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"
35 #include "frontends/Application.h"
36 #include "frontends/alert.h"
38 #include "mathed/InsetMath.h"
39 #include "mathed/InsetMathHull.h"
40 #include "mathed/MathData.h"
41 #include "mathed/MathStream.h"
42 #include "mathed/MathSupport.h"
44 #include "support/debug.h"
45 #include "support/docstream.h"
46 #include "support/FileName.h"
47 #include "support/gettext.h"
48 #include "support/lassert.h"
49 #include "support/lstrings.h"
50 #include "support/textutils.h"
56 #define USE_QT_FOR_SEARCH
57 #if defined(USE_QT_FOR_SEARCH)
58 #include <QtCore> // sets QT_VERSION
59 #if (QT_VERSION >= 0x050000)
60 #include <QRegularExpression>
70 using namespace lyx::support;
74 typedef map<string, string> AccentsMap;
75 static AccentsMap accents = map<string, string>();
77 // Helper class for deciding what should be ignored
81 IgnoreFormats() = default;
83 bool getFamily() const { return ignoreFamily_; }
85 bool getSeries() const { return ignoreSeries_; }
87 bool getShape() const { return ignoreShape_; }
89 bool getUnderline() const { return ignoreUnderline_; }
91 bool getMarkUp() const { return ignoreMarkUp_; }
93 bool getStrikeOut() const { return ignoreStrikeOut_; }
95 bool getSectioning() const { return ignoreSectioning_; }
97 bool getFrontMatter() const { return ignoreFrontMatter_; }
99 bool getColor() const { return ignoreColor_; }
101 bool getLanguage() const { return ignoreLanguage_; }
103 void setIgnoreFormat(string const & type, bool value);
107 bool ignoreFamily_ = false;
109 bool ignoreSeries_ = false;
111 bool ignoreShape_ = false;
113 bool ignoreUnderline_ = false;
115 bool ignoreMarkUp_ = false;
117 bool ignoreStrikeOut_ = false;
119 bool ignoreSectioning_ = false;
121 bool ignoreFrontMatter_ = false;
123 bool ignoreColor_ = false;
125 bool ignoreLanguage_ = false;
129 void IgnoreFormats::setIgnoreFormat(string const & type, bool value)
131 if (type == "color") {
132 ignoreColor_ = value;
134 else if (type == "language") {
135 ignoreLanguage_ = value;
137 else if (type == "sectioning") {
138 ignoreSectioning_ = value;
139 ignoreFrontMatter_ = value;
141 else if (type == "font") {
142 ignoreSeries_ = value;
143 ignoreShape_ = value;
144 ignoreFamily_ = value;
146 else if (type == "series") {
147 ignoreSeries_ = value;
149 else if (type == "shape") {
150 ignoreShape_ = value;
152 else if (type == "family") {
153 ignoreFamily_ = value;
155 else if (type == "markup") {
156 ignoreMarkUp_ = value;
158 else if (type == "underline") {
159 ignoreUnderline_ = value;
161 else if (type == "strike") {
162 ignoreStrikeOut_ = value;
166 // The global variable that can be changed from outside
167 IgnoreFormats ignoreFormats;
170 void setIgnoreFormat(string const & type, bool value)
172 ignoreFormats.setIgnoreFormat(type, value);
178 bool parse_bool(docstring & howto)
183 howto = split(howto, var, ' ');
191 MatchString(docstring const & s, bool cs, bool mw)
192 : str(s), case_sens(cs), whole_words(mw)
195 // returns true if the specified string is at the specified position
196 // del specifies whether deleted strings in ct mode will be considered
197 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
199 return par.find(str, case_sens, whole_words, pos, del);
207 // match whole words only
212 int findForward(DocIterator & cur, MatchString const & match,
213 bool find_del = true)
215 for (; cur; cur.forwardChar())
216 if (cur.inTexted()) {
217 int len = match(cur.paragraph(), cur.pos(), find_del);
225 int findBackwards(DocIterator & cur, MatchString const & match,
226 bool find_del = true)
230 if (cur.inTexted()) {
231 int len = match(cur.paragraph(), cur.pos(), find_del);
240 bool searchAllowed(docstring const & str)
243 frontend::Alert::error(_("Search error"), _("Search string is empty"));
250 bool findOne(BufferView * bv, docstring const & searchstr,
251 bool case_sens, bool whole, bool forward,
252 bool find_del = true, bool check_wrap = false)
254 if (!searchAllowed(searchstr))
257 DocIterator cur = forward
258 ? bv->cursor().selectionEnd()
259 : bv->cursor().selectionBegin();
261 MatchString const match(searchstr, case_sens, whole);
263 int match_len = forward
264 ? findForward(cur, match, find_del)
265 : findBackwards(cur, match, find_del);
268 bv->putSelectionAt(cur, match_len, !forward);
269 else if (check_wrap) {
270 DocIterator cur_orig(bv->cursor());
273 q = _("End of file reached while searching forward.\n"
274 "Continue searching from the beginning?");
276 q = _("Beginning of file reached while searching backward.\n"
277 "Continue searching from the end?");
278 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
279 q, 0, 1, _("&Yes"), _("&No"));
280 if (wrap_answer == 0) {
282 bv->cursor().clear();
283 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
285 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
286 bv->cursor().backwardPos();
288 bv->clearSelection();
289 if (findOne(bv, searchstr, case_sens, whole, forward, find_del, false))
292 bv->cursor().setCursor(cur_orig);
296 return match_len > 0;
300 int replaceAll(BufferView * bv,
301 docstring const & searchstr, docstring const & replacestr,
302 bool case_sens, bool whole)
304 Buffer & buf = bv->buffer();
306 if (!searchAllowed(searchstr) || buf.isReadonly())
309 DocIterator cur_orig(bv->cursor());
311 MatchString const match(searchstr, case_sens, whole);
314 int const rsize = replacestr.size();
315 int const ssize = searchstr.size();
318 cur.setCursor(doc_iterator_begin(&buf));
319 int match_len = findForward(cur, match, false);
320 while (match_len > 0) {
321 // Backup current cursor position and font.
322 pos_type const pos = cur.pos();
323 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
325 int striked = ssize -
326 cur.paragraph().eraseChars(pos, pos + match_len,
327 buf.params().track_changes);
328 cur.paragraph().insert(pos, replacestr, font,
329 Change(buf.params().track_changes
331 : Change::UNCHANGED));
332 for (int i = 0; i < rsize + striked; ++i)
335 match_len = findForward(cur, match, false);
338 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
340 cur_orig.fixIfBroken();
341 bv->setCursor(cur_orig);
347 // the idea here is that we are going to replace the string that
348 // is selected IF it is the search string.
349 // if there is a selection, but it is not the search string, then
350 // we basically ignore it. (FIXME We ought to replace only within
352 // if there is no selection, then:
353 // (i) if some search string has been provided, then we find it.
354 // (think of how the dialog works when you hit "replace" the
356 // (ii) if no search string has been provided, then we treat the
357 // word the cursor is in as the search string. (why? i have no
358 // idea.) but this only works in text?
360 // returns the number of replacements made (one, if any) and
361 // whether anything at all was done.
362 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
363 docstring const & replacestr, bool case_sens,
364 bool whole, bool forward, bool findnext)
366 Cursor & cur = bv->cursor();
367 if (!cur.selection()) {
368 // no selection, non-empty search string: find it
369 if (!searchstr.empty()) {
370 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
371 return make_pair(found, 0);
373 // empty search string
376 return make_pair(false, 0);
377 // select current word and treat it as the search string.
378 // This causes a minor bug as undo will restore this selection,
379 // which the user did not create (#8986).
380 cur.innerText()->selectWord(cur, WHOLE_WORD);
381 searchstr = cur.selectionAsString(false, true);
384 // if we still don't have a search string, report the error
386 if (!searchAllowed(searchstr))
387 return make_pair(false, 0);
389 bool have_selection = cur.selection();
390 docstring const selected = cur.selectionAsString(false, true);
393 ? searchstr == selected
394 : compare_no_case(searchstr, selected) == 0;
396 // no selection or current selection is not search word:
397 // just find the search word
398 if (!have_selection || !match) {
399 bool const found = findOne(bv, searchstr, case_sens, whole, forward, true, findnext);
400 return make_pair(found, 0);
403 // we're now actually ready to replace. if the buffer is
404 // read-only, we can't, though.
405 if (bv->buffer().isReadonly())
406 return make_pair(false, 0);
408 cap::replaceSelectionWithString(cur, replacestr);
410 cur.pos() += replacestr.length();
411 LASSERT(cur.pos() <= cur.lastpos(),
412 cur.pos() = cur.lastpos());
415 findOne(bv, searchstr, case_sens, whole, forward, false, findnext);
417 return make_pair(true, 1);
423 docstring const find2string(docstring const & search,
424 bool casesensitive, bool matchword, bool forward)
428 << int(casesensitive) << ' '
429 << int(matchword) << ' '
435 docstring const replace2string(docstring const & replace,
436 docstring const & search,
437 bool casesensitive, bool matchword,
438 bool all, bool forward, bool findnext)
441 ss << replace << '\n'
443 << int(casesensitive) << ' '
444 << int(matchword) << ' '
446 << int(forward) << ' '
452 bool lyxfind(BufferView * bv, FuncRequest const & ev)
454 if (!bv || ev.action() != LFUN_WORD_FIND)
457 //lyxerr << "find called, cmd: " << ev << endl;
459 // data is of the form
461 // <casesensitive> <matchword> <forward>"
463 docstring howto = split(ev.argument(), search, '\n');
465 bool casesensitive = parse_bool(howto);
466 bool matchword = parse_bool(howto);
467 bool forward = parse_bool(howto);
469 return findOne(bv, search, casesensitive, matchword, forward, false, true);
473 bool lyxreplace(BufferView * bv, FuncRequest const & ev)
475 if (!bv || ev.action() != LFUN_WORD_REPLACE)
478 // data is of the form
481 // <casesensitive> <matchword> <all> <forward> <findnext>"
484 docstring howto = split(ev.argument(), rplc, '\n');
485 howto = split(howto, search, '\n');
487 bool casesensitive = parse_bool(howto);
488 bool matchword = parse_bool(howto);
489 bool all = parse_bool(howto);
490 bool forward = parse_bool(howto);
491 bool findnext = howto.empty() ? true : parse_bool(howto);
495 int replace_count = 0;
497 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
498 update = replace_count > 0;
501 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
503 replace_count = rv.second;
506 Buffer const & buf = bv->buffer();
508 // emit message signal.
509 buf.message(_("String not found."));
511 if (replace_count == 0) {
512 buf.message(_("String found."));
513 } else if (replace_count == 1) {
514 buf.message(_("String has been replaced."));
516 docstring const str =
517 bformat(_("%1$d strings have been replaced."), replace_count);
525 bool findNextChange(BufferView * bv, Cursor & cur, bool const check_wrap)
527 for (; cur; cur.forwardPos())
528 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
532 DocIterator cur_orig(bv->cursor());
533 docstring q = _("End of file reached while searching forward.\n"
534 "Continue searching from the beginning?");
535 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
536 q, 0, 1, _("&Yes"), _("&No"));
537 if (wrap_answer == 0) {
538 bv->cursor().clear();
539 bv->cursor().push_back(CursorSlice(bv->buffer().inset()));
540 bv->clearSelection();
541 cur.setCursor(bv->cursor().selectionBegin());
542 if (findNextChange(bv, cur, false))
545 bv->cursor().setCursor(cur_orig);
552 bool findPreviousChange(BufferView * bv, Cursor & cur, bool const check_wrap)
554 for (cur.backwardPos(); cur; cur.backwardPos()) {
555 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos()))
560 DocIterator cur_orig(bv->cursor());
561 docstring q = _("Beginning of file reached while searching backward.\n"
562 "Continue searching from the end?");
563 int wrap_answer = frontend::Alert::prompt(_("Wrap search?"),
564 q, 0, 1, _("&Yes"), _("&No"));
565 if (wrap_answer == 0) {
566 bv->cursor().setCursor(doc_iterator_end(&bv->buffer()));
567 bv->cursor().backwardPos();
568 bv->clearSelection();
569 cur.setCursor(bv->cursor().selectionBegin());
570 if (findPreviousChange(bv, cur, false))
573 bv->cursor().setCursor(cur_orig);
580 bool selectChange(Cursor & cur, bool forward)
582 if (!cur.inTexted() || !cur.paragraph().isChanged(cur.pos()))
584 Change ch = cur.paragraph().lookupChange(cur.pos());
586 CursorSlice tip1 = cur.top();
587 for (; tip1.pit() < tip1.lastpit() || tip1.pos() < tip1.lastpos(); tip1.forwardPos()) {
588 Change ch2 = tip1.paragraph().lookupChange(tip1.pos());
589 if (!ch2.isSimilarTo(ch))
592 CursorSlice tip2 = cur.top();
593 for (; tip2.pit() > 0 || tip2.pos() > 0;) {
595 Change ch2 = tip2.paragraph().lookupChange(tip2.pos());
596 if (!ch2.isSimilarTo(ch)) {
597 // take a step forward to correctly set the selection
605 cur.bv().mouseSetCursor(cur, false);
607 cur.bv().mouseSetCursor(cur, true);
615 bool findChange(BufferView * bv, bool forward)
618 cur.setCursor(forward ? bv->cursor().selectionEnd()
619 : bv->cursor().selectionBegin());
620 forward ? findNextChange(bv, cur, true) : findPreviousChange(bv, cur, true);
621 return selectChange(cur, forward);
626 bool findNextChange(BufferView * bv)
628 return findChange(bv, true);
632 bool findPreviousChange(BufferView * bv)
634 return findChange(bv, false);
641 typedef vector<pair<string, string> > Escapes;
643 string string2regex(string in)
645 static std::regex specialChars { R"([-[\]{}()*+?.,\^$|#\s\$\\])" };
646 string temp = std::regex_replace(in, specialChars, R"(\$&)" );
651 while (fl_pos < temp.size()) {
652 fl_pos = temp.find("\\\\foreignlanguage", lastpos + offset);
653 if (fl_pos == string::npos)
656 temp2 += temp.substr(lastpos, fl_pos - lastpos);
662 if (lastpos < temp.size()) {
663 temp2 += temp.substr(lastpos, temp.size() - lastpos);
668 string correctRegex(string t, bool withformat)
670 /* Convert \backslash => \
671 * and \{, \}, \[, \] => {, }, [, ]
674 regex wordre("(\\\\)*(\\\\((backslash|mathcircumflex) ?|[\\[\\]\\{\\}]))");
677 bool backslashed = false;
678 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
681 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
685 if (sub.str(4) == "backslash") {
688 // transforms '\backslash \{' into '\{'
690 string next = t.substr(sub.position(2) + sub.str(2).length(), 2);
691 if ((next == "\\{") || (next == "\\}")) {
697 else if (sub.str(4) == "mathcircumflex")
699 else if (backslashed) {
701 if (withformat && (sub.str(3) == "{"))
702 replace = accents["braceleft"];
703 else if (withformat && (sub.str(3) == "}"))
704 replace = accents["braceright"];
706 // else part should not exist
711 replace = sub.str(3);
713 if (lastpos < (size_t) sub.position(2))
714 s += t.substr(lastpos, sub.position(2) - lastpos);
716 lastpos = sub.position(2) + sub.length(2);
720 else if (lastpos < t.length())
721 s += t.substr(lastpos, t.length() - lastpos);
725 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
726 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
727 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
728 string escape_for_regex(string s, bool withformat)
732 while (lastpos < s.size()) {
733 size_t regex_pos = s.find("\\regexp{", lastpos);
734 if (regex_pos == string::npos) {
735 regex_pos = s.size();
737 if (regex_pos > lastpos) {
738 result += string2regex(s.substr(lastpos, regex_pos-lastpos));
740 if (lastpos == s.size())
743 size_t end_pos = s.find("\\endregexp{}}", regex_pos + 8);
744 result += correctRegex(s.substr(regex_pos + 8, end_pos -(regex_pos + 8)), withformat);
745 lastpos = end_pos + 13;
751 /// Wrapper for lyx::regex_replace with simpler interface
752 bool regex_replace(string const & s, string & t, string const & searchstr,
753 string const & replacestr)
755 regex e(searchstr, regex_constants::ECMAScript);
757 ostream_iterator<char, char> it(oss);
758 regex_replace(it, s.begin(), s.end(), e, replacestr);
759 // tolerate t and s be references to the same variable
760 bool rv = (s != oss.str());
779 vector <string> result = vector <string>();
780 MatchResult(int len = 0): match_len(len),match_prefix(0),match2end(0), pos(0),leadsize(0),pos_len(-1),searched_size(0) {};
783 static MatchResult::range interpretMatch(MatchResult &oldres, MatchResult &newres)
785 if (newres.match2end < oldres.match2end)
786 return MatchResult::newIsTooFar;
787 if (newres.match_len < oldres.match_len)
788 return MatchResult::newIsTooFar;
790 if (newres.match_len == oldres.match_len) {
791 if (newres.match2end == oldres.match2end)
792 return MatchResult::newIsBetter;
794 return MatchResult::newIsInvalid;
797 /** The class performing a match between a position in the document and the FindAdvOptions.
