3 * This file is part of LyX, the document processor.
4 * License details can be found in the file COPYING.
6 * \author Lars Gullik Bjønnes
9 * \author Alfredo Braunstein
10 * \author Tommaso Cucinotta
12 * Full author contact details are available in file CREDITS.
20 #include "buffer_funcs.h"
21 #include "BufferList.h"
22 #include "BufferParams.h"
23 #include "BufferView.h"
26 #include "CutAndPaste.h"
27 #include "FuncRequest.h"
29 #include "output_latex.h"
30 #include "OutputParams.h"
31 #include "Paragraph.h"
32 #include "ParIterator.h"
36 #include "frontends/Application.h"
37 #include "frontends/alert.h"
39 #include "mathed/InsetMath.h"
40 #include "mathed/InsetMathGrid.h"
41 #include "mathed/InsetMathHull.h"
42 #include "mathed/MathStream.h"
43 #include "mathed/MathSupport.h"
45 #include "support/convert.h"
46 #include "support/debug.h"
47 #include "support/docstream.h"
48 #include "support/gettext.h"
49 #include "support/lassert.h"
50 #include "support/lstrings.h"
52 #include "support/regex.h"
53 #include <boost/next_prior.hpp>
56 using namespace lyx::support;
62 bool parse_bool(docstring & howto)
67 howto = split(howto, var, ' ');
72 class MatchString : public binary_function<Paragraph, pos_type, int>
75 MatchString(docstring const & str, bool cs, bool mw)
76 : str(str), case_sens(cs), whole_words(mw)
79 // returns true if the specified string is at the specified position
80 // del specifies whether deleted strings in ct mode will be considered
81 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
83 return par.find(str, case_sens, whole_words, pos, del);
91 // match whole words only
96 int findForward(DocIterator & cur, MatchString const & match,
99 for (; cur; cur.forwardChar())
100 if (cur.inTexted()) {
101 int len = match(cur.paragraph(), cur.pos(), find_del);
109 int findBackwards(DocIterator & cur, MatchString const & match,
110 bool find_del = true)
114 if (cur.inTexted()) {
115 int len = match(cur.paragraph(), cur.pos(), find_del);
124 bool findChange(DocIterator & cur, bool next)
128 for (; cur; next ? cur.forwardPos() : cur.backwardPos())
129 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos())) {
131 // if we search backwards, take a step forward
132 // to correctly set the anchor
141 bool searchAllowed(docstring const & str)
144 frontend::Alert::error(_("Search error"), _("Search string is empty"));
151 bool findOne(BufferView * bv, docstring const & searchstr,
152 bool case_sens, bool whole, bool forward, bool find_del = true)
154 if (!searchAllowed(searchstr))
157 DocIterator cur = bv->cursor();
159 MatchString const match(searchstr, case_sens, whole);
161 int match_len = forward ? findForward(cur, match, find_del) :
162 findBackwards(cur, match, find_del);
165 bv->putSelectionAt(cur, match_len, !forward);
167 return match_len > 0;
171 int replaceAll(BufferView * bv,
172 docstring const & searchstr, docstring const & replacestr,
173 bool case_sens, bool whole)
175 Buffer & buf = bv->buffer();
177 if (!searchAllowed(searchstr) || buf.isReadonly())
180 DocIterator cur_orig(bv->cursor());
182 MatchString const match(searchstr, case_sens, whole);
185 int const rsize = replacestr.size();
186 int const ssize = searchstr.size();
189 cur.setCursor(doc_iterator_begin(&buf));
190 int match_len = findForward(cur, match, false);
191 while (match_len > 0) {
192 // Backup current cursor position and font.
193 pos_type const pos = cur.pos();
194 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
196 int striked = ssize - cur.paragraph().eraseChars(pos, pos + match_len,
197 buf.params().trackChanges);
198 cur.paragraph().insert(pos, replacestr, font,
199 Change(buf.params().trackChanges ?
200 Change::INSERTED : Change::UNCHANGED));
201 for (int i = 0; i < rsize + striked; ++i)
204 match_len = findForward(cur, match, false);
207 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
209 cur_orig.fixIfBroken();
210 bv->setCursor(cur_orig);
216 // the idea here is that we are going to replace the string that
217 // is selected IF it is the search string.
