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
11 * \author Kornel Benko
13 * Full author contact details are available in file CREDITS.
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"
36 #include "frontends/Application.h"
37 #include "frontends/alert.h"
39 #include "mathed/InsetMath.h"
40 #include "mathed/InsetMathHull.h"
41 #include "mathed/MathData.h"
42 #include "mathed/MathStream.h"
43 #include "mathed/MathSupport.h"
45 #include "support/debug.h"
46 #include "support/docstream.h"
47 #include "support/FileName.h"
48 #include "support/gettext.h"
49 #include "support/lassert.h"
50 #include "support/lstrings.h"
51 #include "support/textutils.h"
56 //#define ResultsDebug
57 #define USE_QT_FOR_SEARCH
58 #if defined(USE_QT_FOR_SEARCH)
59 #include <QtCore> // sets QT_VERSION
60 #if (QT_VERSION >= 0x050000)
61 #include <QRegularExpression>
71 using namespace lyx::support;
75 typedef map<string, string> AccentsMap;
76 static AccentsMap accents = map<string, string>();
78 // Helper class for deciding what should be ignored
82 IgnoreFormats() = default;
84 bool getFamily() const { return ignoreFamily_; }
86 bool getSeries() const { return ignoreSeries_; }
88 bool getShape() const { return ignoreShape_; }
90 bool getUnderline() const { return ignoreUnderline_; }
92 bool getMarkUp() const { return ignoreMarkUp_; }
94 bool getStrikeOut() const { return ignoreStrikeOut_; }
96 bool getSectioning() const { return ignoreSectioning_; }
98 bool getFrontMatter() const { return ignoreFrontMatter_; }
100 bool getColor() const { return ignoreColor_; }
102 bool getLanguage() const { return ignoreLanguage_; }
104 bool getDeleted() const { return ignoreDeleted_; }
106 void setIgnoreDeleted(bool value);
108 void setIgnoreFormat(string const & type, bool value);
112 bool ignoreFamily_ = false;
114 bool ignoreSeries_ = false;
116 bool ignoreShape_ = false;
118 bool ignoreUnderline_ = false;
120 bool ignoreMarkUp_ = false;
122 bool ignoreStrikeOut_ = false;
124 bool ignoreSectioning_ = false;
126 bool ignoreFrontMatter_ = false;
128 bool ignoreColor_ = false;
130 bool ignoreLanguage_ = false;
132 bool ignoreDeleted_ = true;
135 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
137 if (type == "color") {
138 ignoreColor_ = value;
140 else if (type == "language") {
141 ignoreLanguage_ = value;
143 else if (type == "sectioning") {
144 ignoreSectioning_ = value;
145 ignoreFrontMatter_ = value;
147 else if (type == "font") {
148 ignoreSeries_ = value;
149 ignoreShape_ = value;
150 ignoreFamily_ = value;
152 else if (type == "series") {
153 ignoreSeries_ = value;
155 else if (type == "shape") {
156 ignoreShape_ = value;
158 else if (type == "family") {
159 ignoreFamily_ = value;
161 else if (type == "markup") {
162 ignoreMarkUp_ = value;
164 else if (type == "underline") {
165 ignoreUnderline_ = value;
167 else if (type == "strike") {
168 ignoreStrikeOut_ = value;
170 else if (type == "deleted") {
171 ignoreDeleted_ = value;
175 // The global variable that can be changed from outside
176 IgnoreFormats ignoreFormats;
179 void setIgnoreFormat(string const & type, bool value)
181 ignoreFormats.setIgnoreFormat(type, value);
187 bool parse_bool(docstring & howto)
192 howto = split(howto, var, ' ');
200 MatchString(docstring const & s, bool cs, bool mw)
201 : str(s), case_sens(cs), whole_words(mw)
204 // returns true if the specified string is at the specified position
205 // del specifies whether deleted strings in ct mode will be considered
206 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
208 return par.find(str, case_sens, whole_words, pos, del);
216 // match whole words only
221 int findForward(DocIterator & cur, MatchString const & match,
222 bool find_del = true)
224 for (; cur; cur.forwardChar())
225 if (cur.inTexted()) {
226 int len = match(cur.paragraph(), cur.pos(), find_del);
234 int findBackwards(DocIterator & cur, MatchString const & match,
235 bool find_del = true)
239 if (cur.inTexted()) {
240 int len = match(cur.paragraph(), cur.pos(), find_del);
249 bool searchAllowed(docstring const & str)
252 frontend::Alert::error(_("Search error"), _("Search string is empty"));
259 bool findOne(BufferView * bv, docstring const & searchstr,
260 bool case_sens, bool whole, bool forward,
261 bool find_del = true, bool check_wrap = false)
263 if (!searchAllowed(searchstr))
266 DocIterator cur = forward
267 ? bv->cursor().selectionEnd()
268 : bv->cursor().selectionBegin();
270 MatchString const match(searchstr, case_sens, whole);
272 int match_len = forward
273 ? findForward(cur, match, find_del)
274 : findBackwards(cur, match, find_del);
277 bv->putSelectionAt(cur, match_len, !forward);
278 else if (check_wrap) {
279 DocIterator cur_orig(bv->cursor());
282 q = _("End of file reached while searching forward.\n"
283 "Continue searching from the beginning?");
285 q = _("Beginning of file reached while searching backward.\n"
286 "Continue searching from the end?");
287 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
288 q, 0, 1, _("&Yes"), _("&No"));
289 if (wrap_answer == 0) {
291 bv->cursor().clear();
292 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
294 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
295 bv->cursor().backwardPos();
297 bv->clearSelection();
298 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
301 bv->cursor().setCursor(cur_orig);
305 return match_len > 0;
309 int replaceAll(BufferView * bv,
310 docstring const & searchstr, docstring const & replacestr,
311 bool case_sens, bool whole)
313 Buffer & buf = bv->buffer();
315 if (!searchAllowed(searchstr) || buf.isReadonly())
318 DocIterator cur_orig(bv->cursor());
320 MatchString const match(searchstr, case_sens, whole);
323 int const rsize = replacestr.size();
324 int const ssize = searchstr.size();
327 cur.setCursor(doc_iterator_begin(&buf));
328 int match_len = findForward(cur, match, false);
329 while (match_len > 0) {
330 // Backup current cursor position and font.
331 pos_type const pos = cur.pos();
332 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
334 int striked = ssize -
335 cur.paragraph().eraseChars(pos, pos + match_len,
336 buf.params().track_changes);
337 cur.paragraph().insert(pos, replacestr, font,
338 Change(buf.params().track_changes
340 : Change::UNCHANGED));
341 for (int i = 0; i < rsize + striked; ++i)
344 match_len = findForward(cur, match, false);
347 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
349 cur_orig.fixIfBroken();
350 bv->setCursor(cur_orig);
356 // the idea here is that we are going to replace the string that
357 // is selected IF it is the search string.
358 // if there is a selection, but it is not the search string, then
359 // we basically ignore it. (FIXME We ought to replace only within
361 // if there is no selection, then:
362 // (i) if some search string has been provided, then we find it.
363 // (think of how the dialog works when you hit "replace" the
365 // (ii) if no search string has been provided, then we treat the
366 // word the cursor is in as the search string. (why? i have no
367 // idea.) but this only works in text?
369 // returns the number of replacements made (one, if any) and
370 // whether anything at all was done.
371 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
372 docstring const & replacestr, bool case_sens,
373 bool whole, bool forward, bool findnext)
375 Cursor & cur = bv->cursor();
376 if (!cur.selection()) {
377 // no selection, non-empty search string: find it
378 if (!searchstr.empty()) {
379 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
380 return make_pair(found, 0);
382 // empty search string
385 return make_pair(false, 0);
386 // select current word and treat it as the search string.
387 // This causes a minor bug as undo will restore this selection,
388 // which the user did not create (#8986).
389 cur.innerText()->selectWord(cur, WHOLE_WORD);
390 searchstr = cur.selectionAsString(false, true);
393 // if we still don't have a search string, report the error
395 if (!searchAllowed(searchstr))
396 return make_pair(false, 0);
398 bool have_selection = cur.selection();
399 docstring const selected = cur.selectionAsString(false, true);
402 ? searchstr == selected
403 : compare_no_case(searchstr, selected) == 0;
405 // no selection or current selection is not search word:
406 // just find the search word
407 if (!have_selection || !match) {
408 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
409 return make_pair(found, 0);
412 // we're now actually ready to replace. if the buffer is
413 // read-only, we can't, though.
414 if (bv->buffer().isReadonly())
415 return make_pair(false, 0);
417 cap::replaceSelectionWithString(cur, replacestr);
419 cur.pos() += replacestr.length();
420 LASSERT(cur.pos() <= cur.lastpos(),
421 cur.pos() = cur.lastpos());
424 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
426 return make_pair(true, 1);
432 docstring const find2string(docstring const & search,
433 bool casesensitive, bool matchword, bool forward)
437 << int(casesensitive) << ' '
438 << int(matchword) << ' '
444 docstring const replace2string(docstring const & replace,
445 docstring const & search,
446 bool casesensitive, bool matchword,
447 bool all, bool forward, bool findnext)
450 ss << replace << '\n'
452 << int(casesensitive) << ' '
453 << int(matchword) << ' '
455 << int(forward) << ' '
461 bool lyxfind(BufferView * bv, FuncRequest const & ev)
463 if (!bv || ev.action() != LFUN_WORD_FIND)
466 //lyxerr << "find called, cmd: " << ev << endl;
468 // data is of the form
470 // <casesensitive> <matchword> <forward>"
472 docstring howto = split(ev.argument(), search, '\n');
474 bool casesensitive = parse_bool(howto);
475 bool matchword = parse_bool(howto);
476 bool forward = parse_bool(howto);
478 return findOne(bv, search, casesensitive, matchword, forward, false, true);
482 bool lyxreplace(BufferView * bv, FuncRequest const & ev)
484 if (!bv || ev.action() != LFUN_WORD_REPLACE)
487 // data is of the form
490 // <casesensitive> <matchword> <all> <forward> <findnext>"
493 docstring howto = split(ev.argument(), rplc, '\n');
494 howto = split(howto, search, '\n');
496 bool casesensitive = parse_bool(howto);
497 bool matchword = parse_bool(howto);
498 bool all = parse_bool(howto);
499 bool forward = parse_bool(howto);
500 bool findnext = howto.empty() ? true : parse_bool(howto);
504 int replace_count = 0;
506 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
507 update = replace_count > 0;
510 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
512 replace_count = rv.second;
515 Buffer const & buf = bv->buffer();
517 // emit message signal.