800 class MatchStringAdv {
802 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt);
804 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
805 ** constructor as opt.search, under the opt.* options settings.
808 ** If set, then match is searched only against beginning of text starting at cur.
809 ** If unset, then match is searched anywhere in text starting at cur.
812 ** The length of the matching text, or zero if no match was found.
814 MatchResult operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
823 /// first buffer on which search was started
824 lyx::Buffer * const p_first_buf;
826 FindAndReplaceOptions const & opt;
829 /// Auxiliary find method (does not account for opt.matchword)
830 MatchResult findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
831 void CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string = "");
833 /** Normalize a stringified or latexified LyX paragraph.
837 ** <li>if search is not casesensitive, then lowercase the string;
838 ** <li>remove any newline at begin or end of the string;
839 ** <li>replace any newline in the middle of the string with a simple space;
840 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
843 ** @todo Normalization should also expand macros, if the corresponding
844 ** search option was checked.
846 string normalize(docstring const & s) const;
847 // normalized string to search
848 string par_as_string;
849 // regular expression to use for searching
850 // regexp2 is same as regexp, but prefixed with a ".*?"
852 QRegularExpression regexp;
853 QRegularExpression regexp2;
858 // leading format material as string
859 string lead_as_string;
860 // par_as_string after removal of lead_as_string
861 string par_as_string_nolead;
862 // unmatched open braces in the search string/regexp
864 // number of (.*?) subexpressions added at end of search regexp for closing
865 // environments, math mode, styles, etc...
868 // Are we searching with regular expressions ?
870 static int valid_matches;
871 static vector <string> matches;
872 void FillResults(MatchResult &found_mr);
875 int MatchStringAdv::valid_matches = 0;
876 vector <string> MatchStringAdv::matches = vector <string>(10);
878 void MatchStringAdv::FillResults(MatchResult &found_mr)
880 if (found_mr.match_len > 0) {
881 valid_matches = found_mr.result.size();
882 for (size_t i = 0; i < found_mr.result.size(); i++)
883 matches[i] = found_mr.result[i];
889 static docstring buffer_to_latex(Buffer & buffer)
891 //OutputParams runparams(&buffer.params().encoding());
892 OutputParams runparams(encodings.fromLyXName("utf8"));
893 odocstringstream ods;
895 runparams.nice = true;
896 runparams.flavor = Flavor::XeTeX;
897 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
898 // No side effect of file copying and image conversion
899 runparams.dryrun = true;
900 runparams.for_search = true;
901 pit_type const endpit = buffer.paragraphs().size();
902 for (pit_type pit = 0; pit != endpit; ++pit) {
903 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
904 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
910 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt)
913 if (!opt.ignoreformat) {
914 str = buffer_to_latex(buffer);
916 // OutputParams runparams(&buffer.params().encoding());
917 OutputParams runparams(encodings.fromLyXName("utf8"));
918 runparams.nice = true;
919 runparams.flavor = Flavor::XeTeX;
920 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
921 runparams.dryrun = true;
922 runparams.for_search = true;
923 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
924 Paragraph const & par = buffer.paragraphs().at(pit);
925 LYXERR(Debug::FIND, "Adding to search string: '"
926 << par.asString(pos_type(0), par.size(),
927 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
930 str += par.asString(pos_type(0), par.size(),
931 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
934 // Even in ignore-format we have to remove "\text{}, \lyxmathsym{}" parts
935 string t = to_utf8(str);
936 while (regex_replace(t, t, "\\\\(text|lyxmathsym)\\{([^\\}]*)\\}", "$2"));
943 /// Return separation pos between the leading material and the rest
944 static size_t identifyLeading(string const & s)
947 // @TODO Support \item[text]
948 // Kornel: Added textsl, textsf, textit, texttt and noun
949 // + allow to search for colored text too
950 while (regex_replace(t, t, "^\\\\(("
951 "(author|title|subtitle|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|"
952 "lyxaddress|lyxrightaddress|"
953 "footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
954 "emph|noun|minisec|text(bf|md|sl|sf|it|tt))|"
955 "((textcolor|foreignlanguage|latexenvironment)\\{[a-z]+\\*?\\})|"
956 "(u|uu)line|(s|x)out|uwave)|((sub)?(((sub)?section)|paragraph)|part|chapter)\\*?)\\{", "")
957 || regex_replace(t, t, "^\\$", "")
958 || regex_replace(t, t, "^\\\\\\[", "")
959 || regex_replace(t, t, "^ ?\\\\item\\{[a-z]+\\}", "")
960 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\}", ""))
962 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
967 * Given a latexified string, retrieve some handled features
968 * The features of the regex will later be compared with the features
969 * of the searched text. If the regex features are not a
970 * subset of the analized, then, in not format ignoring search
971 * we can early stop the search in the relevant inset.
973 typedef map<string, bool> Features;
975 static Features identifyFeatures(string const & s)
977 static regex const feature("\\\\(([a-zA-Z]+(\\{([a-z]+\\*?)\\}|\\*)?))\\{");
978 static regex const valid("^("
980 "(footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge|"
981 "emph|noun|text(bf|md|sl|sf|it|tt)|"
982 "(textcolor|foreignlanguage|item|listitem|latexenvironment)\\{[a-z]+\\*?\\})|"
983 "(u|uu)line|(s|x)out|uwave|"
984 "(sub|extra)?title|author|subject|publishers|dedication|(upper|lower)titleback|lyx(right)?address)|"
985 "((sub)?(((sub)?section)|paragraph)|part|chapter|lyxslide)\\*?)$");
990 for (sregex_iterator it(s.begin(), s.end(), feature), end; it != end; ++it) {
993 if (sub.str(1).compare("regexp") == 0) {
997 string token = sub.str(1);
999 if (regex_match(token, sub2, valid)) {
1007 if (sub.str(1).compare("endregexp") == 0) {
1017 * defines values features of a key "\\[a-z]+{"
1022 /* Char type with content discarded
1023 * like \hspace{1cm} */
1025 /* Char, like \backslash */
1027 /* replace starting backslash with '#' */
1029 /* \part, \section*, ... */
1031 /* title, author etc */
1033 /* \foreignlanguage{ngerman}, ... */
1036 * to discard language in content */
1039 /* \begin{eqnarray}...\end{eqnarray}, ... $...$ */
1041 /* fonts, colors, markups, ... */
1043 /* footnotesize, ... large, ...
1044 * Ignore all of them */
1047 /* inputencoding, ...
1048 * Discard also content, because they do not help in search */
1051 * like remove, but also all arguments */
1053 /* item, listitem */
1055 /* tex, latex, ... like isChar */
1057 /* like \lettrine[lines=5]{}{} */
1059 // like isStandard, but always remove head
1061 /* End of arguments marker for lettrine,
1062 * so that they can be ignored */
1065 KeyInfo() = default;
1066 KeyInfo(KeyType type, int parcount, bool disable)
1068 parenthesiscount(parcount),
1069 disabled(disable) {}
1070 KeyType keytype = invalid;
1072 int _tokensize = -1;
1073 int _tokenstart = -1;
1074 int _dataStart = -1;
1076 int parenthesiscount = 1;
1077 bool disabled = false;
1078 bool used = false; /* by pattern */
1083 Border(int l=0, int u=0) : low(l), upper(u) {};
1088 #define MAXOPENED 30
1090 bool isPatternString_;
1092 explicit Intervall(bool isPattern, string const & p) :
1093 isPatternString_(isPattern), par(p), ignoreidx(-1), actualdeptindex(0),
1094 hasTitle(false), langcount(0)
1102 static vector<Border> borders;
1103 int depts[MAXOPENED];
1104 int closes[MAXOPENED];
1105 int actualdeptindex;
1106 int previousNotIgnored(int) const;
1107 int nextNotIgnored(int) const;
1108 void handleOpenP(int i);
1109 void handleCloseP(int i, bool closingAllowed);
1110 void resetOpenedP(int openPos);
1111 void addIntervall(int upper);
1112 void addIntervall(int low, int upper); /* if explicit */
1113 void removeAccents();
1114 void setForDefaultLang(KeyInfo const & defLang) const;
1115 int findclosing(int start, int end, char up, char down, int repeat);
1116 void handleParentheses(int lastpos, bool closingAllowed);
1118 int langcount; // Number of disabled language specs up to current position in actual interval
1119 int isOpeningPar(int pos) const;
1121 void output(ostringstream &os, int lastpos);
1122 // string show(int lastpos);
1125 vector<Border> Intervall::borders = vector<Border>(30);
1127 int Intervall::isOpeningPar(int pos) const
1129 if ((pos < 0) || (size_t(pos) >= par.size()))
1131 if (par[pos] != '{')
1133 if (size_t(pos) + 2 >= par.size())
1135 if (par[pos+2] != '}')
1137 if (par[pos+1] == '[' || par[pos+1] == ']')
1142 void Intervall::setForDefaultLang(KeyInfo const & defLang) const
1144 // Enable the use of first token again
1145 if (ignoreidx >= 0) {
1146 int value = defLang._tokenstart + defLang._tokensize;
1152 if (borders[borderidx].low < value)
1153 borders[borderidx].low = value;
1154 if (borders[borderidx].upper < value)
1155 borders[borderidx].upper = value;
1160 static void checkDepthIndex(int val)
1162 static int maxdepthidx = MAXOPENED-2;
1163 static int lastmaxdepth = 0;
1164 if (val > lastmaxdepth) {
1165 LYXERR(Debug::INFO, "Depth reached " << val);
1168 if (val > maxdepthidx) {
1170 LYXERR(Debug::INFO, "maxdepthidx now " << val);
1175 // Not needed, because borders are now dynamically expanded
1176 static void checkIgnoreIdx(int val)
1178 static int lastmaxignore = -1;
1179 if ((lastmaxignore < val) && (size_t(val+1) >= borders.size())) {
1180 LYXERR(Debug::INFO, "IgnoreIdx reached " << val);
1181 lastmaxignore = val;
1187 * Expand the region of ignored parts of the input latex string
1188 * The region is only relevant in output()
1190 void Intervall::addIntervall(int low, int upper)
1193 if (low == upper) return;
1194 for (idx = ignoreidx+1; idx > 0; --idx) {
1195 if (low > borders[idx-1].upper) {
1199 Border br(low, upper);
1200 if (idx > ignoreidx) {
1201 if (borders.size() <= size_t(idx)) {
1202 borders.push_back(br);
1208 // checkIgnoreIdx(ignoreidx);
1212 // Expand only if one of the new bound is inside the interwall
1213 // We know here that br.low > borders[idx-1].upper
1214 if (br.upper < borders[idx].low) {
1215 // We have to insert at this pos
1216 if (size_t(ignoreidx+1) >= borders.size()) {
1217 borders.push_back(borders[ignoreidx]);
1220 borders[ignoreidx+1] = borders[ignoreidx];
1222 for (int i = ignoreidx; i > idx; --i) {
1223 borders[i] = borders[i-1];
1227 // checkIgnoreIdx(ignoreidx);
1230 // Here we know, that we are overlapping
1231 if (br.low > borders[idx].low)
1232 br.low = borders[idx].low;
1233 // check what has to be concatenated
1235 for (int i = idx; i <= ignoreidx; i++) {
1236 if (br.upper >= borders[i].low) {
1238 if (br.upper < borders[i].upper)
1239 br.upper = borders[i].upper;
1245 // count should be >= 1 here
1248 for (int i = idx + count; i <= ignoreidx; i++) {
1249 borders[i-count+1] = borders[i];
1251 ignoreidx -= count - 1;
1257 static void buildaccent(string n, string param, string values)
1261 const char delim = '|';
1262 while (getline(s, name, delim)) {
1264 for (char c : param) {
1265 string key = name + "{" + c + "}";
1266 // get the corresponding utf8-value
1267 if ((values[start] & 0xc0) != 0xc0) {
1268 // should not happen, utf8 encoding starts at least with 11xxxxxx
1269 // but value for '\dot{i}' is 'i', which is ascii
1270 if ((values[start] & 0x80) == 0) {
1272 accents[key] = values.substr(start, 1);
1273 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1278 for (int j = 1; ;j++) {
1279 if (start + j >= values.size()) {
1280 accents[key] = values.substr(start, j);
1281 start = values.size() - 1;
1284 else if ((values[start+j] & 0xc0) != 0x80) {
1285 // This is the first byte of following utf8 char
1286 accents[key] = values.substr(start, j);
1288 // LYXERR(Debug::INFO, "" << key << "=" << accents[key]);
1297 static string getutf8(unsigned uchar)
1300 string ret = string();
1302 if (uchar <= 0x7f) {