218 // if there is a selection, but it is not the search string, then
219 // we basically ignore it. (FIXME We ought to replace only within
221 // if there is no selection, then:
222 // (i) if some search string has been provided, then we find it.
223 // (think of how the dialog works when you hit "replace" the
225 // (ii) if no search string has been provided, then we treat the
226 // word the cursor is in as the search string. (why? i have no
227 // idea.) but this only works in text?
229 // returns the number of replacements made (one, if any) and
230 // whether anything at all was done.
231 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
232 docstring const & replacestr, bool case_sens,
233 bool whole, bool forward, bool findnext)
235 Cursor & cur = bv->cursor();
236 if (!cur.selection()) {
237 // no selection, non-empty search string: find it
238 if (!searchstr.empty()) {
239 findOne(bv, searchstr, case_sens, whole, forward);
240 return pair<bool, int>(true, 0);
242 // empty search string
245 return pair<bool, int>(false, 0);
246 // select current word and treat it as the search string
247 cur.innerText()->selectWord(cur, WHOLE_WORD);
248 searchstr = cur.selectionAsString(false);
251 // if we still don't have a search string, report the error
253 if (!searchAllowed(searchstr))
254 return pair<bool, int>(false, 0);
256 bool have_selection = cur.selection();
257 docstring const selected = cur.selectionAsString(false);
259 case_sens ? searchstr == selected
260 : compare_no_case(searchstr, selected) == 0;
262 // no selection or current selection is not search word:
263 // just find the search word
264 if (!have_selection || !match) {
265 findOne(bv, searchstr, case_sens, whole, forward);
266 return pair<bool, int>(true, 0);
269 // we're now actually ready to replace. if the buffer is
270 // read-only, we can't, though.
271 if (bv->buffer().isReadonly())
272 return pair<bool, int>(false, 0);
274 cap::replaceSelectionWithString(cur, replacestr, forward);
276 cur.pos() += replacestr.length();
277 LASSERT(cur.pos() <= cur.lastpos(), /* */);
279 cur.pos() -= replacestr.length();
280 LASSERT(cur.pos() >= 0, /* */);
283 findOne(bv, searchstr, case_sens, whole, forward, false);
285 return pair<bool, int>(true, 1);
291 docstring const find2string(docstring const & search,
292 bool casesensitive, bool matchword, bool forward)
296 << int(casesensitive) << ' '
297 << int(matchword) << ' '
303 docstring const replace2string(docstring const & replace,
304 docstring const & search, bool casesensitive, bool matchword,
305 bool all, bool forward, bool findnext)
308 ss << replace << '\n'
310 << int(casesensitive) << ' '
311 << int(matchword) << ' '
313 << int(forward) << ' '
319 bool lyxfind(BufferView * bv, FuncRequest const & ev)
321 if (!bv || ev.action() != LFUN_WORD_FIND)
324 //lyxerr << "find called, cmd: " << ev << endl;
326 // data is of the form
328 // <casesensitive> <matchword> <forward>"
330 docstring howto = split(ev.argument(), search, '\n');
332 bool casesensitive = parse_bool(howto);
333 bool matchword = parse_bool(howto);
334 bool forward = parse_bool(howto);
336 return findOne(bv, search, casesensitive, matchword, forward);
340 bool lyxreplace(BufferView * bv,
341 FuncRequest const & ev, bool has_deleted)
343 if (!bv || ev.action() != LFUN_WORD_REPLACE)
346 // data is of the form
349 // <casesensitive> <matchword> <all> <forward> <findnext>"
352 docstring howto = split(ev.argument(), rplc, '\n');
353 howto = split(howto, search, '\n');
355 bool casesensitive = parse_bool(howto);
356 bool matchword = parse_bool(howto);
357 bool all = parse_bool(howto);
358 bool forward = parse_bool(howto);
359 bool findnext = howto.empty() ? true : parse_bool(howto);
361 int replace_count = 0;
366 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
367 update = replace_count > 0;
370 replaceOne(bv, search, rplc, casesensitive, matchword, forward, findnext);
372 replace_count = rv.second;
375 Buffer const & buf = bv->buffer();
377 // emit message signal.
