729 opt.scope = FindAndReplaceOptions::SearchScope(i);
3731 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
3733 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
3734 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
3735 << opt.scope << ' ' << opt.restr);