1303 c[maxc-1] = uchar & 0x7f;
1306 unsigned char rest = 0x40;
1307 unsigned char first = 0x80;
1309 for (int i = start; i >=0; --i) {
1311 c[i] = first + uchar;
1314 c[i] = 0x80 | (uchar & 0x3f);
1321 for (int i = 0; i < maxc; i++) {
1322 if (c[i] == 0) continue;
1328 static void buildAccentsMap()
1330 accents["imath"] = "ı";
1332 accents["jmath"] = "ȷ";
1333 accents["cdot"] = "·";
1334 accents["textasciicircum"] = "^";
1335 accents["mathcircumflex"] = "^";
1336 accents["sim"] = "~";
1337 accents["guillemotright"] = "»";
1338 accents["guillemotleft"] = "«";
1339 accents["hairspace"] = getutf8(0xf0000); // select from free unicode plane 15
1340 accents["thinspace"] = getutf8(0xf0002); // and used _only_ by findadv
1341 accents["negthinspace"] = getutf8(0xf0003); // to omit backslashed latex macros
1342 accents["medspace"] = getutf8(0xf0004); // See https://en.wikipedia.org/wiki/Private_Use_Areas
1343 accents["negmedspace"] = getutf8(0xf0005);
1344 accents["thickspace"] = getutf8(0xf0006);
1345 accents["negthickspace"] = getutf8(0xf0007);
1346 accents["lyx"] = getutf8(0xf0010); // Used logos
1347 accents["LyX"] = getutf8(0xf0010);
1348 accents["tex"] = getutf8(0xf0011);
1349 accents["TeX"] = getutf8(0xf0011);
1350 accents["latex"] = getutf8(0xf0012);
1351 accents["LaTeX"] = getutf8(0xf0012);
1352 accents["latexe"] = getutf8(0xf0013);
1353 accents["LaTeXe"] = getutf8(0xf0013);
1354 accents["lyxarrow"] = getutf8(0xf0020);
1355 accents["braceleft"] = getutf8(0xf0030);
1356 accents["braceright"] = getutf8(0xf0031);
1357 accents["backslash lyx"] = getutf8(0xf0010); // Used logos inserted with starting \backslash
1358 accents["backslash LyX"] = getutf8(0xf0010);
1359 accents["backslash tex"] = getutf8(0xf0011);
1360 accents["backslash TeX"] = getutf8(0xf0011);
1361 accents["backslash latex"] = getutf8(0xf0012);
1362 accents["backslash LaTeX"] = getutf8(0xf0012);
1363 accents["backslash latexe"] = getutf8(0xf0013);
1364 accents["backslash LaTeXe"] = getutf8(0xf0013);
1365 accents["backslash lyxarrow"] = getutf8(0xf0020);
1366 accents["ddot{\\imath}"] = "ï";
1367 buildaccent("ddot", "aAeEhHiIioOtuUwWxXyY",
1368 "äÄëËḧḦïÏïöÖẗüÜẅẄẍẌÿŸ"); // umlaut
1369 buildaccent("dot|.", "aAbBcCdDeEfFGghHIimMnNoOpPrRsStTwWxXyYzZ",
1370 "ȧȦḃḂċĊḋḊėĖḟḞĠġḣḢİİṁṀṅṄȯȮṗṖṙṘṡṠṫṪẇẆẋẊẏẎżŻ"); // dot{i} can only happen if ignoring case, but there is no lowercase of 'İ'
1371 accents["acute{\\imath}"] = "í";
1372 buildaccent("acute", "aAcCeEgGkKlLmMoOnNpPrRsSuUwWyYzZiI",
1373 "áÁćĆéÉǵǴḱḰĺĹḿḾóÓńŃṕṔŕŔśŚúÚẃẂýÝźŹíÍ");
1374 buildaccent("dacute|H|h", "oOuU", "őŐűŰ"); // double acute
1375 buildaccent("mathring|r", "aAuUwy",
1377 accents["check{\\imath}"] = "ǐ";
1378 accents["check{\\jmath}"] = "ǰ";
1379 buildaccent("check|v", "cCdDaAeEiIoOuUgGkKhHlLnNrRsSTtzZ",
1380 "čČďĎǎǍěĚǐǏǒǑǔǓǧǦǩǨȟȞľĽňŇřŘšŠŤťžŽ"); // caron
1381 accents["hat{\\imath}"] = "î";
1382 accents["hat{\\jmath}"] = "ĵ";
1383 buildaccent("hat|^", "aAcCeEgGhHiIjJoOsSuUwWyYzZ",
1384 "âÂĉĈêÊĝĜĥĤîÎĵĴôÔŝŜûÛŵŴŷŶẑẐ"); // circ
1385 accents["bar{\\imath}"] = "ī";
1386 buildaccent("bar|=", "aAeEiIoOuUyY",
1387 "āĀēĒīĪōŌūŪȳȲ"); // macron
1388 accents["tilde{\\imath}"] = "ĩ";
1389 buildaccent("tilde", "aAeEiInNoOuUvVyY",
1390 "ãÃẽẼĩĨñÑõÕũŨṽṼỹỸ"); // tilde
1391 accents["breve{\\imath}"] = "ĭ";
1392 buildaccent("breve|u", "aAeEgGiIoOuU",
1393 "ăĂĕĔğĞĭĬŏŎŭŬ"); // breve
1394 accents["grave{\\imath}"] = "ì";
1395 buildaccent("grave|`", "aAeEiIoOuUnNwWyY",
1396 "àÀèÈìÌòÒùÙǹǸẁẀỳỲ"); // grave
1397 buildaccent("subdot|d", "BbDdHhKkLlMmNnRrSsTtVvWwZzAaEeIiOoUuYy",
1398 "ḄḅḌḍḤḥḲḳḶḷṂṃṆṇṚṛṢṣṬṭṾṿẈẉẒẓẠạẸẹỊịỌọỤụỴỵ"); // dot below
1399 buildaccent("ogonek|k", "AaEeIiUuOo",
1400 "ĄąĘęĮįŲųǪǫ"); // ogonek
1401 buildaccent("cedilla|c", "CcGgKkLlNnRrSsTtEeDdHh",
1402 "ÇçĢģĶķĻļŅņŖŗŞşŢţȨȩḐḑḨḩ"); // cedilla
1403 buildaccent("subring|textsubring", "Aa",
1405 buildaccent("subhat|textsubcircum", "DdEeLlNnTtUu",
1406 "ḒḓḘḙḼḽṊṋṰṱṶṷ"); // subcircum
1407 buildaccent("subtilde|textsubtilde", "EeIiUu",
1408 "ḚḛḬḭṴṵ"); // subtilde
1409 accents["dgrave{\\imath}"] = "ȉ";
1410 accents["textdoublegrave{\\i}"] = "ȉ";
1411 buildaccent("dgrave|textdoublegrave", "AaEeIiOoRrUu",
1412 "ȀȁȄȅȈȉȌȍȐȑȔȕ"); // double grave
1413 accents["rcap{\\imath}"] = "ȉ";
1414 accents["textroundcap{\\i}"] = "ȉ";
1415 buildaccent("rcap|textroundcap", "AaEeIiOoRrUu",
1416 "ȂȃȆȇȊȋȎȏȒȓȖȗ"); // inverted breve
1417 buildaccent("slashed", "oO",
1422 * Created accents in math or regexp environment
1423 * are macros, but we need the utf8 equivalent
1425 void Intervall::removeAccents()
1427 if (accents.empty())
1429 static regex const accre("\\\\(([\\S]|grave|breve|ddot|dot|acute|dacute|mathring|check|hat|bar|tilde|subdot|ogonek|"
1430 "cedilla|subring|textsubring|subhat|textsubcircum|subtilde|textsubtilde|dgrave|textdoublegrave|rcap|textroundcap|slashed)\\{[^\\{\\}]+\\}"
1431 "|((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]))");
1433 for (sregex_iterator itacc(par.begin(), par.end(), accre), end; itacc != end; ++itacc) {
1435 string key = sub.str(1);
1436 if (accents.find(key) != accents.end()) {
1437 string val = accents[key];
1438 size_t pos = sub.position(size_t(0));
1439 for (size_t i = 0; i < val.size(); i++) {
1440 par[pos+i] = val[i];
1442 // Remove possibly following space too
1443 if (par[pos+sub.str(0).size()] == ' ')
1444 addIntervall(pos+val.size(), pos + sub.str(0).size()+1);
1446 addIntervall(pos+val.size(), pos + sub.str(0).size());
1447 for (size_t i = pos+val.size(); i < pos + sub.str(0).size(); i++) {
1448 // remove traces of any remaining chars
1453 LYXERR(Debug::INFO, "Not added accent for \"" << key << "\"");
1458 void Intervall::handleOpenP(int i)
1461 depts[actualdeptindex] = i+1;
1462 closes[actualdeptindex] = -1;
1463 checkDepthIndex(actualdeptindex);
1466 void Intervall::handleCloseP(int i, bool closingAllowed)
1468 if (actualdeptindex <= 0) {
1469 if (! closingAllowed)
1470 LYXERR(Debug::FIND, "Bad closing parenthesis in latex"); /* should not happen, but the latex input may be wrong */
1471 // if we are at the very end
1472 addIntervall(i, i+1);
1475 closes[actualdeptindex] = i+1;
1480 void Intervall::resetOpenedP(int openPos)
1482 // Used as initializer for foreignlanguage entry
1483 actualdeptindex = 1;
1484 depts[1] = openPos+1;
1488 int Intervall::previousNotIgnored(int start) const
1490 int idx = 0; /* int intervalls */
1491 for (idx = ignoreidx; idx >= 0; --idx) {
1492 if (start > borders[idx].upper)
1494 if (start >= borders[idx].low)
1495 start = borders[idx].low-1;
1500 int Intervall::nextNotIgnored(int start) const
1502 int idx = 0; /* int intervalls */
1503 for (idx = 0; idx <= ignoreidx; idx++) {
1504 if (start < borders[idx].low)
1506 if (start < borders[idx].upper)
1507 start = borders[idx].upper;
1512 typedef map<string, KeyInfo> KeysMap;
1513 typedef vector< KeyInfo> Entries;
1514 static KeysMap keys = map<string, KeyInfo>();
1520 Intervall interval_;
1521 void buildKeys(bool);
1522 void buildEntries(bool);
1523 void makeKey(const string &, KeyInfo, bool isPatternString);
1524 void processRegion(int start, int region_end); /* remove {} parts */
1525 void removeHead(KeyInfo const &, int count=0);
1528 LatexInfo(string const & par, bool isPatternString)
1529 : entidx_(-1), interval_(isPatternString, par)
1531 buildKeys(isPatternString);
1532 entries_ = vector<KeyInfo>();
1533 buildEntries(isPatternString);
1537 if (entries_.empty()) {
1540 if (entries_[0].keytype == KeyInfo::isTitle) {
1541 interval_.hasTitle = true;
1542 if (! entries_[0].disabled) {
1543 interval_.titleValue = entries_[0].head;
1546 interval_.titleValue = "";
1548 removeHead(entries_[0]);
1549 if (entries_.size() > 1)
1558 if (int(entries_.size()) > entidx_) {
1565 bool setNextKey(int idx) {
1566 if ((idx == entidx_) && (entidx_ >= 0)) {
1573 int find(int start, KeyInfo::KeyType keytype) const {
1577 while (tmpIdx < int(entries_.size())) {
1578 if (entries_[tmpIdx].keytype == keytype)
1584 int process(ostringstream & os, KeyInfo const & actual);
1585 int dispatch(ostringstream & os, int previousStart, KeyInfo & actual);
1586 // string show(int lastpos) { return interval.show(lastpos);};
1587 int nextNotIgnored(int start) { return interval_.nextNotIgnored(start);};
1588 KeyInfo &getKeyInfo(int keyinfo) {
1589 static KeyInfo invalidInfo = KeyInfo();
1590 if ((keyinfo < 0) || ( keyinfo >= int(entries_.size())))
1593 return entries_[keyinfo];
1595 void setForDefaultLang(KeyInfo const & defLang) {interval_.setForDefaultLang(defLang);};
1596 void addIntervall(int low, int up) { interval_.addIntervall(low, up); };
1600 int Intervall::findclosing(int start, int end, char up = '{', char down = '}', int repeat = 1)
1604 for (int i = start; i < end; i += 1 + skip) {
1608 if (c == '\\') skip = 1;
1612 else if (c == down) {
1615 if ((repeat <= 0) || (par[i+1] != up))
1633 vector<MathEntry> entries_;
1638 void insert(string const & wait, size_t start, size_t end) {
1639 MathEntry m = MathEntry();
1641 m.mathStart = start;
1643 m.mathSize = end - start;
1644 entries_.push_back(m);
1646 bool empty() const { return entries_.empty(); };
1647 size_t getEndPos() const {
1648 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1651 return entries_[actualIdx_].mathEnd;
1653 size_t getStartPos() const {
1654 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1655 return 100000; /* definitely enough? */
1657 return entries_[actualIdx_].mathStart;
1659 size_t getFirstPos() {
1661 return getStartPos();
1663 size_t getSize() const {
1664 if (entries_.empty() || (actualIdx_ >= entries_.size())) {
1667 return entries_[actualIdx_].mathSize;
1669 void incrEntry() { actualIdx_++; };
1672 void LatexInfo::buildEntries(bool isPatternString)
1674 static regex const rmath("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\(begin|end)\\{((eqnarray|equation|flalign|gather|multline|align|alignat)\\*?)\\})");
1675 static regex const rkeys("(\\\\)*(\\$|\\\\\\[|\\\\\\]|\\\\((([a-zA-Z]+\\*?)(\\{([a-z]+\\*?)\\}|=[0-9]+[a-z]+)?)))");
1676 static bool disableLanguageOverride = false;
1677 smatch sub, submath;
1678 bool evaluatingRegexp = false;
1680 bool evaluatingMath = false;
1681 bool evaluatingCode = false;
1683 bool evaluatingOptional = false;
1684 size_t optionalEnd = 0;
1687 bool math_end_waiting = false;
1688 size_t math_pos = 10000;
1690 static vector<string> usedText = vector<string>();
1692 interval_.removeAccents();
1694 for (sregex_iterator itmath(interval_.par.begin(), interval_.par.end(), rmath), end; itmath != end; ++itmath) {
1696 if ((submath.position(2) - submath.position(0)) %2 == 1) {
1697 // prefixed by odd count of '\\'
1700 if (math_end_waiting) {
1701 size_t pos = submath.position(size_t(2));
1702 if ((math_end == "$") &&
1703 (submath.str(2) == "$")) {
1704 mi.insert("$", math_pos, pos + 1);
1705 math_end_waiting = false;
1707 else if ((math_end == "\\]") &&
1708 (submath.str(2) == "\\]")) {
1709 mi.insert("\\]", math_pos, pos + 2);
1710 math_end_waiting = false;
1712 else if ((submath.str(3).compare("end") == 0) &&
1713 (submath.str(4).compare(math_end) == 0)) {
1714 mi.insert(math_end, math_pos, pos + submath.str(2).length());
1715 math_end_waiting = false;
1721 if (submath.str(3).compare("begin") == 0) {
1722 math_end_waiting = true;
1723 math_end = submath.str(4);
1724 math_pos = submath.position(size_t(2));
1726 else if (submath.str(2).compare("\\[") == 0) {
1727 math_end_waiting = true;
1729 math_pos = submath.position(size_t(2));
1731 else if (submath.str(2) == "$") {
1732 size_t pos = submath.position(size_t(2));
1733 math_end_waiting = true;
1739 // Ignore language if there is math somewhere in pattern-string
1740 if (isPatternString) {
1741 for (auto s: usedText) {
1742 // Remove entries created in previous search runs
1745 usedText = vector<string>();
1748 keys["foreignlanguage"].disabled = true;
1749 disableLanguageOverride = true;
1752 disableLanguageOverride = false;
1755 if (disableLanguageOverride) {
1756 keys["foreignlanguage"].disabled = true;
1759 math_pos = mi.getFirstPos();
1760 for (sregex_iterator it(interval_.par.begin(), interval_.par.end(), rkeys), end; it != end; ++it) {
1762 if ((sub.position(2) - sub.position(0)) %2 == 1) {
1763 // prefixed by odd count of '\\'
1766 string key = sub.str(5);
1768 if (sub.str(2)[0] == '\\')