378 buf.message(_("String not found!"));
380 if (replace_count == 0) {
381 buf.message(_("String found."));
382 } else if (replace_count == 1) {
383 buf.message(_("String has been replaced."));
385 docstring const str =
386 bformat(_("%1$d strings have been replaced."), replace_count);
390 } else if (findnext) {
391 // if we have deleted characters, we do not replace at all, but
392 // rather search for the next occurence
393 if (findOne(bv, search, casesensitive, matchword, forward))
396 bv->message(_("String not found!"));
402 bool findNextChange(BufferView * bv)
404 return findChange(bv, true);
408 bool findPreviousChange(BufferView * bv)
410 return findChange(bv, false);
414 bool findChange(BufferView * bv, bool next)
416 if (bv->cursor().selection()) {
417 // set the cursor at the beginning or at the end of the selection
418 // before searching. Otherwise, the current change will be found.
419 if (next != (bv->cursor().top() > bv->cursor().normalAnchor()))
420 bv->cursor().setCursorToAnchor();
423 DocIterator cur = bv->cursor();
425 // Are we within a change ? Then first search forward (backward),
426 // clear the selection and search the other way around (see the end
427 // of this function). This will avoid changes to be selected half.
428 bool search_both_sides = false;
429 DocIterator tmpcur = cur;
431 while (tmpcur.inMathed())
433 Change change_next_pos
434 = tmpcur.paragraph().lookupChange(tmpcur.pos());
435 if (change_next_pos.changed() && cur.inMathed()) {
437 search_both_sides = true;
438 } else if (tmpcur.pos() > 0 && tmpcur.inTexted()) {
439 Change change_prev_pos
440 = tmpcur.paragraph().lookupChange(tmpcur.pos() - 1);
441 if (change_next_pos.isSimilarTo(change_prev_pos))
442 search_both_sides = true;
445 if (!findChange(cur, next))
448 bv->cursor().setCursor(cur);
449 bv->cursor().resetAnchor();
452 // take a step into the change
455 Change orig_change = cur.paragraph().lookupChange(cur.pos());
457 CursorSlice & tip = cur.top();
459 for (; !tip.at_end(); tip.forwardPos()) {
460 Change change = tip.paragraph().lookupChange(tip.pos());
461 if (!change.isSimilarTo(orig_change))
465 for (; !tip.at_begin();) {
467 Change change = tip.paragraph().lookupChange(tip.pos());
468 if (!change.isSimilarTo(orig_change)) {
469 // take a step forward to correctly set the selection
476 // Now put cursor to end of selection:
477 bv->cursor().setCursor(cur);
478 bv->cursor().setSelection();
480 if (search_both_sides) {
481 bv->cursor().setSelection(false);
482 findChange(bv, !next);
490 typedef vector<pair<string, string> > Escapes;
492 /// A map of symbols and their escaped equivalent needed within a regex.
493 /// @note Beware of order
494 Escapes const & get_regexp_escapes()
496 static Escapes escape_map;
497 if (escape_map.empty()) {
498 escape_map.push_back(pair<string, string>("$", "\\$"));
499 escape_map.push_back(pair<string, string>("{", "\\{"));
500 escape_map.push_back(pair<string, string>("}", "\\}"));
501 escape_map.push_back(pair<string, string>("[", "\\["));
502 escape_map.push_back(pair<string, string>("]", "\\]"));
503 escape_map.push_back(pair<string, string>("(", "\\("));
504 escape_map.push_back(pair<string, string>(")", "\\)"));
505 escape_map.push_back(pair<string, string>("+", "\\+"));
506 escape_map.push_back(pair<string, string>("*", "\\*"));
507 escape_map.push_back(pair<string, string>(".", "\\."));
508 escape_map.push_back(pair<string, string>("\\", "(?:\\\\|\\\\backslash)"));
509 escape_map.push_back(pair<string, string>("~", "(?:\\\\textasciitilde|\\\\sim)"));
510 escape_map.push_back(pair<string, string>("^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\mathcircumflex"));
515 /// A map of lyx escaped strings and their unescaped equivalent.