518 buf.message(_("String not found."));
520 if (replace_count == 0) {
521 buf.message(_("String found."));
522 } else if (replace_count == 1) {
523 buf.message(_("String has been replaced."));
525 docstring const str =
526 bformat(_("%1$d strings have been replaced."), replace_count);
534 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
536 for (; cur; cur.forwardPos())
537 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
541 DocIterator cur_orig(bv->cursor());
542 docstring q = _("End of file reached while searching forward.\n"
543 "Continue searching from the beginning?");
544 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
545 q, 0, 1, _("&Yes"), _("&No"));
546 if (wrap_answer == 0) {
547 bv->cursor().clear();
548 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
549 bv->clearSelection();
550 cur.setCursor(bv->cursor().selectionBegin());
551 if (findNextChange(bv, cur, false))
554 bv->cursor().setCursor(cur_orig);
561 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
563 for (cur.backwardPos(); cur; cur.backwardPos()) {
564 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
569 DocIterator cur_orig(bv->cursor());
570 docstring q = _("Beginning of file reached while searching backward.\n"
571 "Continue searching from the end?");
572 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
573 q, 0, 1, _("&Yes"), _("&No"));
574 if (wrap_answer == 0) {
575 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
576 bv->cursor().backwardPos();
577 bv->clearSelection();
578 cur.setCursor(bv->cursor().selectionBegin());
579 if (findPreviousChange(bv, cur, false))
582 bv->cursor().setCursor(cur_orig);
589 bool selectChange(Cursor & cur, bool forward)
591 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
593 Change ch = cur.paragraph().lookupChange(cur.pos());
595 CursorSlice tip1 = cur.top();
596 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
597 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
598 if (!ch2.isSimilarTo(ch))
601 CursorSlice tip2 = cur.top();
602 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
604 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
605 if (!ch2.isSimilarTo(ch)) {
606 // take a step forward to correctly set the selection
614 cur.bv().mouseSetCursor(cur, false);
616 cur.bv().mouseSetCursor(cur, true);
624 bool findChange(BufferView * bv, bool forward)
627 cur.setCursor(forward ? bv->cursor().selectionEnd()
628 : bv->cursor().selectionBegin());
629 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
630 return selectChange(cur, forward);
635 bool findNextChange(BufferView * bv)
637 return findChange(bv, true);
641 bool findPreviousChange(BufferView * bv)
643 return findChange(bv, false);
650 typedef vector<pair<string, string> > Escapes;
652 string string2regex(string in)
654 static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\$\\])" };
655 string temp = std::regex_replace(in, specialChars, R"(\$&)" );
660 while (fl_pos < temp.size()) {
661 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
662 if (fl_pos == string::npos)
665 temp2 += temp.substr(lastpos, fl_pos - lastpos);
671 if (lastpos < temp.size()) {
672 temp2 += temp.substr(lastpos, temp.size() - lastpos);
677 string correctRegex(string t, bool withformat)
679 /* Convert \backslash => \
680 * and \{, \}, \[, \] => {, }, [, ]
683 regex wordre("(\\\\)*(\\\\((backslash|mathcircumflex) ?|[\\[\\]\\{\\}]))");
686 bool backslashed = false;
687 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
690 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
694 if (sub.str(4) == "backslash") {
697 // transforms '\backslash \{' into '\{'
699 string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
700 if ((next == "\\{") || (next == "\\}")) {
706 else if (sub.str(4) == "mathcircumflex")
708 else if (backslashed) {
710 if (withformat && (sub.str(3) == "{"))
711 replace = accents["braceleft"];
712 else if (withformat && (sub.str(3) == "}"))
713 replace = accents["braceright"];
715 // else part should not exist
720 replace = sub.str(3);
722 if (lastpos < (size_t) sub.position(2))
723 s += t.substr(lastpos, sub.position(2) - lastpos);
725 lastpos = sub.position(2) + sub.length(2);
729 else if (lastpos < t.length())
730 s += t.substr(lastpos, t.length() - lastpos);
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 withformat)
741 while (lastpos < s.size()) {
742 size_t regex_pos = s.find("\\regexp{", lastpos);
743 if (regex_pos == string::npos) {
744 regex_pos = s.size();
746 if (regex_pos > lastpos) {
747 result += string2regex(s.substr(lastpos, regex_pos-lastpos));
749 if (lastpos == s.size())
752 size_t end_pos = s.find("\\endregexp{}}", regex_pos + 8);
753 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
754 lastpos = end_pos + 13;
760 /// Wrapper for lyx::regex_replace with simpler interface
761 bool regex_replace(string const & s, string & t, string const & searchstr,
762 string const & replacestr)
764 regex e(searchstr, regex_constants::ECMAScript);
766 ostream_iterator<char, char> it(oss);
767 regex_replace(it, s.begin(), s.end(), e, replacestr);
768 // tolerate t and s be references to the same variable
769 bool rv = (s != oss.str());
788 vector <string> result = vector <string>();
789 MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {};
792 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
794 if (newres.match2end < oldres.match2end)
795 return MatchResult::newIsTooFar;
796 if (newres.match_len < oldres.match_len)
797 return MatchResult::newIsTooFar;
799 if (newres.match_len == oldres.match_len) {
800 if (newres.match2end == oldres.match2end)
801 return MatchResult::newIsBetter;
803 return MatchResult::newIsInvalid;
806 /** The class performing a match between a position in the document and the FindAdvOptions.
809 class MatchStringAdv {
811 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
813 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
814 ** constructor as opt.search, under the opt.* options settings.
817 ** If set, then match is searched only against beginning of text starting at cur.
818 ** If unset, then match is searched anywhere in text starting at cur.
821 ** The length of the matching text, or zero if no match was found.
823 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
832 /// first buffer on which search was started
833 lyx::Buffer * const p_first_buf;
835 FindAndReplaceOptions const & opt;
838 /// Auxiliary find method (does not account for opt.matchword)
839 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
840 void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
842 /** Normalize a stringified or latexified LyX paragraph.
846 ** <li>if search is not casesensitive, then lowercase the string;
847 ** <li>remove any newline at begin or end of the string;
848 ** <li>replace any newline in the middle of the string with a simple space;
849 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
852 ** @todo Normalization should also expand macros, if the corresponding
853 ** search option was checked.
855 string normalize(docstring const & s) const;
856 // normalized string to search
857 string par_as_string;
858 // regular expression to use for searching
859 // regexp2 is same as regexp, but prefixed with a ".*?"
861 QRegularExpression regexp;
862 QRegularExpression regexp2;
867 // leading format material as string
868 string lead_as_string;
869 // par_as_string after removal of lead_as_string
870 string par_as_string_nolead;
871 // unmatched open braces in the search string/regexp
873 // number of (.*?) subexpressions added at end of search regexp for closing
874 // environments, math mode, styles, etc...
877 // Are we searching with regular expressions ?
879 static int valid_matches;
880 static vector <string> matches;
881 void FillResults(MatchResult &found_mr);
884 int MatchStringAdv::valid_matches = 0;
885 vector <string> MatchStringAdv::matches = vector <string>(10);
887 void MatchStringAdv::FillResults(MatchResult &found_mr)
889 if (found_mr.match_len > 0) {
890 valid_matches = found_mr.result.size();
891 for (size_t i = 0; i < found_mr.result.size(); i++)
892 matches[i] = found_mr.result[i];
898 static docstring buffer_to_latex(Buffer & buffer)
900 //OutputParams runparams(&buffer.params().encoding());
901 OutputParams runparams(encodings.fromLyXName("utf8"));
902 odocstringstream ods;
904 runparams.nice = true;
905 runparams.flavor = Flavor::XeTeX;
906 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
907 // No side effect of file copying and image conversion
908 runparams.dryrun = true;
909 if (ignoreFormats.getDeleted())
910 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
912 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
913 pit_type const endpit = buffer.paragraphs().size();
914 for (pit_type pit = 0; pit != endpit; ++pit) {
915 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
916 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
922 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
925 if (!opt.ignoreformat) {
926 str = buffer_to_latex(buffer);
928 // OutputParams runparams(&buffer.params().encoding());
929 OutputParams runparams(encodings.fromLyXName("utf8"));
930 runparams.nice = true;
931 runparams.flavor = Flavor::XeTeX;
932 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
933 runparams.dryrun = true;
934 int option = AS_STR_INSETS |AS_STR_PLAINTEXT;
935 if (ignoreFormats.getDeleted()) {
936 option |= AS_STR_SKIPDELETE;
937 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
940 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
942 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
943 Paragraph const & par = buffer.paragraphs().at(pit);
944 LYXERR(Debug::FIND, "Adding to search string: '"
945 << par.asString(pos_type(0), par.size(),
949 str += par.asString(pos_type(0), par.size(),
953 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
954 string t = to_utf8(str);
955 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
962 /// Return separation pos between the leading material and the rest
963 static size_t identifyLeading(string const & s)
966 // @TODO Support \item[text]
967 // Kornel: Added textsl, textsf, textit, texttt and noun
968 // + allow to search for colored text too
969 while (regex_replace(t, t, "^\\\\(("
970 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
971 "lyxaddress|lyxrightaddress|"
972 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
973 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
974 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
975 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
976 || regex_replace(t, t, "^\\$", "")
977 || regex_replace(t, t, "^\\\\\\[", "")
978 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
979 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
981 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
986 * Given a latexified string, retrieve some handled features
987 * The features of the regex will later be compared with the features
988 * of the searched text. If the regex features are not a
989 * subset of the analized, then, in not format ignoring search
990 * we can early stop the search in the relevant inset.
992 typedef map<string, bool> Features;
994 static Features identifyFeatures(string const & s)
996 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
997 static regex const valid("^("
999 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
1000 "emph|noun|text(bf|md|sl|sf|it|tt)|"
1001 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
1002 "(u|uu)line|(s|x)out|uwave|"
1003 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
1004 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
1009 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
1012 if (sub.str(1).compare("regexp") == 0) {
1016 string token = sub.str(1);
1018 if (regex_match(token, sub2, valid)) {
1026 if (sub.str(1).compare("endregexp") == 0) {
1036 * defines values features of a key "\\[a-z]+{"
1041 /* Char type with content discarded
1042 * like \hspace{1cm} */
1044 /* Char, like \backslash */
1046 /* replace starting backslash with '#' */
1048 /* \part, \section*, ... */
1050 /* title, author etc */
1052 /* \foreignlanguage{ngerman}, ... */
1055 * to discard language in content */
1058 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1060 /* fonts, colors, markups, ... */
1062 /* footnotesize, ... large, ...
1063 * Ignore all of them */
1066 /* inputencoding, ...