1769 key = sub.str(2)[1];
1774 if (keys.find(key) != keys.end()) {
1775 if (keys[key].keytype == KeyInfo::headRemove) {
1776 KeyInfo found1 = keys[key];
1777 found1.disabled = true;
1778 found1.head = "\\" + key + "{";
1779 found1._tokenstart = sub.position(size_t(2));
1780 found1._tokensize = found1.head.length();
1781 found1._dataStart = found1._tokenstart + found1.head.length();
1782 int endpos = interval_.findclosing(found1._dataStart, interval_.par.length(), '{', '}', 1);
1783 found1._dataEnd = endpos;
1788 if (evaluatingRegexp) {
1789 if (sub.str(3).compare("endregexp") == 0) {
1790 evaluatingRegexp = false;
1791 // found._tokenstart already set
1792 found._dataEnd = sub.position(size_t(2)) + 13;
1793 found._dataStart = found._dataEnd;
1794 found._tokensize = found._dataEnd - found._tokenstart;
1795 found.parenthesiscount = 0;
1796 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1803 if (evaluatingMath) {
1804 if (size_t(sub.position(size_t(2))) < mi.getEndPos())
1806 evaluatingMath = false;
1808 math_pos = mi.getStartPos();
1810 if (keys.find(key) == keys.end()) {
1811 found = KeyInfo(KeyInfo::isStandard, 0, true);
1812 LYXERR(Debug::INFO, "Undefined key " << key << " ==> will be used as text");
1813 found = KeyInfo(KeyInfo::isText, 0, false);
1814 if (isPatternString) {
1815 found.keytype = KeyInfo::isChar;
1816 found.disabled = false;
1820 usedText.push_back(key);
1824 if (key.compare("regexp") == 0) {
1825 evaluatingRegexp = true;
1826 found._tokenstart = sub.position(size_t(2));
1827 found._tokensize = 0;
1831 // Handle the other params of key
1832 if (found.keytype == KeyInfo::isIgnored)
1834 else if (found.keytype == KeyInfo::isMath) {
1835 if (size_t(sub.position(size_t(2))) == math_pos) {
1837 found._tokenstart = sub.position(size_t(2));
1838 found._tokensize = mi.getSize();
1839 found._dataEnd = found._tokenstart + found._tokensize;
1840 found._dataStart = found._dataEnd;
1841 found.parenthesiscount = 0;
1842 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1843 evaluatingMath = true;
1846 // begin|end of unknown env, discard
1847 // First handle tables
1848 // longtable|tabular
1849 bool discardComment;
1851 found.keytype = KeyInfo::doRemove;
1852 if ((sub.str(7).compare("longtable") == 0) ||
1853 (sub.str(7).compare("tabular") == 0)) {
1854 discardComment = true; /* '%' */
1857 discardComment = false;
1858 static regex const removeArgs("^(multicols|multipar|sectionbox|subsectionbox|tcolorbox)$");
1860 string token = sub.str(7);
1861 if (regex_match(token, sub2, removeArgs)) {
1862 found.keytype = KeyInfo::removeWithArg;
1865 // discard spaces before pos(2)
1866 int pos = sub.position(size_t(2));
1868 for (count = 0; pos - count > 0; count++) {
1869 char c = interval_.par[pos-count-1];
1870 if (discardComment) {
1871 if ((c != ' ') && (c != '%'))
1877 found._tokenstart = pos - count;
1878 if (sub.str(3).compare(0, 5, "begin") == 0) {
1879 size_t pos1 = pos + sub.str(2).length();
1880 if (sub.str(7).compare("cjk") == 0) {
1881 pos1 = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1882 if ((interval_.par[pos1] == '{') && (interval_.par[pos1+1] == '}'))
1884 found.keytype = KeyInfo::isMain;
1885 found._dataStart = pos1;
1886 found._dataEnd = interval_.par.length();
1887 found.disabled = keys["foreignlanguage"].disabled;
1888 found.used = keys["foreignlanguage"].used;
1889 found._tokensize = pos1 - found._tokenstart;
1890 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1893 // Swallow possible optional params
1894 while (interval_.par[pos1] == '[') {
1895 pos1 = interval_.findclosing(pos1+1, interval_.par.length(), '[', ']')+1;
1897 // Swallow also the eventual parameter
1898 if (interval_.par[pos1] == '{') {
1899 found._dataEnd = interval_.findclosing(pos1+1, interval_.par.length()) + 1;
1902 found._dataEnd = pos1;
1904 found._dataStart = found._dataEnd;
1905 found._tokensize = count + found._dataEnd - pos;
1906 found.parenthesiscount = 0;
1907 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1908 found.disabled = true;
1912 // Handle "\end{...}"
1913 found._dataStart = pos + sub.str(2).length();
1914 found._dataEnd = found._dataStart;
1915 found._tokensize = count + found._dataEnd - pos;
1916 found.parenthesiscount = 0;
1917 found.head = interval_.par.substr(found._tokenstart, found._tokensize);
1918 found.disabled = true;
1922 else if (found.keytype != KeyInfo::isRegex) {
1923 found._tokenstart = sub.position(size_t(2));
1924 if (found.parenthesiscount == 0) {
1925 // Probably to be discarded
1926 size_t following_pos = sub.position(size_t(2)) + sub.str(5).length() + 1;
1927 char following = interval_.par[following_pos];
1928 if (following == ' ')
1929 found.head = "\\" + sub.str(5) + " ";
1930 else if (following == '=') {
1931 // like \uldepth=1000pt
1932 found.head = sub.str(2);
1935 found.head = "\\" + key;
1936 found._tokensize = found.head.length();
1937 found._dataEnd = found._tokenstart + found._tokensize;
1938 found._dataStart = found._dataEnd;
1941 int params = found._tokenstart + key.length() + 1;
1942 if (evaluatingOptional) {
1943 if (size_t(found._tokenstart) > optionalEnd) {
1944 evaluatingOptional = false;
1947 found.disabled = true;
1950 int optend = params;
1951 while (interval_.par[optend] == '[') {
1952 // discard optional parameters
1953 optend = interval_.findclosing(optend+1, interval_.par.length(), '[', ']') + 1;
1955 if (optend > params) {
1956 key += interval_.par.substr(params, optend-params);
1957 evaluatingOptional = true;
1958 optionalEnd = optend;
1959 if (found.keytype == KeyInfo::isSectioning) {
1960 // Remove optional values (but still keep in header)
1961 interval_.addIntervall(params, optend);
1964 string token = sub.str(7);
1966 if (interval_.par[optend] != '{') {
1968 found.parenthesiscount = 0;
1969 found.head = "\\" + key;
1972 closings = found.parenthesiscount;
1973 if (found.parenthesiscount == 1) {
1974 found.head = "\\" + key + "{";
1976 else if (found.parenthesiscount > 1) {
1978 found.head = sub.str(2) + "{";
1979 closings = found.parenthesiscount - 1;
1982 found.head = "\\" + key + "{";
1985 found._tokensize = found.head.length();
1986 found._dataStart = found._tokenstart + found.head.length();
1987 if (found.keytype == KeyInfo::doRemove) {
1989 size_t endpar = 2 + interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
1990 if (endpar >= interval_.par.length())
1991 found._dataStart = interval_.par.length();
1993 found._dataStart = endpar;
1994 found._tokensize = found._dataStart - found._tokenstart;
1997 found._dataStart = found._tokenstart + found._tokensize;
2001 if (interval_.par.substr(found._dataStart, 15).compare("\\endarguments{}") == 0) {
2002 found._dataStart += 15;
2006 endpos = found._dataStart - 1;
2008 endpos = interval_.findclosing(found._dataStart, interval_.par.length(), '{', '}', closings);
2009 if (found.keytype == KeyInfo::isList) {
2010 // Check if it really is list env
2011 static regex const listre("^([a-z]+)$");
2013 if (!regex_match(token, sub2, listre)) {
2014 // Change the key of this entry. It is not in a list/item environment
2015 found.keytype = KeyInfo::endArguments;
2018 if (found.keytype == KeyInfo::noMain) {
2019 evaluatingCode = true;
2021 codeStart = found._dataStart;
2023 else if (evaluatingCode) {
2024 if (size_t(found._dataStart) > codeEnd)
2025 evaluatingCode = false;
2026 else if (found.keytype == KeyInfo::isMain) {
2027 // Disable this key, treate it as standard
2028 found.keytype = KeyInfo::isStandard;
2029 found.disabled = true;
2030 if ((codeEnd +1 >= interval_.par.length()) &&
2031 (found._tokenstart == codeStart)) {
2032 // trickery, because the code inset starts
2033 // with \selectlanguage ...
2035 if (entries_.size() > 1) {
2036 entries_[entries_.size()-1]._dataEnd = codeEnd;
2041 if ((endpos == interval_.par.length()) &&
2042 (found.keytype == KeyInfo::doRemove)) {
2043 // Missing closing => error in latex-input?
2044 // therefore do not delete remaining data
2045 found._dataStart -= 1;
2046 found._dataEnd = found._dataStart;
2049 found._dataEnd = endpos;
2051 if (isPatternString) {
2052 keys[key].used = true;
2055 entries_.push_back(found);
2059 void LatexInfo::makeKey(const string &keysstring, KeyInfo keyI, bool isPatternString)
2061 stringstream s(keysstring);
2063 const char delim = '|';
2064 while (getline(s, key, delim)) {
2065 KeyInfo keyII(keyI);
2066 if (isPatternString) {
2069 else if ( !keys[key].used)
2070 keyII.disabled = true;
2075 void LatexInfo::buildKeys(bool isPatternString)
2078 static bool keysBuilt = false;
2079 if (keysBuilt && !isPatternString) return;
2081 // Keys to ignore in any case
2082 makeKey("text|textcyrillic|lyxmathsym", KeyInfo(KeyInfo::headRemove, 1, true), true);
2083 // Known standard keys with 1 parameter.
2084 // Split is done, if not at start of region
2085 makeKey("textsf|textss|texttt", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getFamily()), isPatternString);
2086 makeKey("textbf", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getSeries()), isPatternString);
2087 makeKey("textit|textsc|textsl", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getShape()), isPatternString);
2088 makeKey("uuline|uline|uwave", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getUnderline()), isPatternString);
2089 makeKey("emph|noun", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getMarkUp()), isPatternString);
2090 makeKey("sout|xout", KeyInfo(KeyInfo::isStandard, 1, ignoreFormats.getStrikeOut()), isPatternString);
2092 makeKey("section|subsection|subsubsection|paragraph|subparagraph|minisec",
2093 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2094 makeKey("section*|subsection*|subsubsection*|paragraph*",
2095 KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2096 makeKey("part|part*|chapter|chapter*", KeyInfo(KeyInfo::isSectioning, 1, ignoreFormats.getSectioning()), isPatternString);
2097 makeKey("title|subtitle|author|subject|publishers|dedication|uppertitleback|lowertitleback|extratitle|lyxaddress|lyxrightaddress", KeyInfo(KeyInfo::isTitle, 1, ignoreFormats.getFrontMatter()), isPatternString);
2099 makeKey("regexp", KeyInfo(KeyInfo::isRegex, 1, false), isPatternString);
2101 // Split is done, if not at start of region
2102 makeKey("textcolor", KeyInfo(KeyInfo::isStandard, 2, ignoreFormats.getColor()), isPatternString);
2103 makeKey("latexenvironment", KeyInfo(KeyInfo::isStandard, 2, false), isPatternString);
2105 // Split is done always.
2106 makeKey("foreignlanguage", KeyInfo(KeyInfo::isMain, 2, ignoreFormats.getLanguage()), isPatternString);
2110 makeKey("backslash|textbackslash|slash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2111 makeKey("textasciicircum|textasciitilde", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2112 makeKey("textasciiacute|texemdash", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2113 makeKey("dots|ldots", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2115 makeKey("quad|qquad|hfill|dotfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2116 makeKey("textvisiblespace|nobreakspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2117 makeKey("negthickspace|negmedspace|negthinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2118 makeKey("thickspace|medspace|thinspace", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2120 // makeKey("enskip|smallskip|medskip|bigskip|vfill", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2121 // Custom space/skip, remove the content (== length value)
2122 makeKey("vspace|vspace*|hspace|hspace*|mspace", KeyInfo(KeyInfo::noContent, 1, false), isPatternString);
2123 // Found in fr/UserGuide.lyx
2124 makeKey("og|fg", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2126 makeKey("textquotedbl|quotesinglbase|lyxarrow", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2127 makeKey("textquotedblleft|textquotedblright", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2128 // Known macros to remove (including their parameter)
2130 makeKey("input|inputencoding|label|ref|index|bibitem", KeyInfo(KeyInfo::doRemove, 1, false), isPatternString);
2131 makeKey("addtocounter|setlength", KeyInfo(KeyInfo::noContent, 2, true), isPatternString);
2132 // handle like standard keys with 1 parameter.