516 Escapes const & get_lyx_unescapes() {
517 static Escapes escape_map;
518 if (escape_map.empty()) {
519 escape_map.push_back(pair<string, string>("\\%", "%"));
520 escape_map.push_back(pair<string, string>("\\{", "{"));
521 escape_map.push_back(pair<string, string>("\\}", "}"));
522 escape_map.push_back(pair<string, string>("\\mathcircumflex ", "^"));
523 escape_map.push_back(pair<string, string>("\\mathcircumflex", "^"));
524 escape_map.push_back(pair<string, string>("\\backslash ", "\\"));
525 escape_map.push_back(pair<string, string>("\\backslash", "\\"));
526 escape_map.push_back(pair<string, string>("\\sim ", "~"));
527 escape_map.push_back(pair<string, string>("\\sim", "~"));
532 /// A map of escapes turning a regexp matching text to one matching latex.
533 Escapes const & get_regexp_latex_escapes() {
534 static Escapes escape_map;
535 if (escape_map.empty()) {
536 escape_map.push_back(pair<string, string>("\\\\", "(?:\\\\\\\\|\\\\backslash|\\\\textbackslash\\{\\})"));
537 escape_map.push_back(pair<string, string>("(<?!\\\\\\\\textbackslash)\\{", "\\\\\\{"));
538 escape_map.push_back(pair<string, string>("(<?!\\\\\\\\textbackslash\\\\\\{)\\}", "\\\\\\}"));
539 escape_map.push_back(pair<string, string>("\\[", "\\{\\[\\}"));
540 escape_map.push_back(pair<string, string>("\\]", "\\{\\]\\}"));
541 escape_map.push_back(pair<string, string>("\\^", "(?:\\^|\\\\textasciicircum\\{\\}|\\\\mathcircumflex)"));
542 escape_map.push_back(pair<string, string>("%", "\\\\\\%"));
547 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
548 ** the found occurrence were escaped.
550 string apply_escapes(string s, Escapes const & escape_map)
552 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
553 Escapes::const_iterator it;
554 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
555 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
556 unsigned int pos = 0;
557 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
558 s.replace(pos, it->first.length(), it->second);
559 LYXERR(Debug::FIND, "After escape: " << s);
560 pos += it->second.length();
561 // LYXERR(Debug::FIND, "pos: " << pos);
564 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
568 /** Return the position of the closing brace matching the open one at s[pos],
569 ** or s.size() if not found.
571 static size_t find_matching_brace(string const & s, size_t pos)
573 LASSERT(s[pos] == '{', /* */);
575 for (++pos; pos < s.size(); ++pos) {
578 else if (s[pos] == '{')
580 else if (s[pos] == '}') {
582 if (open_braces == 0)
589 /// Within \regexp{} apply get_lyx_unescapes() only (i.e., preserve regexp semantics of the string),
590 /// while outside apply get_lyx_unescapes()+get_regexp_escapes().
591 /// If match_latex is true, then apply regexp_latex_escapes() to \regexp{} contents as well.