1067 * Discard also content, because they do not help in search */
1070 * like remove, but also all arguments */
1072 /* item, listitem */
1074 /* tex, latex, ... like isChar */
1076 /* like \lettrine[lines=5]{}{} */
1078 // like isStandard, but always remove head
1080 /* End of arguments marker for lettrine,
1081 * so that they can be ignored */
1084 KeyInfo() = default;
1085 KeyInfo(KeyType type, int parcount, bool disable)
1087 parenthesiscount(parcount),
1088 disabled(disable) {}
1089 KeyType keytype = invalid;
1091 int _tokensize = -1;
1092 int _tokenstart = -1;
1093 int _dataStart = -1;
1095 int parenthesiscount = 1;
1096 bool disabled = false;
1097 bool used = false; /* by pattern */
1102 Border(int l=0, int u=0) : low(l), upper(u) {};
1107 #define MAXOPENED 30
1109 bool isPatternString_;
1111 explicit Intervall(bool isPattern, string const & p) :
1112 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1113 hasTitle(false), langcount(0)
1121 static vector<Border> borders;
1122 int depts[MAXOPENED];
1123 int closes[MAXOPENED];
1124 int actualdeptindex;
1125 int previousNotIgnored(int) const;
1126 int nextNotIgnored(int) const;
1127 void handleOpenP(int i);
1128 void handleCloseP(int i, bool closingAllowed);
1129 void resetOpenedP(int openPos);
1130 void addIntervall(int upper);
1131 void addIntervall(int low, int upper); /* if explicit */
1132 void removeAccents();
1133 void setForDefaultLang(KeyInfo const & defLang) const;
1134 int findclosing(int start, int end, char up, char down, int repeat);
1135 void handleParentheses(int lastpos, bool closingAllowed);
1137 int langcount; // Number of disabled language specs up to current position in actual interval
1138 int isOpeningPar(int pos) const;
1140 void output(ostringstream &os, int lastpos);
1141 // string show(int lastpos);
1144 vector<Border> Intervall::borders = vector<Border>(30);
1146 int Intervall::isOpeningPar(int pos) const
1148 if ((pos < 0) || (size_t(pos) >= par.size()))
1150 if (par[pos] != '{')
1152 if (size_t(pos) + 2 >= par.size())
1154 if (par[pos+2] != '}')
1156 if (par[pos+1] == '[' || par[pos+1] == ']')
1161 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1163 // Enable the use of first token again
1164 if (ignoreidx >= 0) {
1165 int value = defLang._tokenstart + defLang._tokensize;
1171 if (borders[borderidx].low < value)
1172 borders[borderidx].low = value;
1173 if (borders[borderidx].upper < value)
1174 borders[borderidx].upper = value;
1179 static void checkDepthIndex(int val)
1181 static int maxdepthidx = MAXOPENED-2;
1182 static int lastmaxdepth = 0;
1183 if (val > lastmaxdepth) {
1184 LYXERR(Debug::INFO, "Depth reached " << val);
1187 if (val > maxdepthidx) {
1189 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1194 // Not needed, because borders are now dynamically expanded
1195 static void checkIgnoreIdx(int val)
1197 static int lastmaxignore = -1;
1198 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1199 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1200 lastmaxignore = val;
1206 * Expand the region of ignored parts of the input latex string
1207 * The region is only relevant in output()
1209 void Intervall::addIntervall(int low, int upper)
1212 if (low == upper) return;
1213 for (idx = ignoreidx+1; idx > 0; --idx) {
1214 if (low > borders[idx-1].upper) {
1218 Border br(low, upper);
1219 if (idx > ignoreidx) {
1220 if (borders.size() <= size_t(idx)) {
1221 borders.push_back(br);
1227 // checkIgnoreIdx(ignoreidx);
1231 // Expand only if one of the new bound is inside the interwall
1232 // We know here that br.low > borders[idx-1].upper
1233 if (br.upper < borders[idx].low) {
1234 // We have to insert at this pos
1235 if (size_t(ignoreidx+1) >= borders.size()) {
1236 borders.push_back(borders[ignoreidx]);
1239 borders[ignoreidx+1] = borders[ignoreidx];
1241 for (int i = ignoreidx; i > idx; --i) {
1242 borders[i] = borders[i-1];
1246 // checkIgnoreIdx(ignoreidx);
1249 // Here we know, that we are overlapping
1250 if (br.low > borders[idx].low)
1251 br.low = borders[idx].low;
1252 // check what has to be concatenated
1254 for (int i = idx; i <= ignoreidx; i++) {
1255 if (br.upper >= borders[i].low) {
1257 if (br.upper < borders[i].upper)
1258 br.upper = borders[i].upper;
1264 // count should be >= 1 here
1267 for (int i = idx + count; i <= ignoreidx; i++) {
1268 borders[i-count+1] = borders[i];
1270 ignoreidx -= count - 1;
1276 static void buildaccent(string n, string param, string values)
1280 const char delim = '|';
1281 while (getline(s, name, delim)) {
1283 for (char c : param) {
1284 string key = name + "{" + c + "}";
1285 // get the corresponding utf8-value
1286 if ((values[start] & 0xc0) != 0xc0) {
1287 // should not happen, utf8 encoding starts at least with 11xxxxxx
1288 // but value for '\dot{i}' is 'i', which is ascii
1289 if ((values[start] & 0x80) == 0) {
1291 accents[key] = values.substr(start, 1);
1292 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1297 for (int j = 1; ;j++) {
1298 if (start + j >= values.size()) {
1299 accents[key] = values.substr(start, j);
1300 start = values.size() - 1;
1303 else if ((values[start+j] & 0xc0) != 0x80) {
1304 // This is the first byte of following utf8 char
1305 accents[key] = values.substr(start, j);
1307 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1316 static string getutf8(unsigned uchar)
1319 string ret = string();
1321 if (uchar <= 0x7f) {
1322 c[maxc-1] = uchar & 0x7f;
1325 unsigned char rest = 0x40;
1326 unsigned char first = 0x80;
1328 for (int i = start; i >=0; --i) {
1330 c[i] = first + uchar;
1333 c[i] = 0x80 | (uchar & 0x3f);
1340 for (int i = 0; i < maxc; i++) {
1341 if (c[i] == 0) continue;
1347 static void buildAccentsMap()
1349 accents["imath"] = "ı";
1351 accents["jmath"] = "ȷ";
1352 accents["cdot"] = "·";
1353 accents["textasciicircum"] = "^";
1354 accents["mathcircumflex"] = "^";
1355 accents["sim"] = "~";
1356 accents["guillemotright"] = "»";
1357 accents["guillemotleft"] = "«";
1358 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1359 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1360 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1361 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1362 accents["negmedspace"] = getutf8(0xf0005);
1363 accents["thickspace"] = getutf8(0xf0006);
1364 accents["negthickspace"] = getutf8(0xf0007);
1365 accents["lyx"] = getutf8(0xf0010); // Used logos
1366 accents["LyX"] = getutf8(0xf0010);
1367 accents["tex"] = getutf8(0xf0011);
1368 accents["TeX"] = getutf8(0xf0011);
1369 accents["latex"] = getutf8(0xf0012);
1370 accents["LaTeX"] = getutf8(0xf0012);
1371 accents["latexe"] = getutf8(0xf0013);
1372 accents["LaTeXe"] = getutf8(0xf0013);
1373 accents["lyxarrow"] = getutf8(0xf0020);
1374 accents["braceleft"] = getutf8(0xf0030);
1375 accents["braceright"] = getutf8(0xf0031);
1376 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1377 accents["backslash LyX"] = getutf8(0xf0010);
1378 accents["backslash tex"] = getutf8(0xf0011);
1379 accents["backslash TeX"] = getutf8(0xf0011);
1380 accents["backslash latex"] = getutf8(0xf0012);
1381 accents["backslash LaTeX"] = getutf8(0xf0012);
1382 accents["backslash latexe"] = getutf8(0xf0013);
1383 accents["backslash LaTeXe"] = getutf8(0xf0013);
1384 accents["backslash lyxarrow"] = getutf8(0xf0020);
1385 accents["ddot{\\imath}"] = "ï";
1386 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1387 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1388 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1389 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1390 accents["acute{\\imath}"] = "í";
1391 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1392 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1393 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1394 buildaccent("mathring|r", "aAuUwy",
1396 accents["check{\\imath}"] = "ǐ";
1397 accents["check{\\jmath}"] = "ǰ";
1398 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1399 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1400 accents["hat{\\imath}"] = "î";
1401 accents["hat{\\jmath}"] = "ĵ";
1402 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1403 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1404 accents["bar{\\imath}"] = "ī";
1405 buildaccent("bar|=", "aAeEiIoOuUyY",
1406 "āĀēĒīĪōŌūŪȳȲ"); // macron
1407 accents["tilde{\\imath}"] = "ĩ";
1408 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1409 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1410 accents["breve{\\imath}"] = "ĭ";
1411 buildaccent("breve|u", "aAeEgGiIoOuU",
1412 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1413 accents["grave{\\imath}"] = "ì";
1414 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1415 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1416 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1417 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1418 buildaccent("ogonek|k", "AaEeIiUuOo",
1419 "ĄąĘęĮįŲųǪǫ"); // ogonek
1420 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1421 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1422 buildaccent("subring|textsubring", "Aa",
1424 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1425 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1426 buildaccent("subtilde|textsubtilde", "EeIiUu",
1427 "ḚḛḬḭṴṵ"); // subtilde
1428 accents["dgrave{\\imath}"] = "ȉ";
1429 accents["textdoublegrave{\\i}"] = "ȉ";
1430 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1431 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1432 accents["rcap{\\imath}"] = "ȉ";
1433 accents["textroundcap{\\i}"] = "ȉ";
1434 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1435 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1436 buildaccent("slashed", "oO",
1441 * Created accents in math or regexp environment
1442 * are macros, but we need the utf8 equivalent
1444 void Intervall::removeAccents()
1446 if (accents.empty())
1448 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1449 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1450 "|((i|imath|jmath|cdot|[a-z]+space)|((backslash )?([lL]y[xX]|[tT]e[xX]|[lL]a[tT]e[xX]e?|lyxarrow))|(brace|guillemot)(left|right)|textasciicircum|mathcircumflex|sim)(?![a-zA-Z]))");
1452 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1454 string key = sub.str(1);
1455 if (accents.find(key) != accents.end()) {
1456 string val = accents[key];
1457 size_t pos = sub.position(size_t(0));
1458 for (size_t i = 0; i < val.size(); i++) {
1459 par[pos+i] = val[i];
1461 // Remove possibly following space too
1462 if (par[pos+sub.str(0).size()] == ' ')
1463 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1465 addIntervall(pos+val.size(), pos + sub.str(0).size());
1466 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1467 // remove traces of any remaining chars
1472 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1477 void Intervall::handleOpenP(int i)
1480 depts[actualdeptindex] = i+1;
1481 closes[actualdeptindex] = -1;
1482 checkDepthIndex(actualdeptindex);
1485 void Intervall::handleCloseP(int i, bool closingAllowed)
1487 if (actualdeptindex <= 0) {
1488 if (! closingAllowed)
1489 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1490 // if we are at the very end
1491 addIntervall(i, i+1);
1494 closes[actualdeptindex] = i+1;
1499 void Intervall::resetOpenedP(int openPos)
1501 // Used as initializer for foreignlanguage entry
1502 actualdeptindex = 1;
1503 depts[1] = openPos+1;
1507 int Intervall::previousNotIgnored(int start) const
1509 int idx = 0; /* int intervalls */
1510 for (idx = ignoreidx; idx >= 0; --idx) {
1511 if (start > borders[idx].upper)
1513 if (start >= borders[idx].low)
1514 start = borders[idx].low-1;
1519 int Intervall::nextNotIgnored(int start) const
1521 int idx = 0; /* int intervalls */
1522 for (idx = 0; idx <= ignoreidx; idx++) {
1523 if (start < borders[idx].low)
1525 if (start < borders[idx].upper)
1526 start = borders[idx].upper;
1531 typedef map<string, KeyInfo> KeysMap;
1532 typedef vector< KeyInfo> Entries;
1533 static KeysMap keys = map<string, KeyInfo>();
1539 Intervall interval_;
1540 void buildKeys(bool);
1541 void buildEntries(bool);
1542 void makeKey(const string &, KeyInfo, bool isPatternString);
1543 void processRegion(int start, int region_end); /* remove {} parts */