2133 makeKey("url|href|vref|thanks", KeyInfo(KeyInfo::isStandard, 1, false), isPatternString);
2135 // Ignore deleted text
2136 makeKey("lyxdeleted", KeyInfo(KeyInfo::doRemove, 3, false), isPatternString);
2137 // but preserve added text
2138 makeKey("lyxadded", KeyInfo(KeyInfo::doRemove, 2, false), isPatternString);
2140 // Macros to remove, but let the parameter survive
2142 makeKey("menuitem|textmd|textrm", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2144 // Remove language spec from content of these insets
2145 makeKey("code", KeyInfo(KeyInfo::noMain, 1, false), isPatternString);
2147 // Same effect as previous, parameter will survive (because there is no one anyway)
2149 makeKey("noindent|textcompwordmark|maketitle", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2150 // Remove table decorations
2151 makeKey("hline|tabularnewline|toprule|bottomrule|midrule", KeyInfo(KeyInfo::doRemove, 0, true), isPatternString);
2152 // Discard shape-header.
2153 // For footnote or shortcut too, because of lang settings
2154 // and wrong handling if used 'KeyInfo::noMain'
2155 makeKey("circlepar|diamondpar|heartpar|nutpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2156 makeKey("trianglerightpar|hexagonpar|starpar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2157 makeKey("triangleuppar|triangledownpar|droppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2158 makeKey("triangleleftpar|shapepar|dropuppar", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2159 makeKey("hphantom|vphantom|footnote|shortcut|include|includegraphics", KeyInfo(KeyInfo::isStandard, 1, true), isPatternString);
2160 makeKey("parbox", KeyInfo(KeyInfo::doRemove, 1, true), isPatternString);
2161 // like ('tiny{}' or '\tiny ' ... )
2162 makeKey("footnotesize|tiny|scriptsize|small|large|Large|LARGE|huge|Huge", KeyInfo(KeyInfo::isSize, 0, false), isPatternString);
2164 // Survives, like known character
2165 // makeKey("lyx|LyX|latex|LaTeX|latexe|LaTeXe|tex|TeX", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2166 makeKey("tableofcontents", KeyInfo(KeyInfo::isChar, 0, false), isPatternString);
2167 makeKey("item|listitem", KeyInfo(KeyInfo::isList, 1, false), isPatternString);
2169 makeKey("begin|end", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2170 makeKey("[|]", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2171 makeKey("$", KeyInfo(KeyInfo::isMath, 1, false), isPatternString);
2173 makeKey("par|uldepth|ULdepth|protect|nobreakdash|medskip|relax", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2174 // Remove RTL/LTR marker
2175 makeKey("l|r|textlr|textfr|textar|beginl|endl", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2176 makeKey("lettrine", KeyInfo(KeyInfo::cleanToStart, 0, true), isPatternString);
2177 makeKey("lyxslide", KeyInfo(KeyInfo::isSectioning, 1, true), isPatternString);
2178 makeKey("endarguments", KeyInfo(KeyInfo::endArguments, 0, true), isPatternString);
2179 makeKey("twocolumn", KeyInfo(KeyInfo::removeWithArg, 2, true), isPatternString);
2180 makeKey("tnotetext|ead|fntext|cortext|address", KeyInfo(KeyInfo::removeWithArg, 0, true), isPatternString);
2181 makeKey("lyxend", KeyInfo(KeyInfo::isStandard, 0, true), isPatternString);
2182 if (isPatternString) {
2183 // Allow the first searched string to rebuild the keys too
2187 // no need to rebuild again
2193 * Keep the list of actual opened parentheses actual
2194 * (e.g. depth == 4 means there are 4 '{' not processed yet)
2196 void Intervall::handleParentheses(int lastpos, bool closingAllowed)
2199 for (int i = depts[actualdeptindex]; i < lastpos; i+= 1 + skip) {
2203 if (c == '\\') skip = 1;
2204 else if (c == '{') {
2207 else if (c == '}') {
2208 handleCloseP(i, closingAllowed);
2214 string Intervall::show(int lastpos)
2216 int idx = 0; /* int intervalls */
2219 for (idx = 0; idx <= ignoreidx; idx++) {
2220 while (i < lastpos) {
2222 if (i <= borders[idx].low) {
2223 if (borders[idx].low > lastpos)
2224 printsize = lastpos - i;
2226 printsize = borders[idx].low - i;
2227 s += par.substr(i, printsize);
2229 if (i >= borders[idx].low)
2230 i = borders[idx].upper;
2233 i = borders[idx].upper;
2239 s += par.substr(i, lastpos-i);
2245 void Intervall::output(ostringstream &os, int lastpos)
2247 // get number of chars to output
2248 int idx = 0; /* int intervalls */
2251 string startTitle = titleValue;
2252 for (idx = 0; idx <= ignoreidx; idx++) {
2254 if (i <= borders[idx].low) {
2256 if (borders[idx].low > lastpos)
2257 printsize = lastpos - i;
2259 printsize = borders[idx].low - i;
2260 if (printsize > 0) {
2261 os << startTitle << par.substr(i, printsize);
2263 printed += printsize;
2266 handleParentheses(i, false);
2267 if (i >= borders[idx].low)
2268 i = borders[idx].upper;
2271 i = borders[idx].upper;
2278 os << startTitle << par.substr(i, lastpos-i);
2279 printed += lastpos-i;
2281 handleParentheses(lastpos, false);
2283 if (keys["foreignlanguage"].disabled)
2284 startindex = actualdeptindex-langcount;
2286 startindex = actualdeptindex;
2287 for (int i = startindex; i > 0; --i) {
2290 if (hasTitle && (printed > 0))
2292 if (! isPatternString_)
2294 handleParentheses(lastpos, true); /* extra closings '}' allowed here */
2297 void LatexInfo::processRegion(int start, int region_end)
2299 while (start < region_end) { /* Let {[} and {]} survive */
2300 int cnt = interval_.isOpeningPar(start);
2302 // Closing is allowed past the region
2303 int closing = interval_.findclosing(start+1, interval_.par.length());
2304 interval_.addIntervall(start, start+1);
2305 interval_.addIntervall(closing, closing+1);
2309 start = interval_.nextNotIgnored(start+1);
2313 void LatexInfo::removeHead(KeyInfo const & actual, int count)
2315 if (actual.parenthesiscount == 0) {
2316 // "{\tiny{} ...}" ==> "{{} ...}"
2317 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart + actual._tokensize);
2320 // Remove header hull, that is "\url{abcd}" ==> "abcd"
2321 interval_.addIntervall(actual._tokenstart - count, actual._dataStart);
2322 interval_.addIntervall(actual._dataEnd, actual._dataEnd+1);
2326 int LatexInfo::dispatch(ostringstream &os, int previousStart, KeyInfo &actual)
2329 switch (actual.keytype)
2331 case KeyInfo::isTitle: {
2333 nextKeyIdx = getNextKey();
2336 case KeyInfo::cleanToStart: {
2337 actual._dataEnd = actual._dataStart;
2338 nextKeyIdx = getNextKey();
2339 // Search for end of arguments
2340 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2342 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2343 entries_[i].disabled = true;
2345 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2347 while (interval_.par[actual._dataEnd] == ' ')
2349 interval_.addIntervall(0, actual._dataEnd+1);
2350 interval_.actualdeptindex = 0;
2351 interval_.depts[0] = actual._dataEnd+1;
2352 interval_.closes[0] = -1;
2355 case KeyInfo::isText:
2356 interval_.par[actual._tokenstart] = '#';
2357 //interval_.addIntervall(actual._tokenstart, actual._tokenstart+1);
2358 nextKeyIdx = getNextKey();
2360 case KeyInfo::noContent: { /* char like "\hspace{2cm}" */
2361 if (actual.disabled)
2362 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2364 interval_.addIntervall(actual._dataStart, actual._dataEnd);
2367 case KeyInfo::isChar: {
2368 nextKeyIdx = getNextKey();
2371 case KeyInfo::isSize: {
2372 if (actual.disabled || (interval_.par[actual._dataStart] != '{') || (interval_.par[actual._dataStart-1] == ' ')) {
2373 if (actual.parenthesiscount == 0)
2374 interval_.addIntervall(actual._tokenstart, actual._dataEnd);
2376 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2378 nextKeyIdx = getNextKey();
2380 // Here _dataStart points to '{', so correct it
2381 actual._dataStart += 1;
2382 actual._tokensize += 1;
2383 actual.parenthesiscount = 1;
2384 if (interval_.par[actual._dataStart] == '}') {
2385 // Determine the end if used like '{\tiny{}...}'
2386 actual._dataEnd = interval_.findclosing(actual._dataStart+1, interval_.par.length()) + 1;
2387 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2390 // Determine the end if used like '\tiny{...}'
2391 actual._dataEnd = interval_.findclosing(actual._dataStart, interval_.par.length()) + 1;
2393 // Split on this key if not at start
2394 int start = interval_.nextNotIgnored(previousStart);
2395 if (start < actual._tokenstart) {
2396 interval_.output(os, actual._tokenstart);
2397 interval_.addIntervall(start, actual._tokenstart);
2399 // discard entry if at end of actual
2400 nextKeyIdx = process(os, actual);
2404 case KeyInfo::endArguments: {
2405 // Remove trailing '{}' too
2406 actual._dataStart += 1;
2407 actual._dataEnd += 1;
2408 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2409 nextKeyIdx = getNextKey();
2412 case KeyInfo::noMain:
2414 case KeyInfo::isStandard: {
2415 if (actual.disabled) {
2417 processRegion(actual._dataStart, actual._dataStart+1);
2418 nextKeyIdx = getNextKey();
2420 // Split on this key if not at datastart of calling entry
2421 int start = interval_.nextNotIgnored(previousStart);
2422 if (start < actual._tokenstart) {
2423 interval_.output(os, actual._tokenstart);
2424 interval_.addIntervall(start, actual._tokenstart);
2426 // discard entry if at end of actual
2427 nextKeyIdx = process(os, actual);
2431 case KeyInfo::removeWithArg: {
2432 nextKeyIdx = getNextKey();
2433 // Search for end of arguments
2434 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2436 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2437 entries_[i].disabled = true;
2439 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2441 interval_.addIntervall(actual._tokenstart, actual._dataEnd+1);
2444 case KeyInfo::doRemove: {
2445 // Remove the key with all parameters and following spaces
2448 if (interval_.par[actual._dataEnd-1] == ' ')
2449 start = actual._dataEnd;
2451 start = actual._dataEnd+1;
2452 for (pos = start; pos < interval_.par.length(); pos++) {
2453 if ((interval_.par[pos] != ' ') && (interval_.par[pos] != '%'))
2456 // Remove also enclosing parentheses [] and {}
2459 while (actual._tokenstart > numpars) {
2460 if (pos+numpars >= interval_.par.size())
2462 else if (interval_.par[pos+numpars] == ']' && interval_.par[actual._tokenstart-numpars-1] == '[')
2464 else if (interval_.par[pos+numpars] == '}' && interval_.par[actual._tokenstart-numpars-1] == '{')
2470 if (interval_.par[pos+numpars] == ' ')
2474 interval_.addIntervall(actual._tokenstart-numpars, pos+numpars+spaces);
2475 nextKeyIdx = getNextKey();
2478 case KeyInfo::isList: {
2479 // Discard space before _tokenstart
2481 for (count = 0; count < actual._tokenstart; count++) {
2482 if (interval_.par[actual._tokenstart-count-1] != ' ')
2485 nextKeyIdx = getNextKey();
2486 int tmpIdx = find(nextKeyIdx, KeyInfo::endArguments);
2488 // Special case: \item is not a list, but a command (like in Style Author_Biography in maa-monthly.layout)
2490 // How else can we catch this one?
2491 for (int i = nextKeyIdx; i <= tmpIdx; i++) {
2492 entries_[i].disabled = true;
2494 actual._dataEnd = entries_[tmpIdx]._dataEnd;
2496 else if (nextKeyIdx > 0) {
2497 // Ignore any lang entries inside data region
2498 for (int i = nextKeyIdx; i < int(entries_.size()) && entries_[i]._tokenstart < actual._dataEnd; i++) {
2499 if (entries_[i].keytype == KeyInfo::isMain)
2500 entries_[i].disabled = true;
2503 if (actual.disabled) {
2504 interval_.addIntervall(actual._tokenstart-count, actual._dataEnd+1);
2507 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2509 if (interval_.par[actual._dataEnd+1] == '[') {
2510 int posdown = interval_.findclosing(actual._dataEnd+2, interval_.par.length(), '[', ']');
2511 if ((interval_.par[actual._dataEnd+2] == '{') &&
2512 (interval_.par[posdown-1] == '}')) {
2513 interval_.addIntervall(actual._dataEnd+1,actual._dataEnd+3);
2514 interval_.addIntervall(posdown-1, posdown+1);
2517 interval_.addIntervall(actual._dataEnd+1, actual._dataEnd+2);
2518 interval_.addIntervall(posdown, posdown+1);
2520 int blk = interval_.nextNotIgnored(actual._dataEnd+1);
2521 if (blk > posdown) {
2522 // Discard at most 1 space after empty item
2524 for (count = 0; count < 1; count++) {
2525 if (interval_.par[blk+count] != ' ')
2529 interval_.addIntervall(blk, blk+count);
2534 case KeyInfo::isSectioning: {
2535 // Discard spaces before _tokenstart
2537 int val = actual._tokenstart;
2538 for (count = 0; count < actual._tokenstart;) {
2539 val = interval_.previousNotIgnored(val-1);
2540 if (val < 0 || interval_.par[val] != ' ')
2543 count = actual._tokenstart - val;
2546 if (actual.disabled) {
2547 removeHead(actual, count);
2548 nextKeyIdx = getNextKey();
2550 interval_.addIntervall(actual._tokenstart-count, actual._tokenstart);
2551 nextKeyIdx = process(os, actual);
2555 case KeyInfo::isMath: {
2556 // Same as regex, use the content unchanged
2557 nextKeyIdx = getNextKey();
2560 case KeyInfo::isRegex: {
2561 // DO NOT SPLIT ON REGEX
2563 nextKeyIdx = getNextKey();
2566 case KeyInfo::isIgnored: {
2567 // Treat like a character for now
2568 nextKeyIdx = getNextKey();
2571 case KeyInfo::isMain: {
2572 if (interval_.par.substr(actual._dataStart, 2) == "% ")
2573 interval_.addIntervall(actual._dataStart, actual._dataStart+2);
2574 if (actual._tokenstart > 0) {
2575 int prev = interval_.previousNotIgnored(actual._tokenstart - 1);
2576 if ((prev >= 0) && interval_.par[prev] == '%')
2577 interval_.addIntervall(prev, prev+1);
2579 if (actual.disabled) {
2581 interval_.langcount++;
2582 if ((interval_.par.substr(actual._dataStart, 3) == " \\[") ||
2583 (interval_.par.substr(actual._dataStart, 8) == " \\begin{")) {
2584 // Discard also the space before math-equation
2585 interval_.addIntervall(actual._dataStart, actual._dataStart+1);
2587 nextKeyIdx = getNextKey();
2588 // interval.resetOpenedP(actual._dataStart-1);
2591 if (actual._tokenstart < 26) {
2592 // for the first (and maybe dummy) language
2593 interval_.setForDefaultLang(actual);
2595 interval_.resetOpenedP(actual._dataStart-1);
2599 case KeyInfo::invalid:
2600 case KeyInfo::headRemove:
2601 // These two cases cannot happen, already handled
2604 // LYXERR(Debug::INFO, "Unhandled keytype");
2605 nextKeyIdx = getNextKey();
2612 int LatexInfo::process(ostringstream & os, KeyInfo const & actual )
2614 int end = interval_.nextNotIgnored(actual._dataEnd);
2615 int oldStart = actual._dataStart;
2616 int nextKeyIdx = getNextKey();
2618 if ((nextKeyIdx < 0) ||
2619 (entries_[nextKeyIdx]._tokenstart >= actual._dataEnd) ||
2620 (entries_[nextKeyIdx].keytype == KeyInfo::invalid)) {
2621 if (oldStart <= end) {
2622 processRegion(oldStart, end);
2627 KeyInfo &nextKey = getKeyInfo(nextKeyIdx);
2629 if ((nextKey.keytype == KeyInfo::isMain) && !nextKey.disabled) {
2630 (void) dispatch(os, actual._dataStart, nextKey);
2631 end = nextKey._tokenstart;
2634 processRegion(oldStart, nextKey._tokenstart);
2635 nextKeyIdx = dispatch(os, actual._dataStart, nextKey);
2637 oldStart = nextKey._dataEnd+1;
2639 // now nextKey is either invalid or is outside of actual._dataEnd
2640 // output the remaining and discard myself
2641 if (oldStart <= end) {
2642 processRegion(oldStart, end);
2644 if (interval_.par.size() > (size_t) end && interval_.par[end] == '}') {
2646 // This is the normal case.