592 string escape_for_regex(string s, bool match_latex)
595 while (pos < s.size()) {
596 size_t new_pos = s.find("\\regexp{{{", pos);
597 if (new_pos == string::npos)
599 LYXERR(Debug::FIND, "new_pos: " << new_pos);
600 string t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
601 LYXERR(Debug::FIND, "t [lyx]: " << t);
602 t = apply_escapes(t, get_regexp_escapes());
603 LYXERR(Debug::FIND, "t [rxp]: " << t);
604 s.replace(pos, new_pos - pos, t);
605 new_pos = pos + t.size();
606 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
607 LYXERR(Debug::FIND, "new_pos: " << new_pos);
608 if (new_pos == s.size())
610 size_t end_pos = s.find("}}}", new_pos + 10); // find_matching_brace(s, new_pos + 7);
611 LYXERR(Debug::FIND, "end_pos: " << end_pos);
612 t = apply_escapes(s.substr(new_pos + 10, end_pos - (new_pos + 10)), get_lyx_unescapes());
614 t = apply_escapes(t, get_regexp_latex_escapes());
615 LYXERR(Debug::FIND, "t : " << t);
616 if (end_pos == s.size()) {
617 s.replace(new_pos, end_pos - new_pos, t);
619 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
622 s.replace(new_pos, end_pos + 3 - new_pos, t);
623 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
624 pos = new_pos + t.size();
625 LYXERR(Debug::FIND, "pos: " << pos);
630 /// Wrapper for lyx::regex_replace with simpler interface
631 bool regex_replace(string const & s, string & t, string const & searchstr,
632 string const & replacestr)
634 lyx::regex e(searchstr);
636 ostream_iterator<char, char> it(oss);
637 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
638 // tolerate t and s be references to the same variable
639 bool rv = (s != oss.str());
644 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
646 ** Verify that closed braces exactly match open braces. This avoids that, for example,
647 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
650 ** Number of open braces that must remain open at the end for the verification to succeed.
652 bool braces_match(string::const_iterator const & beg,
653 string::const_iterator const & end,
657 string::const_iterator it = beg;
658 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
659 for (; it != end; ++it) {
660 // Skip escaped braces in the count
665 } else if (*it == '{') {
667 } else if (*it == '}') {
668 if (open_pars == 0) {
669 LYXERR(Debug::FIND, "Found unmatched closed brace");
675 if (open_pars != unmatched) {
676 LYXERR(Debug::FIND, "Found " << open_pars
677 << " instead of " << unmatched
678 << " unmatched open braces at the end of count");
681 LYXERR(Debug::FIND, "Braces match as expected");
685 /** The class performing a match between a position in the document and the FindAdvOptions.
687 class MatchStringAdv {
689 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
691 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
692 ** constructor as opt.search, under the opt.* options settings.
695 ** If set, then match is searched only against beginning of text starting at cur.
696 ** If unset, then match is searched anywhere in text starting at cur.
699 ** The length of the matching text, or zero if no match was found.
701 int operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
706 /// first buffer on which search was started
707 lyx::Buffer * const p_first_buf;
709 FindAndReplaceOptions const & opt;
712 /// Auxiliary find method (does not account for opt.matchword)
713 int findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
715 /** Normalize a stringified or latexified LyX paragraph.
719 ** <li>if search is not casesensitive, then lowercase the string;
720 ** <li>remove any newline at begin or end of the string;
721 ** <li>replace any newline in the middle of the string with a simple space;
722 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
725 ** @todo Normalization should also expand macros, if the corresponding
726 ** search option was checked.
728 string normalize(docstring const & s) const;
729 // normalized string to search
730 string par_as_string;
731 // regular expression to use for searching
733 // same as regexp, but prefixed with a ".*"
735 // leading format material as string
736 string lead_as_string;
737 // par_as_string after removal of lead_as_string
738 string par_as_string_nolead;
739 // unmatched open braces in the search string/regexp
741 // number of (.*?) subexpressions added at end of search regexp for closing
742 // environments, math mode, styles, etc...
744 // Are we searching with regular expressions ?