1544 void removeHead(KeyInfo const &, int count=0);
1547 LatexInfo(string const & par, bool isPatternString)
1548 : entidx_(-1), interval_(isPatternString, par)
1550 buildKeys(isPatternString);
1551 entries_ = vector<KeyInfo>();
1552 buildEntries(isPatternString);
1556 if (entries_.empty()) {
1559 if (entries_[0].keytype == KeyInfo::isTitle) {
1560 interval_.hasTitle = true;
1561 if (! entries_[0].disabled) {
1562 interval_.titleValue = entries_[0].head;
1565 interval_.titleValue = "";
1567 removeHead(entries_[0]);
1568 if (entries_.size() > 1)
1577 if (int(entries_.size()) > entidx_) {
1584 bool setNextKey(int idx) {
1585 if ((idx == entidx_) && (entidx_ >= 0)) {
1592 int find(int start, KeyInfo::KeyType keytype) const {
1596 while (tmpIdx < int(entries_.size())) {
1597 if (entries_[tmpIdx].keytype == keytype)
1603 int process(ostringstream & os, KeyInfo const & actual);
1604 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1605 // string show(int lastpos) { return interval.show(lastpos);};
1606 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1607 KeyInfo &getKeyInfo(int keyinfo) {
1608 static KeyInfo invalidInfo = KeyInfo();
1609 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1612 return entries_[keyinfo];
1614 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1615 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1619 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1623 for (int i = start; i < end; i += 1 + skip) {
1627 if (c == '\\') skip = 1;
1631 else if (c == down) {
1634 if ((repeat <= 0) || (par[i+1] != up))
1652 vector<MathEntry> entries_;
1657 void insert(string const & wait, size_t start, size_t end) {
1658 MathEntry m = MathEntry();
1660 m.mathStart = start;
1662 m.mathSize = end - start;
1663 entries_.push_back(m);
1665 bool empty() const { return entries_.empty(); };
1666 size_t getEndPos() const {
1667 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1670 return entries_[actualIdx_].mathEnd;
1672 size_t getStartPos() const {
1673 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1674 return 100000; /* definitely enough? */
1676 return entries_[actualIdx_].mathStart;
1678 size_t getFirstPos() {
1680 return getStartPos();
1682 size_t getSize() const {
1683 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1686 return entries_[actualIdx_].mathSize;
1688 void incrEntry() { actualIdx_++; };
1691 void LatexInfo::buildEntries(bool isPatternString)
1693 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\})");
1694 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1695 static bool disableLanguageOverride = false;
1696 smatch sub, submath;
1697 bool evaluatingRegexp = false;
1699 bool evaluatingMath = false;
1700 bool evaluatingCode = false;
1702 bool evaluatingOptional = false;
1703 size_t optionalEnd = 0;
1706 bool math_end_waiting = false;
1707 size_t math_pos = 10000;
1709 static vector<string> usedText = vector<string>();
1711 interval_.removeAccents();
1713 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1715 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1716 // prefixed by odd count of '\\'
1719 if (math_end_waiting) {
1720 size_t pos = submath.position(size_t(2));
1721 if ((math_end == "$") &&
1722 (submath.str(2) == "$")) {
1723 mi.insert("$", math_pos, pos + 1);
1724 math_end_waiting = false;
1726 else if ((math_end == "\\]") &&
1727 (submath.str(2) == "\\]")) {
1728 mi.insert("\\]", math_pos, pos + 2);
1729 math_end_waiting = false;
1731 else if ((submath.str(3).compare("end") == 0) &&
1732 (submath.str(4).compare(math_end) == 0)) {
1733 mi.insert(math_end, math_pos, pos + submath.str(2).length());
1734 math_end_waiting = false;
1740 if (submath.str(3).compare("begin") == 0) {
1741 math_end_waiting = true;
1742 math_end = submath.str(4);
1743 math_pos = submath.position(size_t(2));
1745 else if (submath.str(2).compare("\\[") == 0) {
1746 math_end_waiting = true;
1748 math_pos = submath.position(size_t(2));
1750 else if (submath.str(2) == "$") {
1751 size_t pos = submath.position(size_t(2));
1752 math_end_waiting = true;
1758 // Ignore language if there is math somewhere in pattern-string
1759 if (isPatternString) {
1760 for (auto s: usedText) {
1761 // Remove entries created in previous search runs
1764 usedText = vector<string>();
1767 keys["foreignlanguage"].disabled = true;
1768 disableLanguageOverride = true;
1771 disableLanguageOverride = false;
1774 if (disableLanguageOverride) {
1775 keys["foreignlanguage"].disabled = true;
1778 math_pos = mi.getFirstPos();
1779 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1781 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1782 // prefixed by odd count of '\\'
1785 string key = sub.str(5);
1787 if (sub.str(2)[0] == '\\')
1788 key = sub.str(2)[1];
1793 if (keys.find(key) != keys.end()) {
1794 if (keys[key].keytype == KeyInfo::headRemove) {
1795 KeyInfo found1 = keys[key];
1796 found1.disabled = true;
1797 found1.head = "\\" + key + "{";
1798 found1._tokenstart = sub.position(size_t(2));
1799 found1._tokensize = found1.head.length();
1800 found1._dataStart = found1._tokenstart + found1.head.length();
1801 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1802 found1._dataEnd = endpos;
1807 if (evaluatingRegexp) {
1808 if (sub.str(3).compare("endregexp") == 0) {
1809 evaluatingRegexp = false;
1810 // found._tokenstart already set
1811 found._dataEnd = sub.position(size_t(2)) + 13;
1812 found._dataStart = found._dataEnd;
1813 found._tokensize = found._dataEnd - found._tokenstart;
1814 found.parenthesiscount = 0;
1815 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1822 if (evaluatingMath) {
1823 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1825 evaluatingMath = false;
1827 math_pos = mi.getStartPos();
1829 if (keys.find(key) == keys.end()) {
1830 found = KeyInfo(KeyInfo::isStandard, 0, true);
1831 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1832 found = KeyInfo(KeyInfo::isText, 0, false);
1833 if (isPatternString) {
1834 found.keytype = KeyInfo::isChar;
1835 found.disabled = false;
1839 usedText.push_back(key);
1843 if (key.compare("regexp") == 0) {
1844 evaluatingRegexp = true;
1845 found._tokenstart = sub.position(size_t(2));
1846 found._tokensize = 0;
1850 // Handle the other params of key
1851 if (found.keytype == KeyInfo::isIgnored)
1853 else if (found.keytype == KeyInfo::isMath) {
1854 if (size_t(sub.position(size_t(2))) == math_pos) {
1856 found._tokenstart = sub.position(size_t(2));
1857 found._tokensize = mi.getSize();
1858 found._dataEnd = found._tokenstart + found._tokensize;
1859 found._dataStart = found._dataEnd;
1860 found.parenthesiscount = 0;
1861 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1862 evaluatingMath = true;
1865 // begin|end of unknown env, discard
1866 // First handle tables
1867 // longtable|tabular
1868 bool discardComment;
1870 found.keytype = KeyInfo::doRemove;
1871 if ((sub.str(7).compare("longtable") == 0) ||
1872 (sub.str(7).compare("tabular") == 0)) {
1873 discardComment = true; /* '%' */
1876 discardComment = false;
1877 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1879 string token = sub.str(7);
1880 if (regex_match(token, sub2, removeArgs)) {
1881 found.keytype = KeyInfo::removeWithArg;
1884 // discard spaces before pos(2)
1885 int pos = sub.position(size_t(2));
1887 for (count = 0; pos - count > 0; count++) {
1888 char c = interval_.par[pos-count-1];
1889 if (discardComment) {
1890 if ((c != ' ') && (c != '%'))
1896 found._tokenstart = pos - count;
1897 if (sub.str(3).compare(0, 5, "begin") == 0) {
1898 size_t pos1 = pos + sub.str(2).length();
1899 if (sub.str(7).compare("cjk") == 0) {
1900 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1901 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1903 found.keytype = KeyInfo::isMain;
1904 found._dataStart = pos1;
1905 found._dataEnd = interval_.par.length();
1906 found.disabled = keys["foreignlanguage"].disabled;
1907 found.used = keys["foreignlanguage"].used;
1908 found._tokensize = pos1 - found._tokenstart;
1909 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1912 // Swallow possible optional params
1913 while (interval_.par[pos1] == '[') {
1914 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1916 // Swallow also the eventual parameter
1917 if (interval_.par[pos1] == '{') {
1918 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1921 found._dataEnd = pos1;
1923 found._dataStart = found._dataEnd;
1924 found._tokensize = count + found._dataEnd - pos;
1925 found.parenthesiscount = 0;
1926 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1927 found.disabled = true;
1931 // Handle "\end{...}"
1932 found._dataStart = pos + sub.str(2).length();
1933 found._dataEnd = found._dataStart;
1934 found._tokensize = count + found._dataEnd - pos;
1935 found.parenthesiscount = 0;
1936 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1937 found.disabled = true;
1941 else if (found.keytype != KeyInfo::isRegex) {
1942 found._tokenstart = sub.position(size_t(2));
1943 if (found.parenthesiscount == 0) {
1944 // Probably to be discarded
1945 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
1946 char following = interval_.par[following_pos];
1947 if (following == ' ')
1948 found.head = "\\" + sub.str(5) + " ";
1949 else if (following == '=') {
1950 // like \uldepth=1000pt
1951 found.head = sub.str(2);
1954 found.head = "\\" + key;
1955 found._tokensize = found.head.length();
1956 found._dataEnd = found._tokenstart + found._tokensize;
1957 found._dataStart = found._dataEnd;
1960 int params = found._tokenstart + key.length() + 1;
1961 if (evaluatingOptional) {
1962 if (size_t(found._tokenstart) > optionalEnd) {
1963 evaluatingOptional = false;
1966 found.disabled = true;
1969 int optend = params;
1970 while (interval_.par[optend] == '[') {
1971 // discard optional parameters
1972 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
1974 if (optend > params) {
1975 key += interval_.par.substr(params, optend-params);
1976 evaluatingOptional = true;
1977 optionalEnd = optend;
1978 if (found.keytype == KeyInfo::isSectioning) {
1979 // Remove optional values (but still keep in header)
1980 interval_.addIntervall(params, optend);
1983 string token = sub.str(7);
1985 if (interval_.par[optend] != '{') {
1987 found.parenthesiscount = 0;
1988 found.head = "\\" + key;
1991 closings = found.parenthesiscount;
1992 if (found.parenthesiscount == 1) {
1993 found.head = "\\" + key + "{";
1995 else if (found.parenthesiscount > 1) {
1997 found.head = sub.str(2) + "{";
1998 closings = found.parenthesiscount - 1;
2001 found.head = "\\" + key + "{";
2004 found._tokensize = found.head.length();
2005 found._dataStart = found._tokenstart + found.head.length();
2006 if (found.keytype == KeyInfo::doRemove) {
2008 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2009 if (endpar >= interval_.par.length())
2010 found._dataStart = interval_.par.length();
2012 found._dataStart = endpar;
2013 found._tokensize = found._dataStart - found._tokenstart;
2016 found._dataStart = found._tokenstart + found._tokensize;
2020 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2021 found._dataStart += 15;
2025 endpos = found._dataStart - 1;
2027 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2028 if (found.keytype == KeyInfo::isList) {
2029 // Check if it really is list env
2030 static regex const listre("^([a-z]+)$");
2032 if (!regex_match(token, sub2, listre)) {
2033 // Change the key of this entry. It is not in a list/item environment
2034 found.keytype = KeyInfo::endArguments;
2037 if (found.keytype == KeyInfo::noMain) {
2038 evaluatingCode = true;
2040 codeStart = found._dataStart;
2042 else if (evaluatingCode) {
2043 if (size_t(found._dataStart) > codeEnd)
2044 evaluatingCode = false;
2045 else if (found.keytype == KeyInfo::isMain) {
2046 // Disable this key, treate it as standard
2047 found.keytype = KeyInfo::isStandard;
2048 found.disabled = true;
2049 if ((codeEnd +1 >= interval_.par.length()) &&
2050 (found._tokenstart == codeStart)) {
2051 // trickery, because the code inset starts
2052 // with \selectlanguage ...
2054 if (entries_.size() > 1) {
2055 entries_[entries_.size()-1]._dataEnd = codeEnd;
2060 if ((endpos == interval_.par.length()) &&
2061 (found.keytype == KeyInfo::doRemove)) {
2062 // Missing closing => error in latex-input?