2647 // But if using the firstlanguage, the closing may be missing
2649 // get minimum of 'end' and 'actual._dataEnd' in case that the nextKey.keytype was 'KeyInfo::isMain'
2651 if (actual._dataEnd < end)
2652 output_end = interval_.nextNotIgnored(actual._dataEnd);
2653 else if (interval_.par.size() > (size_t) end)
2654 output_end = interval_.nextNotIgnored(end);
2656 output_end = interval_.par.size();
2657 if ((actual.keytype == KeyInfo::isMain) && actual.disabled) {
2658 interval_.addIntervall(actual._tokenstart, actual._tokenstart+actual._tokensize);
2660 // Remove possible empty data
2661 int dstart = interval_.nextNotIgnored(actual._dataStart);
2662 while (interval_.isOpeningPar(dstart) == 1) {
2663 interval_.addIntervall(dstart, dstart+1);
2664 int dend = interval_.findclosing(dstart+1, output_end);
2665 interval_.addIntervall(dend, dend+1);
2666 dstart = interval_.nextNotIgnored(dstart+1);
2668 if (dstart < output_end)
2669 interval_.output(os, output_end);
2670 interval_.addIntervall(actual._tokenstart, end);
2674 string splitOnKnownMacros(string par, bool isPatternString)
2677 LatexInfo li(par, isPatternString);
2678 // LYXERR(Debug::INFO, "Berfore split: " << par);
2679 KeyInfo DummyKey = KeyInfo(KeyInfo::KeyType::isMain, 2, true);
2681 DummyKey._tokensize = 0;
2682 DummyKey._dataStart = 0;
2683 DummyKey._dataEnd = par.length();
2684 DummyKey.disabled = true;
2685 int firstkeyIdx = li.getFirstKey();
2687 if (firstkeyIdx >= 0) {
2688 KeyInfo firstKey = li.getKeyInfo(firstkeyIdx);
2689 DummyKey._tokenstart = firstKey._tokenstart;
2691 if ((firstKey.keytype != KeyInfo::isMain) || firstKey.disabled) {
2692 // Use dummy firstKey
2693 firstKey = DummyKey;
2694 (void) li.setNextKey(firstkeyIdx);
2697 if (par.substr(firstKey._dataStart, 2) == "% ")
2698 li.addIntervall(firstKey._dataStart, firstKey._dataStart+2);
2700 nextkeyIdx = li.process(os, firstKey);
2701 while (nextkeyIdx >= 0) {
2702 // Check for a possible gap between the last
2703 // entry and this one
2704 int datastart = li.nextNotIgnored(firstKey._dataStart);
2705 KeyInfo &nextKey = li.getKeyInfo(nextkeyIdx);
2706 if ((nextKey._tokenstart > datastart)) {
2708 firstKey._dataStart = datastart;
2709 firstKey._dataEnd = par.length();
2710 (void) li.setNextKey(nextkeyIdx);
2711 // Fake the last opened parenthesis
2712 li.setForDefaultLang(firstKey);
2713 nextkeyIdx = li.process(os, firstKey);
2716 if (nextKey.keytype != KeyInfo::isMain) {
2717 firstKey._dataStart = datastart;
2718 firstKey._dataEnd = nextKey._dataEnd+1;
2719 (void) li.setNextKey(nextkeyIdx);
2720 li.setForDefaultLang(firstKey);
2721 nextkeyIdx = li.process(os, firstKey);
2724 nextkeyIdx = li.process(os, nextKey);
2728 // Handle the remaining
2729 firstKey._dataStart = li.nextNotIgnored(firstKey._dataStart);
2730 firstKey._dataEnd = par.length();
2732 if ((firstKey._dataStart < firstKey._dataEnd) &&
2733 (par[firstKey._dataStart] != '}')) {
2734 li.setForDefaultLang(firstKey);
2735 (void) li.process(os, firstKey);
2739 // return string definitelly impossible to match
2740 s = "\\foreignlanguage{ignore}{ }";
2744 s = par; /* no known macros found */
2745 // LYXERR(Debug::INFO, "After split: " << s);
2750 * Try to unify the language specs in the latexified text.
2751 * Resulting modified string is set to "", if
2752 * the searched tex does not contain all the features in the search pattern
2754 static string correctlanguagesetting(string par, bool isPatternString, bool withformat)
2756 static Features regex_f;
2757 static int missed = 0;
2758 static bool regex_with_format = false;
2760 int parlen = par.length();
2762 while ((parlen > 0) && (par[parlen-1] == '\n')) {
2765 if (isPatternString && (parlen > 0) && (par[parlen-1] == '~')) {
2766 // Happens to be there in case of description or labeling environment
2771 // Split the latex input into pieces which
2772 // can be digested by our search engine
2773 LYXERR(Debug::FIND, "input: \"" << par << "\"");
2774 result = splitOnKnownMacros(par.substr(0,parlen), isPatternString);
2775 LYXERR(Debug::FIND, "After splitOnKnownMacros:\n\"" << result << "\"");
2778 result = par.substr(0, parlen);
2779 if (isPatternString) {
2782 regex_f = identifyFeatures(result);
2783 string features = "";
2784 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2785 string a = it->first;
2786 regex_with_format = true;
2787 features += " " + a;
2788 // LYXERR(Debug::INFO, "Identified regex format:" << a);
2790 LYXERR(Debug::FIND, "Identified Features" << features);
2793 } else if (regex_with_format) {
2794 Features info = identifyFeatures(result);
2795 for (auto it = regex_f.cbegin(); it != regex_f.cend(); ++it) {
2796 string a = it->first;
2797 bool b = it->second;
2798 if (b && ! info[a]) {
2800 LYXERR(Debug::FIND, "Missed(" << missed << " " << a <<", srclen = " << parlen );
2807 // LYXERR(Debug::INFO, "No regex formats");
2813 // Remove trailing closure of math, macros and environments, so to catch parts of them.
2814 static int identifyClosing(string & t)
2816 int open_braces = 0;
2818 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
2819 if (regex_replace(t, t, "(.*[^\\\\])\\$$", "$1"))
2821 if (regex_replace(t, t, "(.*[^\\\\])\\\\\\]$", "$1"))
2823 if (regex_replace(t, t, "(.*[^\\\\])\\\\end\\{[a-zA-Z_]*\\*?\\}$", "$1"))
2825 if (regex_replace(t, t, "(.*[^\\\\])\\}$", "$1")) {
2834 static int num_replaced = 0;
2835 static bool previous_single_replace = true;
2837 void MatchStringAdv::CreateRegexp(FindAndReplaceOptions const & opt, string regexp_str, string regexp2_str, string par_as_string)
2840 // Handle \w properly
2841 QRegularExpression::PatternOptions popts = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::MultilineOption;
2842 if (! opt.casesensitive) {
2843 popts |= QRegularExpression::CaseInsensitiveOption;
2845 regexp = QRegularExpression(QString::fromStdString(regexp_str), popts);
2846 regexp2 = QRegularExpression(QString::fromStdString(regexp2_str), popts);
2848 if (regexp.isValid() && regexp2.isValid()) {
2849 regexIsValid = true;
2850 // Check '{', '}' pairs inside the regex
2853 for (unsigned i = 0; i < par_as_string.size(); i+= skip) {
2854 char c = par_as_string[i];
2861 else if (c == '}') {
2868 if (balanced != 0) {
2869 regexIsValid = false;
2870 regexError = "Unbalanced curly brackets in regexp \"" + regexp_str + "\"";
2874 regexIsValid = false;
2875 if (!regexp.isValid())
2876 regexError += "Invalid regexp \"" + regexp_str + "\", error = " + regexp.errorString().toStdString();
2878 regexError += "Invalid regexp2 \"" + regexp2_str + "\", error = " + regexp2.errorString().toStdString();
2881 if (opt.casesensitive) {
2882 regexp = regex(regexp_str);
2883 regexp2 = regex(regexp2_str);
2886 regexp = regex(regexp_str, std::regex_constants::icase);
2887 regexp2 = regex(regexp2_str, std::regex_constants::icase);
2892 static void modifyRegexForMatchWord(string &t)
2895 regex wordre("(\\\\)*((\\.|\\\\b))");
2898 for (sregex_iterator it(t.begin(), t.end(), wordre), end; it != end; ++it) {
2900 if ((sub.position(2) - sub.position(0)) % 2 == 1) {
2903 else if (sub.str(2) == "\\\\b")
2905 if (lastpos < (size_t) sub.position(2))
2906 s += t.substr(lastpos, sub.position(2) - lastpos);
2908 lastpos = sub.position(2) + sub.length(2);
2911 s = "\\b" + t + "\\b";
2915 else if (lastpos < t.length())
2916 s += t.substr(lastpos, t.length() - lastpos);
2917 t = "\\b" + s + "\\b";
2920 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
2921 : p_buf(&buf), p_first_buf(&buf), opt(opt)
2923 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
2924 docstring const & ds = stringifySearchBuffer(find_buf, opt);
2925 use_regexp = lyx::to_utf8(ds).find("\\regexp{") != std::string::npos;
2926 if (opt.replace_all && previous_single_replace) {
2927 previous_single_replace = false;
2930 else if (!opt.replace_all) {
2931 num_replaced = 0; // count number of replaced strings
2932 previous_single_replace = true;
2934 // When using regexp, braces are hacked already by escape_for_regex()
2935 par_as_string = normalize(ds);
2937 close_wildcards = 0;
2939 size_t lead_size = 0;
2940 // correct the language settings
2941 par_as_string = correctlanguagesetting(par_as_string, true, !opt.ignoreformat);
2942 opt.matchstart = false;
2944 identifyClosing(par_as_string); // Removes math closings ($, ], ...) at end of string
2945 if (opt.ignoreformat) {
2949 lead_size = identifyLeading(par_as_string);
2951 lead_as_string = par_as_string.substr(0, lead_size);
2952 string lead_as_regex_string = string2regex(lead_as_string);
2953 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2954 string par_as_regex_string_nolead = string2regex(par_as_string_nolead);
2955 /* Handle whole words too in this case
2957 if (opt.matchword) {
2958 par_as_regex_string_nolead = "\\b" + par_as_regex_string_nolead + "\\b";
2959 opt.matchword = false;
2961 string regexp_str = "(" + lead_as_regex_string + ")()" + par_as_regex_string_nolead;
2962 string regexp2_str = "(" + lead_as_regex_string + ")(.*?)" + par_as_regex_string_nolead;
2963 CreateRegexp(opt, regexp_str, regexp2_str);
2965 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
2966 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
2970 if (!opt.ignoreformat) {
2971 lead_size = identifyLeading(par_as_string);
2972 LYXERR(Debug::FIND, "Lead_size: " << lead_size);
2973 lead_as_string = par_as_string.substr(0, lead_size);
2974 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
2977 // Here we are using regexp
2978 LASSERT(use_regexp, /**/);
2980 string lead_as_regexp;
2981 if (lead_size > 0) {
2982 lead_as_regexp = string2regex(par_as_string.substr(0, lead_size));
2983 regex_replace(par_as_string_nolead, par_as_string_nolead, "}$", "");
2984 par_as_string = par_as_string_nolead;
2985 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
2986 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2988 // LYXERR(Debug::FIND, "par_as_string before escape_for_regex() is '" << par_as_string << "'");
2989 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
2990 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
2991 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
2993 size_t lng = par_as_string.size();
2994 if (!opt.ignoreformat) {
2995 // Remove extra '\}' at end if not part of \{\.\}
2997 if (par_as_string.substr(lng-2, 2).compare("\\}") == 0) {
2999 if (par_as_string.substr(lng-6,3).compare("\\{\\") == 0)
3008 if (lng < par_as_string.size())
3009 par_as_string = par_as_string.substr(0,lng);
3011 LYXERR(Debug::FIND, "par_as_string after correctRegex is '" << par_as_string << "'");
3012 if ((lng > 0) && (par_as_string[0] == '^')) {
3013 par_as_string = par_as_string.substr(1);
3015 opt.matchstart = true;
3017 // LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
3018 // LYXERR(Debug::FIND, "Open braces: " << open_braces);
3019 // LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
3021 // If entered regexp must match at begin of searched string buffer
3022 // Kornel: Added parentheses to use $1 for size of the leading string
3026 // TODO: Adapt '\[12345678]' in par_as_string to acount for the first '()
3027 // Unfortunately is '\1', '\2', etc not working for strings with extra format
3028 // so the convert has no effect in that case
3029 for (int i = 7; i > 0; --i) {
3030 string orig = "\\\\" + std::to_string(i);
3031 string dest = "\\" + std::to_string(i+2);
3032 while (regex_replace(par_as_string, par_as_string, orig, dest));
3034 if (opt.matchword) {
3035 modifyRegexForMatchWord(par_as_string);
3036 opt.matchword = false;
3038 regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
3039 regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
3041 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
3042 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
3043 CreateRegexp(opt, regexp_str, regexp2_str, par_as_string);
3047 MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
3051 mres.searched_size = len;
3053 (opt.restr == FindAndReplaceOptions::R_ONLY_MATHS && !cur.inMathed()) )
3056 docstring docstr = stringifyFromForSearch(opt, cur, len);
3058 str = normalize(docstr);
3059 if (!opt.ignoreformat) {
3060 str = correctlanguagesetting(str, false, !opt.ignoreformat);
3061 // remove closing '}' and '\n' to allow for use of '$' in regex
3062 size_t lng = str.size();
3063 while ((lng > 1) && ((str[lng -1] == '}') || (str[lng -1] == '\n')))
3065 if (lng != str.size()) {
3066 str = str.substr(0, lng);
3070 mres.match_len = -1;
3073 LYXERR(Debug::FIND, "After normalization: Matching against:\n'" << str << "'");
3075 LASSERT(use_regexp, /**/);
3077 // use_regexp always true
3078 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
3080 QString qstr = QString::fromStdString(str);
3081 QRegularExpression const *p_regexp;
3082 QRegularExpression::MatchType flags = QRegularExpression::NormalMatch;
3086 p_regexp = ®exp2;
3088 QRegularExpressionMatch match = p_regexp->match(qstr, 0, flags);
3089 if (!match.hasMatch())
3092 regex const *p_regexp;
3093 regex_constants::match_flag_type flags;
3095 flags = regex_constants::match_continuous;
3098 flags = regex_constants::match_default;
3099 p_regexp = ®exp2;
3101 sregex_iterator re_it(str.begin(), str.end(), *p_regexp, flags);
3102 if (re_it == sregex_iterator())
3104 match_results<string::const_iterator> const & m = *re_it;
3106 // Whole found string, including the leading
3107 // std: m[0].second - m[0].first
3108 // Qt: match.capturedEnd(0) - match.capturedStart(0)
3110 // Size of the leading string
3111 // std: m[1].second - m[1].first
3112 // Qt: match.capturedEnd(1) - match.capturedStart(1)
3113 int leadingsize = 0;
3115 if (match.lastCapturedIndex() > 0) {
3116 leadingsize = match.capturedEnd(1) - match.capturedStart(1);
3121 leadingsize = m[1].second - m[1].first;
3125 mres.match_prefix = match.capturedEnd(2) - match.capturedStart(2);
3126 mres.match_len = match.capturedEnd(0) - match.capturedEnd(2);
3127 // because of different number of closing at end of string
3128 // we have to 'unify' the length of the post-match.