2063 // therefore do not delete remaining data
2064 found._dataStart -= 1;
2065 found._dataEnd = found._dataStart;
2068 found._dataEnd = endpos;
2070 if (isPatternString) {
2071 keys[key].used = true;
2074 entries_.push_back(found);
2078 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2080 stringstream s(keysstring);
2082 const char delim = '|';
2083 while (getline(s, key, delim)) {
2084 KeyInfo keyII(keyI);
2085 if (isPatternString) {
2088 else if ( !keys[key].used)
2089 keyII.disabled = true;
2094 void LatexInfo::buildKeys(bool isPatternString)
2097 static bool keysBuilt = false;
2098 if (keysBuilt && !isPatternString) return;
2100 // Keys to ignore in any case
2101 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2102 // Known standard keys with 1 parameter.
2103 // Split is done, if not at start of region
2104 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2105 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2106 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2107 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2108 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2109 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2111 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2112 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2113 makeKey("section*|subsection*|subsubsection*|paragraph*",
2114 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2115 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2116 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2118 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2120 // Split is done, if not at start of region
2121 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2122 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2124 // Split is done always.
2125 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2129 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2130 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2131 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2132 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2134 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2135 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2136 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2137 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2139 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2140 // Custom space/skip, remove the content (== length value)
2141 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2142 // Found in fr/UserGuide.lyx
2143 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2145 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2146 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2147 // Known macros to remove (including their parameter)
2149 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2150 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2151 // handle like standard keys with 1 parameter.
2152 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2154 // Ignore deleted text
2155 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2156 // but preserve added text
2157 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2159 // Macros to remove, but let the parameter survive
2161 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2163 // Remove language spec from content of these insets
2164 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2166 // Same effect as previous, parameter will survive (because there is no one anyway)
2168 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2169 // Remove table decorations
2170 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2171 // Discard shape-header.
2172 // For footnote or shortcut too, because of lang settings
2173 // and wrong handling if used 'KeyInfo::noMain'
2174 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2175 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2176 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2177 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2178 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2179 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2180 // like ('tiny{}' or '\tiny ' ... )
2181 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2183 // Survives, like known character
2184 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2185 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2186 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2188 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2189 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2190 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2192 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2193 // Remove RTL/LTR marker
2194 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2195 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2196 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2197 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2198 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2199 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2200 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2201 if (isPatternString) {
2202 // Allow the first searched string to rebuild the keys too
2206 // no need to rebuild again
2212 * Keep the list of actual opened parentheses actual
2213 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2215 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2218 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2222 if (c == '\\') skip = 1;
2223 else if (c == '{') {
2226 else if (c == '}') {
2227 handleCloseP(i, closingAllowed);
2233 string Intervall::show(int lastpos)
2235 int idx = 0; /* int intervalls */
2238 for (idx = 0; idx <= ignoreidx; idx++) {
2239 while (i < lastpos) {
2241 if (i <= borders[idx].low) {
2242 if (borders[idx].low > lastpos)
2243 printsize = lastpos - i;
2245 printsize = borders[idx].low - i;
2246 s += par.substr(i, printsize);
2248 if (i >= borders[idx].low)
2249 i = borders[idx].upper;
2252 i = borders[idx].upper;
2258 s += par.substr(i, lastpos-i);
2264 void Intervall::output(ostringstream &os, int lastpos)
2266 // get number of chars to output
2267 int idx = 0; /* int intervalls */
2270 string startTitle = titleValue;
2271 for (idx = 0; idx <= ignoreidx; idx++) {
2273 if (i <= borders[idx].low) {
2275 if (borders[idx].low > lastpos)
2276 printsize = lastpos - i;
2278 printsize = borders[idx].low - i;
2279 if (printsize > 0) {
2280 os << startTitle << par.substr(i, printsize);
2282 printed += printsize;
2285 handleParentheses(i, false);
2286 if (i >= borders[idx].low)
2287 i = borders[idx].upper;
2290 i = borders[idx].upper;
2297 os << startTitle << par.substr(i, lastpos-i);
2298 printed += lastpos-i;
2300 handleParentheses(lastpos, false);
2302 if (keys["foreignlanguage"].disabled)
2303 startindex = actualdeptindex-langcount;
2305 startindex = actualdeptindex;
2306 for (int i = startindex; i > 0; --i) {
2309 if (hasTitle && (printed > 0))
2311 if (! isPatternString_)
2313 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2316 void LatexInfo::processRegion(int start, int region_end)
2318 while (start < region_end) { /* Let {[} and {]} survive */
2319 int cnt = interval_.isOpeningPar(start);
2321 // Closing is allowed past the region
2322 int closing = interval_.findclosing(start+1, interval_.par.length());
2323 interval_.addIntervall(start, start+1);
2324 interval_.addIntervall(closing, closing+1);
2328 start = interval_.nextNotIgnored(start+1);
2332 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2334 if (actual.parenthesiscount == 0) {
2335 // "{\tiny{} ...}" ==> "{{} ...}"
2336 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2339 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2340 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2341 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2345 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2348 switch (actual.keytype)
2350 case KeyInfo::isTitle: {
2352 nextKeyIdx = getNextKey();
2355 case KeyInfo::cleanToStart: {
2356 actual._dataEnd = actual._dataStart;
2357 nextKeyIdx = getNextKey();
2358 // Search for end of arguments
2359 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2361 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2362 entries_[i].disabled = true;
2364 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2366 while (interval_.par[actual._dataEnd] == ' ')
2368 interval_.addIntervall(0, actual._dataEnd+1);
2369 interval_.actualdeptindex = 0;
2370 interval_.depts[0] = actual._dataEnd+1;
2371 interval_.closes[0] = -1;
2374 case KeyInfo::isText:
2375 interval_.par[actual._tokenstart] = '#';
2376 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2377 nextKeyIdx = getNextKey();
2379 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2380 if (actual.disabled)
2381 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2383 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2386 case KeyInfo::isChar: {
2387 nextKeyIdx = getNextKey();
2390 case KeyInfo::isSize: {
2391 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2392 if (actual.parenthesiscount == 0)
2393 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2395 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2397 nextKeyIdx = getNextKey();
2399 // Here _dataStart points to '{', so correct it
2400 actual._dataStart += 1;
2401 actual._tokensize += 1;
2402 actual.parenthesiscount = 1;
2403 if (interval_.par[actual._dataStart] == '}') {
2404 // Determine the end if used like '{\tiny{}...}'
2405 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2406 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2409 // Determine the end if used like '\tiny{...}'
2410 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2412 // Split on this key if not at start
2413 int start = interval_.nextNotIgnored(previousStart);
2414 if (start < actual._tokenstart) {
2415 interval_.output(os, actual._tokenstart);
2416 interval_.addIntervall(start, actual._tokenstart);
2418 // discard entry if at end of actual
2419 nextKeyIdx = process(os, actual);
2423 case KeyInfo::endArguments: {
2424 // Remove trailing '{}' too
2425 actual._dataStart += 1;
2426 actual._dataEnd += 1;
2427 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2428 nextKeyIdx = getNextKey();
2431 case KeyInfo::noMain:
2433 case KeyInfo::isStandard: {
2434 if (actual.disabled) {
2436 processRegion(actual._dataStart, actual._dataStart+1);
2437 nextKeyIdx = getNextKey();
2439 // Split on this key if not at datastart of calling entry
2440 int start = interval_.nextNotIgnored(previousStart);
2441 if (start < actual._tokenstart) {
2442 interval_.output(os, actual._tokenstart);
2443 interval_.addIntervall(start, actual._tokenstart);
2445 // discard entry if at end of actual
2446 nextKeyIdx = process(os, actual);
2450 case KeyInfo::removeWithArg: {
2451 nextKeyIdx = getNextKey();
2452 // Search for end of arguments
2453 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2455 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2456 entries_[i].disabled = true;
2458 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2460 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2463 case KeyInfo::doRemove: {
2464 // Remove the key with all parameters and following spaces
2467 if (interval_.par[actual._dataEnd-1] == ' ')
2468 start = actual._dataEnd;
2470 start = actual._dataEnd+1;
2471 for (pos = start; pos < interval_.par.length(); pos++) {
2472 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2475 // Remove also enclosing parentheses [] and {}
2478 while (actual._tokenstart > numpars) {
2479 if (pos+numpars >= interval_.par.size())
2481 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2483 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2489 if (interval_.par[pos+numpars] == ' ')
2493 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2494 nextKeyIdx = getNextKey();
2497 case KeyInfo::isList: {
2498 // Discard space before _tokenstart
2500 for (count = 0; count < actual._tokenstart; count++) {
2501 if (interval_.par[actual._tokenstart-count-1] != ' ')
2504 nextKeyIdx = getNextKey();
2505 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2507 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2509 // How else can we catch this one?
2510 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2511 entries_[i].disabled = true;
2513 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2515 else if (nextKeyIdx > 0) {
2516 // Ignore any lang entries inside data region
2517 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2518 if (entries_[i].keytype == KeyInfo::isMain)
2519 entries_[i].disabled = true;
2522 if (actual.disabled) {
2523 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2526 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2528 if (interval_.par[actual._dataEnd+1] == '[') {
2529 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2530 if ((interval_.par[actual._dataEnd+2] == '{') &&
2531 (interval_.par[posdown-1] == '}')) {
2532 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2533 interval_.addIntervall(posdown-1, posdown+1);
2536 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2537 interval_.addIntervall(posdown, posdown+1);
2539 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2540 if (blk > posdown) {
2541 // Discard at most 1 space after empty item
2543 for (count = 0; count < 1; count++) {
2544 if (interval_.par[blk+count] != ' ')
2548 interval_.addIntervall(blk, blk+count);
2553 case KeyInfo::isSectioning: {
2554 // Discard spaces before _tokenstart
2556 int val = actual._tokenstart;
2557 for (count = 0; count < actual._tokenstart;) {
2558 val = interval_.previousNotIgnored(val-1);
2559 if (val < 0 || interval_.par[val] != ' ')
2562 count = actual._tokenstart - val;
2565 if (actual.disabled) {
2566 removeHead(actual, count);
2567 nextKeyIdx = getNextKey();
2569 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2570 nextKeyIdx = process(os, actual);
2574 case KeyInfo::isMath: {
2575 // Same as regex, use the content unchanged
2576 nextKeyIdx = getNextKey();
2579 case KeyInfo::isRegex: {
2580 // DO NOT SPLIT ON REGEX
2582 nextKeyIdx = getNextKey();
2585 case KeyInfo::isIgnored: {
2586 // Treat like a character for now
2587 nextKeyIdx = getNextKey();
2590 case KeyInfo::isMain: {
2591 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2592 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2593 if (actual._tokenstart > 0) {
2594 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2595 if ((prev >= 0) && interval_.par[prev] == '%')
2596 interval_.addIntervall(prev, prev+1);
2598 if (actual.disabled) {
2600 interval_.langcount++;
2601 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2602 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2603 // Discard also the space before math-equation
2604 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2606 nextKeyIdx = getNextKey();
2607 // interval.resetOpenedP(actual._dataStart-1);
2610 if (actual._tokenstart < 26) {
2611 // for the first (and maybe dummy) language
2612 interval_.setForDefaultLang(actual);
2614 interval_.resetOpenedP(actual._dataStart-1);
2618 case KeyInfo::invalid:
2619 case KeyInfo::headRemove:
2620 // These two cases cannot happen, already handled
2623 // LYXERR(Debug::INFO, "Unhandled keytype");
2624 nextKeyIdx = getNextKey();
2631 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2633 int end = interval_.nextNotIgnored(actual._dataEnd);
2634 int oldStart = actual._dataStart;
2635 int nextKeyIdx = getNextKey();
2637 if ((nextKeyIdx < 0) ||
2638 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2639 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2640 if (oldStart <= end) {
2641 processRegion(oldStart, end);
2646 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2648 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2649 (void) dispatch(os, actual._dataStart, nextKey);
2650 end = nextKey._tokenstart;
2653 processRegion(oldStart, nextKey._tokenstart);
2654 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2656 oldStart = nextKey._dataEnd+1;
2658 // now nextKey is either invalid or is outside of actual._dataEnd
2659 // output the remaining and discard myself
2660 if (oldStart <= end) {
2661 processRegion(oldStart, end);
2663 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2665 // This is the normal case.
2666 // But if using the firstlanguage, the closing may be missing
2668 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2670 if (actual._dataEnd < end)
2671 output_end = interval_.nextNotIgnored(actual._dataEnd);
2672 else if (interval_.par.size() > (size_t) end)
2673 output_end = interval_.nextNotIgnored(end);
2675 output_end = interval_.par.size();
2676 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2677 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2679 // Remove possible empty data
2680 int dstart = interval_.nextNotIgnored(actual._dataStart);
2681 while (interval_.isOpeningPar(dstart) == 1) {
2682 interval_.addIntervall(dstart, dstart+1);
2683 int dend = interval_.findclosing(dstart+1, output_end);
2684 interval_.addIntervall(dend, dend+1);
2685 dstart = interval_.nextNotIgnored(dstart+1);
2687 if (dstart < output_end)
2688 interval_.output(os, output_end);
2689 interval_.addIntervall(actual._tokenstart, end);
2693 string splitOnKnownMacros(string par, bool isPatternString)
2696 LatexInfo li(par, isPatternString);
2697 // LYXERR(Debug::INFO, "Berfore split: " << par);
2698 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2700 DummyKey._tokensize = 0;
2701 DummyKey._dataStart = 0;
2702 DummyKey._dataEnd = par.length();
2703 DummyKey.disabled = true;
2704 int firstkeyIdx = li.getFirstKey();
2706 if (firstkeyIdx >= 0) {
2707 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2708 DummyKey._tokenstart = firstKey._tokenstart;
2710 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2711 // Use dummy firstKey
2712 firstKey = DummyKey;
2713 (void) li.setNextKey(firstkeyIdx);
2716 if (par.substr(firstKey._dataStart, 2) == "% ")
2717 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2719 nextkeyIdx = li.process(os, firstKey);
2720 while (nextkeyIdx >= 0) {
2721 // Check for a possible gap between the last
2722 // entry and this one
2723 int datastart = li.nextNotIgnored(firstKey._dataStart);
2724 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2725 if ((nextKey._tokenstart > datastart)) {
2727 firstKey._dataStart = datastart;
2728 firstKey._dataEnd = par.length();
2729 (void) li.setNextKey(nextkeyIdx);
2730 // Fake the last opened parenthesis
2731 li.setForDefaultLang(firstKey);
2732 nextkeyIdx = li.process(os, firstKey);
2735 if (nextKey.keytype != KeyInfo::isMain) {
2736 firstKey._dataStart = datastart;
2737 firstKey._dataEnd = nextKey._dataEnd+1;
2738 (void) li.setNextKey(nextkeyIdx);
2739 li.setForDefaultLang(firstKey);
2740 nextkeyIdx = li.process(os, firstKey);
2743 nextkeyIdx = li.process(os, nextKey);
2747 // Handle the remaining
2748 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2749 firstKey._dataEnd = par.length();
2751 if ((firstKey._dataStart < firstKey._dataEnd) &&
2752 (par[firstKey._dataStart] != '}')) {
2753 li.setForDefaultLang(firstKey);
2754 (void) li.process(os, firstKey);
2758 // return string definitelly impossible to match
2759 s = "\\foreignlanguage{ignore}{ }";
2763 s = par; /* no known macros found */
2764 // LYXERR(Debug::INFO, "After split: " << s);
2769 * Try to unify the language specs in the latexified text.
2770 * Resulting modified string is set to "", if
2771 * the searched tex does not contain all the features in the search pattern
2773 static string correctlanguagesetting(string par, bool isPatternString, bool withformat, lyx::Buffer *pbuf = nullptr)
2775 static Features regex_f;
2776 static int missed = 0;
2777 static bool regex_with_format = false;
2779 int parlen = par.length();
2781 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2784 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2785 // Happens to be there in case of description or labeling environment
2790 // Split the latex input into pieces which
2791 // can be digested by our search engine
2792 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2793 if (isPatternString && (pbuf != nullptr)) { // Check if we should disable/enable test for language
2794 // We check for polyglossia, because in runparams.flavor we use Flavor::XeTeX
2795 string doclang = pbuf->params().language->polyglossia();
2796 static regex langre("\\\\(foreignlanguage)\\{([^\\}]+)\\}");
2798 bool toIgnoreLang = true;
2799 for (sregex_iterator it(par.begin(), par.end(), langre), end; it != end; ++it) {
2801 if (sub.str(2) != doclang) {
2802 toIgnoreLang = false;
2806 setIgnoreFormat("language", toIgnoreLang);
2809 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2810 LYXERR(Debug::FIND, "After splitOnKnownMacros:\n\"" << result << "\"");
2813 result = par.substr(0, parlen);
2814 if (isPatternString) {
2817 regex_f = identifyFeatures(result);
2818 string features = "";
2819 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2820 string a = it->first;
2821 regex_with_format = true;
2822 features += " " + a;
2823 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2825 LYXERR(Debug::FIND, "Identified Features" << features);
2828 } else if (regex_with_format) {
2829 Features info = identifyFeatures(result);
2830 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2831 string a = it->first;
2832 bool b = it->second;
2833 if (b && ! info[a]) {
2835 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2842 // LYXERR(Debug::INFO, "No regex formats");
2848 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2849 static int identifyClosing(string & t)
2851 int open_braces = 0;
2853 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2854 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2856 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2858 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2860 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2869 static int num_replaced = 0;
2870 static bool previous_single_replace = true;
2872 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
2875 // Handle \w properly
2876 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
2877 if (! opt.casesensitive) {
2878 popts |= QRegularExpression::CaseInsensitiveOption;
2880 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
2881 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
2883 if (regexp.isValid() && regexp2.isValid()) {
2884 regexIsValid = true;
2885 // Check '{', '}' pairs inside the regex
2888 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
2889 char c = par_as_string[i];
2896 else if (c == '}') {
2903 if (balanced != 0) {
2904 regexIsValid = false;
2905 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
2909 regexIsValid = false;
2910 if (!regexp.isValid())
2911 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
2913 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
2916 if (opt.casesensitive) {
2917 regexp = regex(regexp_str);
2918 regexp2 = regex(regexp2_str);
2921 regexp = regex(regexp_str, std::regex_constants::icase);
2922 regexp2 = regex(regexp2_str, std::regex_constants::icase);
2927 static void modifyRegexForMatchWord(string &t)
2930 regex wordre("(\\\\)*((\\.|\\\\b))");
2933 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
2935 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
2938 else if (sub.str(2) == "\\\\b")
2940 if (lastpos < (size_t) sub.position(2))
2941 s += t.substr(lastpos, sub.position(2) - lastpos);
2943 lastpos = sub.position(2) + sub.length(2);
2946 s = "\\b" + t + "\\b";
2950 else if (lastpos < t.length())
2951 s += t.substr(lastpos, t.length() - lastpos);
2952 t = "\\b" + s + "\\b";
2955 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
2956 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2958 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2959 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2960 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2961 if (opt.replace_all && previous_single_replace) {
2962 previous_single_replace = false;
2965 else if (!opt.replace_all) {
2966 num_replaced = 0; // count number of replaced strings
2967 previous_single_replace = true;
2969 // When using regexp, braces are hacked already by escape_for_regex()
2970 par_as_string = normalize(ds);
2972 close_wildcards = 0;
2974 size_t lead_size = 0;
2975 // correct the language settings
2976 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat, &buf);
2977 opt.matchAtStart = false;
2979 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
2980 if (opt.ignoreformat) {
2984 lead_size = identifyLeading(par_as_string);
2986 lead_as_string = par_as_string.substr(0, lead_size);
2987 string lead_as_regex_string = string2regex(lead_as_string);
2988 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2989 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
2990 /* Handle whole words too in this case
2992 if (opt.matchword) {
2993 par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
2994 opt.matchword = false;
2996 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
2997 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
2998 CreateRegexp(opt, regexp_str, regexp2_str);
3000 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3001 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3005 if (!opt.ignoreformat) {
3006 lead_size = identifyLeading(par_as_string);
3007 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
3008 lead_as_string = par_as_string.substr(0, lead_size);
3009 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
3012 // Here we are using regexp
3013 LASSERT(use_regexp, /**/);
3015 string lead_as_regexp;
3016 if (lead_size > 0) {
3017 lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
3018 regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
3019 par_as_string = par_as_string_nolead;
3020 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
3021 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3023 // LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
3024 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
3025 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
3026 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3028 size_t lng = par_as_string.size();
3029 if (!opt.ignoreformat) {
3030 // Remove extra '\}' at end if not part of \{\.\}
3032 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
3034 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3043 if (lng < par_as_string.size())
3044 par_as_string = par_as_string.substr(0,lng);
3046 LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
3047 if ((lng > 0) && (par_as_string[0] == '^')) {
3048 par_as_string = par_as_string.substr(1);
3050 opt.matchAtStart = true;
3052 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3053 // LYXERR(Debug::FIND, "Open braces: " << open_braces);
3054 // LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3056 // If entered regexp must match at begin of searched string buffer
3057 // Kornel: Added parentheses to use $1 for size of the leading string
3061 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3062 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3063 // so the convert has no effect in that case
3064 for (int i = 7; i > 0; --i) {
3065 string orig = "\\\\" + std::to_string(i);
3066 string dest = "\\" + std::to_string(i+2);
3067 while (regex_replace(par_as_string, par_as_string, orig, dest));
3069 if (opt.matchword) {
3070 modifyRegexForMatchWord(par_as_string);
3071 opt.matchword = false;
3073 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3074 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3076 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3077 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3078 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3082 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3086 mres.searched_size = len;
3088 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3091 docstring docstr = stringifyFromForSearch(opt, cur, len);
3093 str = normalize(docstr);
3094 if (!opt.ignoreformat) {
3095 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3096 // remove closing '}' and '\n' to allow for use of '$' in regex
3097 size_t lng = str.size();
3098 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3100 if (lng != str.size()) {
3101 str = str.substr(0, lng);
3105 mres.match_len = -1;
3108 LYXERR(Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
3110 LASSERT(use_regexp, /**/);
3112 // use_regexp always true
3113 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3115 QString qstr = QString::fromStdString(str);
3116 QRegularExpression const *p_regexp;
3117 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3121 p_regexp = ®exp2;
3123 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3124 if (!match.hasMatch())
3127 regex const *p_regexp;
3128 regex_constants::match_flag_type flags;
3130 flags = regex_constants::match_continuous;
3133 flags = regex_constants::match_default;
3134 p_regexp = ®exp2;
3136 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3137 if (re_it == sregex_iterator())
3139 match_results<string::const_iterator> const & m = *re_it;
3141 // Whole found string, including the leading
3142 // std: m[0].second - m[0].first
3143 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3145 // Size of the leading string
3146 // std: m[1].second - m[1].first
3147 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3148 int leadingsize = 0;
3150 if (match.lastCapturedIndex() > 0) {
3151 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3156 leadingsize = m[1].second - m[1].first;
3160 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3161 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3162 // because of different number of closing at end of string
3163 // we have to 'unify' the length of the post-match.