3129 // Done by ignoring closing parenthesis and linefeeds at string end
3130 int matchend = match.capturedEnd(0);
3131 while (mres.match_len > 0) {
3132 QChar c = qstr.at(matchend - 1);
3133 if ((c == '\n') || (c == '}') || (c == '{')) {
3140 size_t strsize = qstr.size();
3141 while (strsize > (size_t) match.capturedEnd(0)) {
3142 QChar c = qstr.at(strsize-1);
3143 if ((c == '\n') || (c == '}')) {
3149 // LYXERR0(qstr.toStdString());
3150 mres.match2end = strsize - matchend;
3151 mres.pos = match.capturedStart(2);
3153 mres.match_prefix = m[2].second - m[2].first;
3154 mres.match_len = m[0].second - m[2].second;
3155 // ignore closing parenthesis and linefeeds at string end
3156 size_t strend = m[0].second - m[0].first;
3157 int matchend = strend;
3158 while (mres.match_len > 0) {
3159 char c = str.at(matchend - 1);
3160 if ((c == '\n') || (c == '}') || (c == '{')) {
3167 size_t strsize = str.size();
3168 while (strsize > strend) {
3169 if ((str.at(strsize-1) == '}') || (str.at(strsize-1) == '\n')) {
3176 mres.match2end = strsize - matchend;
3177 mres.pos = m[2].first - m[0].first;;
3179 if (mres.match2end < 0)
3181 mres.leadsize = leadingsize;
3183 if (mres.match_len > 0) {
3184 string a0 = match.captured(0).mid(mres.pos + mres.match_prefix, mres.match_len).toStdString();
3185 mres.result.push_back(a0);
3186 for (int i = 3; i <= match.lastCapturedIndex(); i++) {
3187 mres.result.push_back(match.captured(i).toStdString());
3191 if (mres.match_len > 0) {
3192 string a0 = m[0].str().substr(mres.pos + mres.match_prefix, mres.match_len);
3193 mres.result.push_back(a0);
3194 for (size_t i = 3; i < m.size(); i++) {
3195 mres.result.push_back(m[i]);
3204 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
3206 MatchResult mres = findAux(cur, len, at_begin);
3207 int res = mres.match_len;
3209 "res=" << res << ", at_begin=" << at_begin
3210 << ", matchstart=" << opt.matchstart
3211 << ", inTexted=" << cur.inTexted());
3212 if (opt.matchstart) {
3215 else if (mres.match_prefix > 0)
3224 static bool simple_replace(string &t, string from, string to)
3226 regex repl("(\\\\)*(" + from + ")");
3230 for (sregex_iterator it(t.begin(), t.end(), repl), end; it != end; ++it) {
3232 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3234 if (lastpos < (size_t) sub.position(2))
3235 s += t.substr(lastpos, sub.position(2) - lastpos);
3237 lastpos = sub.position(2) + sub.length(2);
3241 else if (lastpos < t.length())
3242 s += t.substr(lastpos, t.length() - lastpos);
3248 string MatchStringAdv::normalize(docstring const & s) const
3251 t = lyx::to_utf8(s);
3252 // Remove \n at begin
3253 while (!t.empty() && t[0] == '\n')
3256 while (!t.empty() && t[t.size() - 1] == '\n')
3257 t = t.substr(0, t.size() - 1);
3259 // Handle all other '\n'
3260 while ((pos = t.find("\n")) != string::npos) {
3261 if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
3263 if (isAlnumASCII(t[pos+1])) {
3264 t.replace(pos-2, 3, " ");
3267 t.replace(pos-2, 3, "");
3270 else if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
3271 // '\n' adjacent to non-alpha-numerics, discard
3272 t.replace(pos, 1, "");
3275 // Replace all other \n with spaces
3276 t.replace(pos, 1, " ");
3279 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
3280 // Kornel: Added textsl, textsf, textit, texttt and noun
3281 // + allow to seach for colored text too
3282 LYXERR(Debug::FIND, "Removing stale empty macros from: " << t);
3283 while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
3284 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3285 while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
3286 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
3287 while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
3293 docstring stringifyFromCursor(DocIterator const & cur, int len)
3295 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
3296 if (cur.inTexted()) {
3297 Paragraph const & par = cur.paragraph();
3298 // TODO what about searching beyond/across paragraph breaks ?
3299 // TODO Try adding a AS_STR_INSERTS as last arg
3300 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
3301 int(par.size()) : cur.pos() + len;
3302 // OutputParams runparams(&cur.buffer()->params().encoding());
3303 OutputParams runparams(encodings.fromLyXName("utf8"));
3304 runparams.nice = true;
3305 runparams.flavor = Flavor::XeTeX;
3306 runparams.linelen = 10000; //lyxrc.plaintext_linelen;
3307 // No side effect of file copying and image conversion
3308 runparams.dryrun = true;
3309 runparams.for_search = true;
3310 LYXERR(Debug::FIND, "Stringifying with cur: "
3311 << cur << ", from pos: " << cur.pos() << ", end: " << end);
3312 return par.asString(cur.pos(), end,
3313 AS_STR_INSETS | AS_STR_SKIPDELETE | AS_STR_PLAINTEXT,
3315 } else if (cur.inMathed()) {
3316 CursorSlice cs = cur.top();
3317 MathData md = cs.cell();
3318 MathData::const_iterator it_end =
3319 (( len == -1 || cs.pos() + len > int(md.size()))
3321 : md.begin() + cs.pos() + len );
3323 for (MathData::const_iterator it = md.begin() + cs.pos();
3326 docstring s = asString(md2);
3327 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
3330 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3335 /** Computes the LaTeX export of buf starting from cur and ending len positions
3336 * after cur, if len is positive, or at the paragraph or innermost inset end
3339 docstring latexifyFromCursor(DocIterator const & cur, int len)
3342 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
3343 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
3344 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
3346 Buffer const & buf = *cur.buffer();
3348 odocstringstream ods;
3350 //OutputParams runparams(&buf.params().encoding());
3351 OutputParams runparams(encodings.fromLyXName("utf8"));
3352 runparams.nice = false;
3353 runparams.flavor = Flavor::XeTeX;
3354 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3355 // No side effect of file copying and image conversion
3356 runparams.dryrun = true;
3357 runparams.for_search = true;
3359 if (cur.inTexted()) {
3360 // @TODO what about searching beyond/across paragraph breaks ?
3361 pos_type endpos = cur.paragraph().size();
3362 if (len != -1 && endpos > cur.pos() + len)
3363 endpos = cur.pos() + len;
3364 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
3365 string(), cur.pos(), endpos);
3366 string s = lyx::to_utf8(ods.str());
3367 LYXERR(Debug::FIND, "Latexified +modified text: '" << s << "'");
3368 return(lyx::from_utf8(s));
3369 } else if (cur.inMathed()) {
3370 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
3371 for (int s = cur.depth() - 1; s >= 0; --s) {
3372 CursorSlice const & cs = cur[s];
3373 if (cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
3374 TeXMathStream ws(os);
3375 cs.asInsetMath()->asHullInset()->header_write(ws);
3380 CursorSlice const & cs = cur.top();
3381 MathData md = cs.cell();
3382 MathData::const_iterator it_end =
3383 ((len == -1 || cs.pos() + len > int(md.size()))
3385 : md.begin() + cs.pos() + len);
3387 for (MathData::const_iterator it = md.begin() + cs.pos();
3391 ods << asString(md2);
3392 // Retrieve the math environment type, and add '$' or '$]'
3393 // or others (\end{equation}) accordingly
3394 for (int s = cur.depth() - 1; s >= 0; --s) {
3395 CursorSlice const & cs2 = cur[s];
3396 InsetMath * inset = cs2.asInsetMath();
3397 if (inset && inset->asHullInset()) {
3398 TeXMathStream ws(os);
3399 inset->asHullInset()->footer_write(ws);
3403 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
3405 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
3410 #if defined(ResultsDebug)
3412 static void displayMResult(MatchResult &mres, string from, DocIterator & cur)
3414 LYXERR0( "from:\t\t\t" << from);
3416 if (mres.pos_len > 0) {
3418 status = "FINALSEARCH";
3421 if (mres.match_len > 0) {
3422 if ((mres.match_prefix == 0) && (mres.pos == mres.leadsize))
3423 status = "Good Match";
3425 status = "Matched in";
3428 status = "MissedSearch";
3431 LYXERR0( status << "(" << cur.pos() << " ... " << mres.searched_size + cur.pos() << ") cur.lastpos(" << cur.lastpos() << ")");
3432 if ((mres.leadsize > 0) || (mres.match_len > 0) || (mres.match2end > 0))
3433 LYXERR0( "leadsize(" << mres.leadsize << ") match_len(" << mres.match_len << ") match2end(" << mres.match2end << ")");
3434 if ((mres.pos > 0) || (mres.match_prefix > 0))
3435 LYXERR0( "pos(" << mres.pos << ") match_prefix(" << mres.match_prefix << ")");
3436 for (size_t i = 0; i < mres.result.size(); i++)
3437 LYXERR0( "Match " << i << " = \"" << mres.result[i] << "\"");
3439 #define displayMres(s, txt, cur) displayMResult(s, txt, cur);
3441 #define displayMres(s, txt, cur)
3444 /** Finalize an advanced find operation, advancing the cursor to the innermost
3445 ** position that matches, plus computing the length of the matching text to
3447 ** Return the cur.pos() difference between start and end of found match
3449 MatchResult &findAdvFinalize(DocIterator & cur, MatchStringAdv const & match, MatchResult const & expected = MatchResult(-1))
3451 // Search the foremost position that matches (avoids find of entire math
3452 // inset when match at start of it)
3453 DocIterator old_cur(cur.buffer());
3455 static MatchResult fail = MatchResult();
3456 static MatchResult max_match;
3457 // If (prefix_len > 0) means that forwarding 1 position will remove the complete entry
3458 // Happens with e.g. hyperlinks
3459 // either one sees "http://www.bla.bla" or nothing
3460 // so the search for "www" gives prefix_len = 7 (== sizeof("http://")
3461 // and although we search for only 3 chars, we find the whole hyperlink inset
3462 bool at_begin = (expected.match_prefix == 0);
3463 LASSERT(at_begin, /**/);
3464 if (expected.match_len > 0 && at_begin) {
3465 // Search for deepest match
3467 max_match = expected;
3469 size_t d = cur.depth();
3473 if (cur.depth() < d)
3475 if (cur.depth() == d)
3478 while (cur && cur.depth() > lastd) {
3479 lastd = cur.depth();
3480 mres = match(cur, -1, at_begin);
3481 displayMres(mres, "Checking innermost", cur);
3482 if (mres.match_len > 0)
3487 if (mres.match_len < expected.match_len)
3495 // (expected.match_len <= 0)
3496 mres = match(cur); /* match valid only if not searching whole words */
3497 displayMres(mres, "Start with negative match", cur);
3500 if (max_match.match_len <= 0) return fail;
3501 LYXERR(Debug::FIND, "Ok");
3503 // Compute the match length
3505 if (cur.pos() + len > cur.lastpos())
3508 LASSERT(match.use_regexp, /**/);
3511 int maxl = cur.lastpos() - cur.pos();
3512 // Greedy behaviour while matching regexps
3513 while (maxl > minl) {
3515 mres2 = match(cur, len, at_begin);
3516 displayMres(mres2, "Finalize loop", cur);
3517 int actual_match_len = mres2.match_len;
3518 if (actual_match_len >= max_match.match_len) {
3519 // actual_match_len > max_match _can_ happen,
3520 // if the search area splits
3521 // some following word so that the regex
3522 // (e.g. 'r.*r\b' matches 'r' from the middle of the
3524 // This means, the len value is too big
3525 actual_match_len = max_match.match_len;
3527 max_match.match_len = actual_match_len;
3529 if (maxl - minl < 4)
3530 len = (int)((maxl + minl)/2);
3532 len = (int)(minl + (maxl - minl + 3)/4);
3535 // (actual_match_len < max_match.match_len)
3537 len = (int)((maxl + minl)/2);
3542 // Search for real start of matched characters
3544 MatchResult actual_match;
3547 } while (cur.depth() > old_cur.depth()); /* Skip inner insets */
3548 if (cur.depth() < old_cur.depth()) {
3550 LYXERR(Debug::INFO, "cur.depth() < old_cur.depth(), this should never happen");
3553 if (cur.pos() != old_cur.pos()) {
3554 // OK, forwarded 1 pos in actual inset
3555 actual_match = match(cur, len-1, at_begin);
3556 if (actual_match.match_len == max_match.match_len) {
3557 // Ha, got it! The shorter selection has the same match length
3560 max_match = actual_match;
3563 // OK, the shorter selection matches less chars, revert to previous value
3569 LYXERR(Debug::INFO, "cur.pos() == old_cur.pos(), this should never happen");
3570 actual_match = match(cur, len, at_begin);
3571 if (actual_match.match_len == max_match.match_len) {
3573 max_match = actual_match;
3580 max_match.pos_len = len;
3581 displayMres(max_match, "SEARCH RESULT", cur)
3588 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
3592 bool repeat = false;
3593 DocIterator orig_cur; // to be used if repeat not successful
3594 MatchResult orig_mres;
3595 while (!theApp()->longOperationCancelled() && cur) {
3596 //(void) findAdvForwardInnermost(cur);
3597 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
3598 MatchResult mres = match(cur, -1, false);
3599 string msg = "Starting";
3602 displayMres(mres, msg + " findForwardAdv", cur)
3603 int match_len = mres.match_len;
3604 if ((mres.pos > 100000) || (mres.match2end > 100000) || (match_len > 100000)) {
3605 LYXERR(Debug::INFO, "BIG LENGTHS: " << mres.pos << ", " << match_len << ", " << mres.match2end);
3608 if (match_len <= 0) {
3609 // This should exit nested insets, if any, or otherwise undefine the currsor.