3164 // Done by ignoring closing parenthesis and linefeeds at string end
3165 int matchend = match.capturedEnd(0);
3166 size_t strsize = qstr.size();
3167 if (!opt.ignoreformat) {
3168 while (mres.match_len > 0) {
3169 QChar c = qstr.at(matchend - 1);
3170 if ((c == '\n') || (c == '}') || (c == '{')) {
3177 while (strsize > (size_t) match.capturedEnd(0)) {
3178 QChar c = qstr.at(strsize-1);
3179 if ((c == '\n') || (c == '}')) {
3186 // LYXERR0(qstr.toStdString());
3187 mres.match2end = strsize - matchend;
3188 mres.pos = match.capturedStart(2);
3190 mres.match_prefix = m[2].second - m[2].first;
3191 mres.match_len = m[0].second - m[2].second;
3192 // ignore closing parenthesis and linefeeds at string end
3193 size_t strend = m[0].second - m[0].first;
3194 int matchend = strend;
3195 size_t strsize = str.size();
3196 if (!opt.ignoreformat) {
3197 while (mres.match_len > 0) {
3198 char c = str.at(matchend - 1);
3199 if ((c == '\n') || (c == '}') || (c == '{')) {
3206 while (strsize > strend) {
3207 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3215 mres.match2end = strsize - matchend;
3216 mres.pos = m[2].first - m[0].first;;
3218 if (mres.match2end < 0)
3220 mres.leadsize = leadingsize;
3222 if (mres.match_len > 0) {
3223 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3224 mres.result.push_back(a0);
3225 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3226 mres.result.push_back(match.captured(i).toStdString());
3230 if (mres.match_len > 0) {
3231 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3232 mres.result.push_back(a0);
3233 for (size_t i = 3; i < m.size(); i++) {
3234 mres.result.push_back(m[i]);
3243 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3245 MatchResult mres = findAux(cur, len, at_begin);
3246 int res = mres.match_len;
3248 "res=" << res << ", at_begin=" << at_begin
3249 << ", matchAtStart=" << opt.matchAtStart
3250 << ", inTexted=" << cur.inTexted());
3251 if (opt.matchAtStart) {
3254 else if (mres.match_prefix > 0)
3263 static bool simple_replace(string &t, string from, string to)
3265 regex repl("(\\\\)*(" + from + ")");
3269 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3271 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3273 if (lastpos < (size_t) sub.position(2))
3274 s += t.substr(lastpos, sub.position(2) - lastpos);
3276 lastpos = sub.position(2) + sub.length(2);
3280 else if (lastpos < t.length())
3281 s += t.substr(lastpos, t.length() - lastpos);
3287 string MatchStringAdv::normalize(docstring const & s) const
3290 t = lyx::to_utf8(s);
3291 // Remove \n at begin
3292 while (!t.empty() && t[0] == '\n')
3295 while (!t.empty() && t[t.size() - 1] == '\n')
3296 t = t.substr(0, t.size() - 1);
3298 // Handle all other '\n'
3299 while ((pos = t.find("\n")) != string::npos) {
3300 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3302 if (isAlnumASCII(t[pos+1])) {
3303 t.replace(pos-2, 3, " ");
3306 t.replace(pos-2, 3, "");
3309 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3310 // '\n' adjacent to non-alpha-numerics, discard
3311 t.replace(pos, 1, "");
3314 // Replace all other \n with spaces
3315 t.replace(pos, 1, " ");
3318 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3319 // Kornel: Added textsl, textsf, textit, texttt and noun
3320 // + allow to seach for colored text too
3321 LYXERR(Debug::FIND, "Removing stale empty macros from: " << t);
3322 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3323 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3324 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3325 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3326 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3332 docstring stringifyFromCursor(DocIterator const & cur, int len)
3334 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3335 if (cur.inTexted()) {
3336 Paragraph const & par = cur.paragraph();
3337 // TODO what about searching beyond/across paragraph breaks ?
3338 // TODO Try adding a AS_STR_INSERTS as last arg
3339 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3340 int(par.size()) : cur.pos() + len;
3341 // OutputParams runparams(&cur.buffer()->params().encoding());
3342 OutputParams runparams(encodings.fromLyXName("utf8"));
3343 runparams.nice = true;
3344 runparams.flavor = Flavor::XeTeX;
3345 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3346 // No side effect of file copying and image conversion
3347 runparams.dryrun = true;
3348 int option = AS_STR_INSETS | AS_STR_PLAINTEXT;
3349 if (ignoreFormats.getDeleted()) {
3350 option |= AS_STR_SKIPDELETE;
3351 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
3354 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
3356 LYXERR(Debug::FIND, "Stringifying with cur: "
3357 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3358 return par.asString(cur.pos(), end,
3361 } else if (cur.inMathed()) {
3362 CursorSlice cs = cur.top();
3363 MathData md = cs.cell();
3364 MathData::const_iterator it_end =
3365 (( len == -1 || cs.pos() + len > int(md.size()))
3367 : md.begin() + cs.pos() + len );
3369 for (MathData::const_iterator it = md.begin() + cs.pos();
3372 docstring s = asString(md2);
3373 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3376 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3381 /** Computes the LaTeX export of buf starting from cur and ending len positions
3382 * after cur, if len is positive, or at the paragraph or innermost inset end
3385 docstring latexifyFromCursor(DocIterator const & cur, int len)
3388 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3389 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3390 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3392 Buffer const & buf = *cur.buffer();
3394 odocstringstream ods;
3396 //OutputParams runparams(&buf.params().encoding());
3397 OutputParams runparams(encodings.fromLyXName("utf8"));
3398 runparams.nice = false;
3399 runparams.flavor = Flavor::XeTeX;
3400 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3401 // No side effect of file copying and image conversion
3402 runparams.dryrun = true;
3403 if (ignoreFormats.getDeleted()) {
3404 runparams.for_searchAdv = OutputParams::SearchWithoutDeleted;
3407 runparams.for_searchAdv = OutputParams::SearchWithDeleted;
3410 if (cur.inTexted()) {
3411 // @TODO what about searching beyond/across paragraph breaks ?
3412 pos_type endpos = cur.paragraph().size();
3413 if (len != -1 && endpos > cur.pos() + len)
3414 endpos = cur.pos() + len;
3415 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3416 string(), cur.pos(), endpos);
3417 string s = lyx::to_utf8(ods.str());
3418 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3419 return(lyx::from_utf8(s));
3420 } else if (cur.inMathed()) {
3421 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3422 for (int s = cur.depth() - 1; s >= 0; --s) {
3423 CursorSlice const & cs = cur[s];
3424 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3425 TeXMathStream ws(os);
3426 cs.asInsetMath()->asHullInset()->header_write(ws);
3431 CursorSlice const & cs = cur.top();
3432 MathData md = cs.cell();
3433 MathData::const_iterator it_end =
3434 ((len == -1 || cs.pos() + len > int(md.size()))
3436 : md.begin() + cs.pos() + len);
3438 for (MathData::const_iterator it = md.begin() + cs.pos();
3442 ods << asString(md2);
3443 // Retrieve the math environment type, and add '$' or '$]'
3444 // or others (\end{equation}) accordingly
3445 for (int s = cur.depth() - 1; s >= 0; --s) {
3446 CursorSlice const & cs2 = cur[s];
3447 InsetMath * inset = cs2.asInsetMath();
3448 if (inset && inset->asHullInset()) {
3449 TeXMathStream ws(os);
3450 inset->asHullInset()->footer_write(ws);
3454 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3456 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3461 #if defined(ResultsDebug)
3463 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
3465 LYXERR0( "from:\t\t\t" << from);
3467 if (mres.pos_len > 0) {
3469 status = "FINALSEARCH";
3472 if (mres.match_len > 0) {
3473 if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
3474 status = "Good Match";
3476 status = "Matched in";
3479 status = "MissedSearch";
3482 LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
3483 if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
3484 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
3485 if ((mres.pos > 0) || (mres.match_prefix > 0))
3486 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
3487 for (size_t i = 0; i < mres.result.size(); i++)
3488 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3490 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
3492 #define displayMres(s, txt, cur)
3495 /** Finalize an advanced find operation, advancing the cursor to the innermost
3496 ** position that matches, plus computing the length of the matching text to
3498 ** Return the cur.pos() difference between start and end of found match
3500 MatchResult findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
3502 // Search the foremost position that matches (avoids find of entire math
3503 // inset when match at start of it)
3504 DocIterator old_cur(cur.buffer());
3506 static MatchResult fail = MatchResult();
3507 MatchResult max_match;
3508 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3509 // Happens with e.g. hyperlinks
3510 // either one sees "http://www.bla.bla" or nothing
3511 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3512 // and although we search for only 3 chars, we find the whole hyperlink inset
3513 bool at_begin = (expected.match_prefix == 0);
3514 if (!match.opt.forward && match.opt.ignoreformat) {
3515 if (expected.pos > 0)
3518 LASSERT(at_begin, /**/);
3519 if (expected.match_len > 0 && at_begin) {
3520 // Search for deepest match
3522 max_match = expected;
3524 size_t d = cur.depth();
3528 if (cur.depth() < d)
3530 if (cur.depth() == d)
3533 while (cur && cur.depth() > lastd) {
3534 lastd = cur.depth();
3535 mres = match(cur, -1, at_begin);
3536 displayMres(mres, "Checking innermost", cur);
3537 if (mres.match_len > 0)
3542 if (mres.match_len < expected.match_len)
3550 // (expected.match_len <= 0)
3551 mres = match(cur); /* match valid only if not searching whole words */
3552 displayMres(mres, "Start with negative match", cur);
3555 if (max_match.match_len <= 0) return fail;
3556 LYXERR(Debug::FIND, "Ok");
3558 // Compute the match length
3560 if (cur.pos() + len > cur.lastpos())
3563 LASSERT(match.use_regexp, /**/);
3566 int maxl = cur.lastpos() - cur.pos();
3567 // Greedy behaviour while matching regexps
3568 while (maxl > minl) {
3570 mres2 = match(cur, len, at_begin);
3571 displayMres(mres2, "Finalize loop", cur);
3572 int actual_match_len = mres2.match_len;
3573 if (actual_match_len >= max_match.match_len) {
3574 // actual_match_len > max_match _can_ happen,
3575 // if the search area splits
3576 // some following word so that the regex
3577 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3579 // This means, the len value is too big
3580 actual_match_len = max_match.match_len;
3582 max_match.match_len = actual_match_len;
3584 if (maxl - minl < 4)
3585 len = (int)((maxl + minl)/2);
3587 len = (int)(minl + (maxl - minl + 3)/4);
3590 // (actual_match_len < max_match.match_len)
3592 len = (int)((maxl + minl)/2);
3597 // Search for real start of matched characters
3599 MatchResult actual_match;
3602 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3603 if (cur.depth() < old_cur.depth()) {
3605 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3608 if (cur.pos() != old_cur.pos()) {
3609 // OK, forwarded 1 pos in actual inset
3610 actual_match = match(cur, len-1, at_begin);
3611 if (actual_match.match_len == max_match.match_len) {
3612 // Ha, got it! The shorter selection has the same match length
3615 max_match = actual_match;
3618 // OK, the shorter selection matches less chars, revert to previous value
3624 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3625 actual_match = match(cur, len, at_begin);
3626 if (actual_match.match_len == max_match.match_len) {
3628 max_match = actual_match;
3635 max_match.pos_len = len;
3636 displayMres(max_match, "SEARCH RESULT", cur)
3643 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3647 bool repeat = false;
3648 DocIterator orig_cur; // to be used if repeat not successful
3649 MatchResult orig_mres;
3650 while (!theApp()->longOperationCancelled() && cur) {
3651 //(void) findAdvForwardInnermost(cur);
3652 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3653 MatchResult mres = match(cur, -1, false);
3654 string msg = "Starting";
3657 displayMres(mres, msg + " findForwardAdv", cur)
3658 int match_len = mres.match_len;
3659 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3660 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3663 if (match_len <= 0) {
3664 // This should exit nested insets, if any, or otherwise undefine the currsor.