3610 cur.pos() = cur.lastpos();
3611 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
3614 else { // match_len > 0
3615 // Try to find the begin of searched string
3617 int firstInvalid = 100000;
3619 int incrmatch = (mres.match_prefix + mres.pos - mres.leadsize + 1)*3/4;
3620 int incrcur = (cur.lastpos() - cur.pos() + 1 )*3/4;
3621 if (incrcur < incrmatch)
3622 increment = incrcur;
3624 increment = incrmatch;
3628 LYXERR(Debug::FIND, "Set increment to " << increment);
3629 while (increment > 0) {
3630 DocIterator old_cur = cur;
3631 size_t skipping = cur.depth();
3632 for (int i = 0; i < increment && cur; i++) {
3634 while (cur && cur.depth() > skipping) {
3635 cur.pos() = cur.lastpos();
3639 if (! cur || (cur.pit() > old_cur.pit())) {
3640 // Are we outside of the paragraph?
3641 // This can happen if moving past some UTF8-encoded chars
3646 MatchResult mres2 = match(cur, -1, false);
3647 displayMres(mres2, "findForwardAdv loop", cur)
3648 switch (interpretMatch(mres, mres2)) {
3649 case MatchResult::newIsTooFar:
3650 // behind the expected match
3651 firstInvalid = increment;
3655 case MatchResult::newIsBetter:
3656 // not reached yet, but cur.pos()+increment is bettert
3658 firstInvalid -= increment;
3659 if (increment > firstInvalid*3/4)
3660 increment = firstInvalid*3/4;
3661 if ((mres2.pos == mres2.leadsize) && (increment >= mres2.match_prefix)) {
3662 if (increment >= mres2.match_prefix)
3663 increment = (mres2.match_prefix+1)*3/4;
3668 // Handle not like MatchResult::newIsTooFar
3669 LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
3671 increment = increment*3/4;
3677 if (mres.match_len > 0) {
3678 if (mres.match_prefix + mres.pos - mres.leadsize > 0) {
3687 // seems to never be reached.
3691 // LYXERR0("Leaving first loop");
3692 LYXERR(Debug::FIND, "Finalizing 1");
3693 MatchResult found_match = findAdvFinalize(cur, match, mres);
3694 if (found_match.match_len > 0) {
3695 LASSERT(found_match.pos_len > 0, /**/);
3696 match.FillResults(found_match);
3697 return found_match.pos_len;
3700 // try next possible match
3711 /// Find the most backward consecutive match within same paragraph while searching backwards.
3712 MatchResult &findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
3714 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3715 DocIterator tmp_cur = cur;
3716 static MatchResult mr = findAdvFinalize(tmp_cur, match, MatchResult(-1));
3717 Inset & inset = cur.inset();
3718 for (; cur != cur_begin; cur.backwardPos()) {
3719 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
3720 DocIterator new_cur = cur;
3721 new_cur.backwardPos();
3722 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur).match_len)
3724 MatchResult new_mr = findAdvFinalize(new_cur, match, MatchResult(-1));
3725 if (new_mr.match_len == mr.match_len)
3729 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
3735 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match)
3739 // Backup of original position
3740 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
3741 if (cur == cur_begin)
3744 DocIterator cur_orig(cur);
3745 bool pit_changed = false;
3748 bool found_match = (match(cur, -1, false).match_len > 0);
3752 cur.pos() = cur.lastpos();
3754 cur.pos() = cur_orig.pos();
3755 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
3756 DocIterator cur_prev_iter;
3758 found_match = (match(cur).match_len > 0);
3759 LYXERR(Debug::FIND, "findBackAdv3: found_match="
3760 << found_match << ", cur: " << cur);
3762 MatchResult found_mr = findMostBackwards(cur, match);
3763 match.FillResults(found_mr);
3764 LASSERT(found_mr.pos_len > 0, /**/);
3765 return found_mr.pos_len;
3768 // Stop if begin of document reached
3769 if (cur == cur_begin)
3771 cur_prev_iter = cur;
3775 if (cur == cur_begin)
3782 } while (!theApp()->longOperationCancelled());
3790 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
3791 DocIterator const & cur, int len)
3793 if (cur.pos() < 0 || cur.pos() > cur.lastpos())
3795 if (!opt.ignoreformat)
3796 return latexifyFromCursor(cur, len);
3798 return stringifyFromCursor(cur, len);
3802 FindAndReplaceOptions::FindAndReplaceOptions(
3803 docstring const & _find_buf_name, bool _casesensitive,
3804 bool _matchword, bool _forward, bool _expandmacros, bool _ignoreformat,
3805 docstring const & _repl_buf_name, bool _keep_case,
3806 SearchScope _scope, SearchRestriction _restr, bool _replace_all)
3807 : find_buf_name(_find_buf_name), casesensitive(_casesensitive), matchword(_matchword),
3808 forward(_forward), expandmacros(_expandmacros), ignoreformat(_ignoreformat),
3809 repl_buf_name(_repl_buf_name), keep_case(_keep_case), scope(_scope), restr(_restr), replace_all(_replace_all)
3817 /** Check if 'len' letters following cursor are all non-lowercase */
3818 static bool allNonLowercase(Cursor const & cur, int len)
3820 pos_type beg_pos = cur.selectionBegin().pos();
3821 pos_type end_pos = cur.selectionBegin().pos() + len;
3822 if (len > cur.lastpos() + 1 - beg_pos) {
3823 LYXERR(Debug::FIND, "This should not happen, more debug needed");
3824 len = cur.lastpos() + 1 - beg_pos;
3825 end_pos = beg_pos + len;
3827 for (pos_type pos = beg_pos; pos != end_pos; ++pos)
3828 if (isLowerCase(cur.paragraph().getChar(pos)))
3834 /** Check if first letter is upper case and second one is lower case */
3835 static bool firstUppercase(Cursor const & cur)
3838 pos_type pos = cur.selectionBegin().pos();
3839 if (pos >= cur.lastpos() - 1) {
3840 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
3843 ch1 = cur.paragraph().getChar(pos);
3844 ch2 = cur.paragraph().getChar(pos + 1);
3845 bool result = isUpperCase(ch1) && isLowerCase(ch2);
3846 LYXERR(Debug::FIND, "firstUppercase(): "
3847 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
3848 << ch2 << "(" << char(ch2) << ")"
3849 << ", result=" << result << ", cur=" << cur);
3854 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
3856 ** \fixme What to do with possible further paragraphs in replace buffer ?
3858 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case)
3860 ParagraphList::iterator pit = buffer.paragraphs().begin();
3861 LASSERT(!pit->empty(), /**/);
3862 pos_type right = pos_type(1);
3863 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
3864 right = pit->size();
3865 pit->changeCase(buffer.params(), pos_type(1), right, others_case);
3869 static bool replaceMatches(string &t, int maxmatchnum, vector <string> const & replacements)
3871 // Should replace the string "$" + std::to_string(matchnum) with replacement
3872 // if the char '$' is not prefixed with odd number of char '\\'
3873 static regex const rematch("(\\\\)*(\\$\\$([0-9]))");
3877 for (sregex_iterator it(t.begin(), t.end(), rematch), end; it != end; ++it) {
3879 if ((sub.position(2) - sub.position(0)) % 2 == 1)
3881 int num = stoi(sub.str(3), nullptr, 10);
3882 if (num >= maxmatchnum)
3884 if (lastpos < (size_t) sub.position(2))
3885 s += t.substr(lastpos, sub.position(2) - lastpos);
3886 s += replacements[num];
3887 lastpos = sub.position(2) + sub.length(2);
3891 else if (lastpos < t.length())
3892 s += t.substr(lastpos, t.length() - lastpos);
3898 static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
3900 Cursor & cur = bv->cursor();
3901 if (opt.repl_buf_name.empty()
3902 || theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true) == 0
3903 || theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
3906 DocIterator sel_beg = cur.selectionBegin();
3907 DocIterator sel_end = cur.selectionEnd();
3908 if (&sel_beg.inset() != &sel_end.inset()
3909 || sel_beg.pit() != sel_end.pit()
3910 || sel_beg.idx() != sel_end.idx())
3912 int sel_len = sel_end.pos() - sel_beg.pos();
3913 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
3914 << ", sel_len: " << sel_len << endl);
3917 LASSERT(sel_len > 0, return 0);
3919 if (!matchAdv(sel_beg, sel_len).match_len)
3922 // Build a copy of the replace buffer, adapted to the KeepCase option
3923 Buffer const & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
3925 repl_buffer_orig.write(oss);
3926 string lyx = oss.str();
3927 if (matchAdv.valid_matches > 0) {
3928 replaceMatches(lyx, matchAdv.valid_matches, matchAdv.matches);
3930 Buffer repl_buffer("", false);
3931 repl_buffer.setUnnamed(true);
3932 LASSERT(repl_buffer.readString(lyx), return 0);
3933 if (opt.keep_case && sel_len >= 2) {
3934 LYXERR(Debug::FIND, "keep_case true: cur.pos()=" << cur.pos() << ", sel_len=" << sel_len);
3935 if (cur.inTexted()) {
3936 if (firstUppercase(cur))
3937 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
3938 else if (allNonLowercase(cur, sel_len))
3939 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
3942 cap::cutSelection(cur, false);
3943 if (cur.inTexted()) {
3944 repl_buffer.changeLanguage(
3945 repl_buffer.language(),
3946 cur.getFont().language());
3947 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
3948 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
3949 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
3950 repl_buffer.params().documentClassPtr(),
3951 bv->buffer().errorList("Paste"));
3952 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
3953 sel_len = repl_buffer.paragraphs().begin()->size();
3954 } else if (cur.inMathed()) {
3955 odocstringstream ods;
3957 // OutputParams runparams(&repl_buffer.params().encoding());
3958 OutputParams runparams(encodings.fromLyXName("utf8"));
3959 runparams.nice = false;
3960 runparams.flavor = Flavor::XeTeX;
3961 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
3962 runparams.dryrun = true;
3963 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
3964 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
3965 docstring repl_latex = ods.str();
3966 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
3968 (void)regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
3969 (void)regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
3970 repl_latex = from_utf8(s);
3971 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
3972 MathData ar(cur.buffer());
3973 asArray(repl_latex, ar, Parse::NORMAL);
3975 sel_len = ar.size();
3976 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3978 if (cur.pos() >= sel_len)
3979 cur.pos() -= sel_len;
3982 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
3983 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
3984 bv->processUpdateFlags(Update::Force);
3989 /// Perform a FindAdv operation.
3990 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
3995 // e.g., when invoking word-findadv from mini-buffer wither with
3996 // wrong options syntax or before ever opening advanced F&R pane
3997 if (theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true) == 0)
4001 MatchStringAdv matchAdv(bv->buffer(), opt);
4003 if (!matchAdv.regexIsValid) {
4004 bv->message(lyx::from_utf8(matchAdv.regexError));
4008 int length = bv->cursor().selectionEnd().pos() - bv->cursor().selectionBegin().pos();
4010 bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
4011 num_replaced += findAdvReplace(bv, opt, matchAdv);
4014 pos_len = findForwardAdv(cur, matchAdv);
4016 pos_len = findBackwardsAdv(cur, matchAdv);
4017 } catch (exception & ex) {
4018 bv->message(from_utf8(ex.what()));
4023 if (num_replaced > 0) {
4024 switch (num_replaced)
4027 bv->message(_("One match has been replaced."));
4030 bv->message(_("Two matches have been replaced."));
4033 bv->message(bformat(_("%1$d matches have been replaced."), num_replaced));
4039 bv->message(_("Match not found."));
4044 if (num_replaced > 0)
4045 bv->message(_("Match has been replaced."));
4047 bv->message(_("Match found."));
4049 if (cur.pos() + pos_len > cur.lastpos()) {
4050 // Prevent crash in bv->putSelectionAt()
4051 // Should never happen, maybe LASSERT() here?
4052 pos_len = cur.lastpos() - cur.pos();
4054 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << pos_len);
4055 bv->putSelectionAt(cur, pos_len, !opt.forward);
4061 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
4063 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
4064 << opt.casesensitive << ' '
4065 << opt.matchword << ' '
4066 << opt.forward << ' '
4067 << opt.expandmacros << ' '
4068 << opt.ignoreformat << ' '
4069 << opt.replace_all << ' '
4070 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
4071 << opt.keep_case << ' '
4072 << int(opt.scope) << ' '
4075 LYXERR(Debug::FIND, "built: " << os.str());
4081 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
4083 // LYXERR(Debug::FIND, "parsing");
4087 while (line != "EOSS") {
4091 if (is.eof()) // Tolerate malformed request
4095 // LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
4096 opt.find_buf_name = from_utf8(s);
4097 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat >> opt.replace_all;
4098 is.get(); // Waste space before replace string
4101 while (line != "EOSS") {
4105 if (is.eof()) // Tolerate malformed request
4109 // LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
4110 opt.repl_buf_name = from_utf8(s);
4111 is >> opt.keep_case;
4114 opt.scope = FindAndReplaceOptions::SearchScope(i);
4116 opt.restr = FindAndReplaceOptions::SearchRestriction(i);
4119 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
4120 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case << ' '
4121 << opt.scope << ' ' << opt.restr);