3665 cur.pos() = cur.lastpos();
3666 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3669 else { // match_len > 0
3670 // Try to find the begin of searched string
3672 int firstInvalid = cur.lastpos() - cur.pos();
3674 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3675 int incrcur = (firstInvalid + 1 )*3/4;
3676 if (incrcur < incrmatch)
3677 increment = incrcur;
3679 increment = incrmatch;
3683 LYXERR(Debug::FIND, "Set increment to " << increment);
3684 while (increment > 0) {
3685 DocIterator old_cur = cur;
3686 if (cur.pos() + increment >= cur.lastpos()) {
3690 cur.pos() = cur.pos() + increment;
3691 MatchResult mres2 = match(cur, -1, false);
3692 displayMres(mres2, "findForwardAdv loop", cur)
3693 switch (interpretMatch(mres, mres2)) {
3694 case MatchResult::newIsTooFar:
3695 // behind the expected match
3696 firstInvalid = increment;
3700 case MatchResult::newIsBetter:
3701 // not reached yet, but cur.pos()+increment is bettert
3703 firstInvalid -= increment;
3704 if (increment > firstInvalid*3/4)
3705 increment = firstInvalid*3/4;
3706 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3707 if (increment >= mres2.match_prefix)
3708 increment = (mres2.match_prefix+1)*3/4;
3713 // Handle not like MatchResult::newIsTooFar
3714 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3716 increment = increment*3/4;
3721 if (mres.match_len > 0) {
3722 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
3723 // The match seems to indicate some deeper level
3732 // should never be reached.
3736 // LYXERR0("Leaving first loop");
3737 LYXERR(Debug::FIND, "Finalizing 1");
3738 MatchResult found_match = findAdvFinalize(cur, match, mres);
3739 if (found_match.match_len > 0) {
3740 LASSERT(found_match.pos_len > 0, /**/);
3741 match.FillResults(found_match);
3742 return found_match.pos_len;
3745 // try next possible match
3756 /// Find the most backward consecutive match within same paragraph while searching backwards.
3757 MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, MatchResult &expected)
3759 DocIterator cur_begin = cur;
3760 cur_begin.pos() = 0;
3761 DocIterator tmp_cur = cur;
3762 MatchResult mr = findAdvFinalize(tmp_cur, match, expected);
3763 Inset & inset = cur.inset();
3764 for (; cur != cur_begin; cur.backwardPos()) {
3765 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3766 DocIterator new_cur = cur;
3767 new_cur.backwardPos();
3768 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3770 MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
3771 if (new_mr.match_len == mr.match_len)
3775 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3781 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3785 // Backup of original position
3786 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3787 if (cur == cur_begin)
3790 DocIterator cur_orig(cur);
3791 bool pit_changed = false;
3794 MatchResult found_match = match(cur, -1, false);
3796 if (found_match.match_len > 0) {
3798 cur.pos() = cur.lastpos();
3800 cur.pos() = cur_orig.pos();
3801 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3802 DocIterator cur_prev_iter;
3804 found_match = match(cur);
3805 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3806 << (found_match.match_len > 0) << ", cur: " << cur);
3807 if (found_match.match_len > 0) {
3808 MatchResult found_mr = findMostBackwards(cur, match, found_match);
3809 if (found_mr.pos_len > 0) {
3810 match.FillResults(found_mr);
3811 return found_mr.pos_len;
3815 // Stop if begin of document reached
3816 if (cur == cur_begin)
3818 cur_prev_iter = cur;
3822 if (cur == cur_begin)
3829 } while (!theApp()->longOperationCancelled());
3837 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3838 DocIterator const & cur, int len)
3840 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3842 if (!opt.ignoreformat)
3843 return latexifyFromCursor(cur, len);
3845 return stringifyFromCursor(cur, len);
3849 FindAndReplaceOptions::FindAndReplaceOptions(
3850 docstring const & _find_buf_name, bool _casesensitive,
3851 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3852 docstring const & _repl_buf_name, bool _keep_case,
3853 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3854 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3855 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3856 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3864 /** Check if 'len' letters following cursor are all non-lowercase */
3865 static bool allNonLowercase(Cursor const & cur, int len)
3867 pos_type beg_pos = cur.selectionBegin().pos();
3868 pos_type end_pos = cur.selectionBegin().pos() + len;
3869 if (len > cur.lastpos() + 1 - beg_pos) {
3870 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3871 len = cur.lastpos() + 1 - beg_pos;
3872 end_pos = beg_pos + len;
3874 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3875 if (isLowerCase(cur.paragraph().getChar(pos)))
3881 /** Check if first letter is upper case and second one is lower case */
3882 static bool firstUppercase(Cursor const & cur)
3885 pos_type pos = cur.selectionBegin().pos();
3886 if (pos >= cur.lastpos() - 1) {
3887 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3890 ch1 = cur.paragraph().getChar(pos);
3891 ch2 = cur.paragraph().getChar(pos + 1);
3892 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3893 LYXERR(Debug::FIND, "firstUppercase(): "
3894 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3895 << ch2 << "(" << char(ch2) << ")"
3896 << ", result=" << result << ", cur=" << cur);
3901 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3903 ** \fixme What to do with possible further paragraphs in replace buffer ?
3905 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3907 ParagraphList::iterator pit = buffer.paragraphs().begin();
3908 LASSERT(!pit->empty(), /**/);
3909 pos_type right = pos_type(1);
3910 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3911 right = pit->size();
3912 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3916 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
3918 // Should replace the string "$" + std::to_string(matchnum) with replacement
3919 // if the char '$' is not prefixed with odd number of char '\\'
3920 static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
3924 for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
3926 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3928 int num = stoi(sub.str(3), nullptr, 10);
3929 if (num >= maxmatchnum)
3931 if (lastpos < (size_t) sub.position(2))
3932 s += t.substr(lastpos, sub.position(2) - lastpos);
3933 s += replacements[num];
3934 lastpos = sub.position(2) + sub.length(2);
3938 else if (lastpos < t.length())
3939 s += t.substr(lastpos, t.length() - lastpos);
3945 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3947 Cursor & cur = bv->cursor();
3948 if (opt.repl_buf_name.empty()
3949 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3950 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3953 DocIterator sel_beg = cur.selectionBegin();
3954 DocIterator sel_end = cur.selectionEnd();
3955 if (&sel_beg.inset() != &sel_end.inset()
3956 || sel_beg.pit() != sel_end.pit()
3957 || sel_beg.idx() != sel_end.idx())
3959 int sel_len = sel_end.pos() - sel_beg.pos();
3960 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3961 << ", sel_len: " << sel_len << endl);
3964 LASSERT(sel_len > 0, return 0);
3966 if (!matchAdv(sel_beg, sel_len).match_len)
3969 // Build a copy of the replace buffer, adapted to the KeepCase option
3970 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3972 repl_buffer_orig.write(oss);
3973 string lyx = oss.str();
3974 if (matchAdv.valid_matches > 0) {
3975 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
3977 Buffer repl_buffer("", false);
3978 repl_buffer.setUnnamed(true);
3979 LASSERT(repl_buffer.readString(lyx), return 0);
3980 if (opt.keep_case && sel_len >= 2) {
3981 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3982 if (cur.inTexted()) {
3983 if (firstUppercase(cur))
3984 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3985 else if (allNonLowercase(cur, sel_len))
3986 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3989 cap::cutSelection(cur, false);
3990 if (cur.inTexted()) {
3991 repl_buffer.changeLanguage(
3992 repl_buffer.language(),
3993 cur.getFont().language());
3994 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3995 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3996 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3997 repl_buffer.params().documentClassPtr(),
3998 bv->buffer().errorList("Paste"));
3999 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
4000 sel_len = repl_buffer.paragraphs().begin()->size();
4001 } else if (cur.inMathed()) {
4002 odocstringstream ods;
4004 // OutputParams runparams(&repl_buffer.params().encoding());
4005 OutputParams runparams(encodings.fromLyXName("utf8"));
4006 runparams.nice = false;
4007 runparams.flavor = Flavor::XeTeX;
4008 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
4009 runparams.dryrun = true;
4010 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
4011 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
4012 docstring repl_latex = ods.str();
4013 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
4015 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
4016 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
4017 repl_latex = from_utf8(s);
4018 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
4019 MathData ar(cur.buffer());
4020 asArray(repl_latex, ar, Parse::NORMAL);
4022 sel_len = ar.size();
4023 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4025 if (cur.pos() >= sel_len)
4026 cur.pos() -= sel_len;
4029 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
4030 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
4031 bv->processUpdateFlags(Update::Force);
4036 /// Perform a FindAdv operation.
4037 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
4042 // e.g., when invoking word-findadv from mini-buffer wither with
4043 // wrong options syntax or before ever opening advanced F&R pane
4044 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4048 MatchStringAdv matchAdv(bv->buffer(), opt);
4050 if (!matchAdv.regexIsValid) {
4051 bv->message(lyx::from_utf8(matchAdv.regexError));
4055 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4057 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4058 num_replaced += findAdvReplace(bv, opt, matchAdv);
4061 pos_len = findForwardAdv(cur, matchAdv);
4063 pos_len = findBackwardsAdv(cur, matchAdv);
4064 } catch (exception & ex) {
4065 bv->message(from_utf8(ex.what()));
4070 if (num_replaced > 0) {
4071 switch (num_replaced)
4074 bv->message(_("One match has been replaced."));
4077 bv->message(_("Two matches have been replaced."));
4080 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4086 bv->message(_("Match not found."));
4091 if (num_replaced > 0)
4092 bv->message(_("Match has been replaced."));
4094 bv->message(_("Match found."));
4096 if (cur.pos() + pos_len > cur.lastpos()) {
4097 // Prevent crash in bv->putSelectionAt()
4098 // Should never happen, maybe LASSERT() here?
4099 pos_len = cur.lastpos() - cur.pos();
4101 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4102 bv->putSelectionAt(cur, pos_len, !opt.forward);
4108 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4110 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4111 << opt.casesensitive << ' '
4112 << opt.matchword << ' '
4113 << opt.forward << ' '
4114 << opt.expandmacros << ' '
4115 << opt.ignoreformat << ' '
4116 << opt.replace_all << ' '
4117 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4118 << opt.keep_case << ' '
4119 << int(opt.scope) << ' '
4122 LYXERR(Debug::FIND, "built: " << os.str());
4128 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4130 // LYXERR(Debug::FIND, "parsing");
4134 while (line != "EOSS") {
4138 if (is.eof()) // Tolerate malformed request
4142 // LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4143 opt.find_buf_name = from_utf8(s);
4144 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4145 is.get(); // Waste space before replace string
4148 while (line != "EOSS") {
4152 if (is.eof()) // Tolerate malformed request
4156 // LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4157 opt.repl_buf_name = from_utf8(s);
4158 is >> opt.keep_case;
4161 opt.scope = FindAndReplaceOptions::SearchScope(i);
4163 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4166 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4167 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4168 << opt.scope << ' ' << opt.restr);