749 static docstring buffer_to_latex(Buffer & buffer)
751 OutputParams runparams(&buffer.params().encoding());
753 odocstringstream ods;
754 otexstream os(ods, texrow);
755 runparams.nice = true;
756 runparams.flavor = OutputParams::LATEX;
757 runparams.linelen = 80; //lyxrc.plaintext_linelen;
758 // No side effect of file copying and image conversion
759 runparams.dryrun = true;
760 pit_type const endpit = buffer.paragraphs().size();
761 for (pit_type pit = 0; pit != endpit; ++pit) {
762 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
763 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
769 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt) {
771 if (!opt.ignoreformat) {
772 str = buffer_to_latex(buffer);
774 OutputParams runparams(&buffer.params().encoding());
775 runparams.nice = true;
776 runparams.flavor = OutputParams::LATEX;
777 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
778 runparams.dryrun = true;
779 for (pos_type pit = pos_type(0); pit < (pos_type)buffer.paragraphs().size(); ++pit) {
780 Paragraph const & par = buffer.paragraphs().at(pit);
781 LYXERR(Debug::FIND, "Adding to search string: '"
782 << par.stringify(pos_type(0), par.size(),
783 AS_STR_INSETS, runparams)
785 str += par.stringify(pos_type(0), par.size(),
786 AS_STR_INSETS, runparams);
793 /// Return separation pos between the leading material and the rest
794 static size_t identifyLeading(string const & s) {
796 // @TODO Support \item[text]
797 while (regex_replace(t, t, "^\\\\(emph|textbf|subsubsection|subsection|section|subparagraph|paragraph|part)\\*?\\{", "")
798 || regex_replace(t, t, "^\\$", "")
799 || regex_replace(t, t, "^\\\\\\[ ", "")
800 || regex_replace(t, t, "^\\\\item ", "")
801 || regex_replace(t, t, "^\\\\begin\\{[a-zA-Z_]*\\*?\\} ", ""))
802 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: '" << t << "'");
807 // Remove trailing closure of math, macros and environments, so to catch parts of them.
808 static int identifyClosing(string & t) {
811 LYXERR(Debug::FIND, "identifyClosing(): t now is '" << t << "'");
812 if (regex_replace(t, t, "(.*[^\\\\])\\$\\'", "$1"))
814 if (regex_replace(t, t, "(.*[^\\\\]) \\\\\\]\\'", "$1"))
816 if (regex_replace(t, t, "(.*[^\\\\]) \\\\end\\{[a-zA-Z_]*\\*?\\}\\'", "$1"))
818 if (regex_replace(t, t, "(.*[^\\\\])\\}\\'", "$1")) {
828 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
829 : p_buf(&buf), p_first_buf(&buf), opt(opt)
831 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
832 par_as_string = normalize(stringifySearchBuffer(find_buf, opt));
836 use_regexp = par_as_string.find("\\regexp") != std::string::npos;
838 size_t lead_size = 0;
839 if (!opt.ignoreformat) {
840 lead_size = identifyLeading(par_as_string);
841 lead_as_string = par_as_string.substr(0, lead_size);
842 par_as_string_nolead = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
846 open_braces = identifyClosing(par_as_string);
847 identifyClosing(par_as_string_nolead);
848 LYXERR(Debug::FIND, "Open braces: " << open_braces);
849 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
851 string lead_as_regexp;
853 // @todo No need to search for \regexp{} insets in leading material
854 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size), !opt.ignoreformat);
855 par_as_string = par_as_string_nolead;
856 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
857 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
859 par_as_string = escape_for_regex(par_as_string, !opt.ignoreformat);
860 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
861 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
863 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
864 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
865 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
866 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])( \\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
867 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
868 || regex_replace(par_as_string, par_as_string,
869 "(.*[^\\\\])( \\\\\\\\end\\\\\\{[a-zA-Z_]*)(\\\\\\*)?(\\\\\\})\\'", "$1(.*?)$2$3$4")
870 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
871 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
875 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
876 LYXERR(Debug::FIND, "Open braces: " << open_braces);
877 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
878 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
879 // If entered regexp must match at begin of searched string buffer
880 string regexp_str = string("\\`") + lead_as_regexp + par_as_string;
881 LYXERR(Debug::FIND, "Setting regexp to : '" << regexp_str << "'");
882 regexp = lyx::regex(regexp_str);
884 // If entered regexp may match wherever in searched string buffer
885 string regexp2_str = string("\\`.*") + lead_as_regexp + ".*" + par_as_string;
886 LYXERR(Debug::FIND, "Setting regexp2 to: '" << regexp2_str << "'");
887 regexp2 = lyx::regex(regexp2_str);
892 int MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
894 docstring docstr = stringifyFromForSearch(opt, cur, len);
895 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
896 string str = normalize(docstr);
897 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
899 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='" << par_as_string << "', str='" << str << "'");
900 LYXERR(Debug::FIND, "Searching in normal mode: lead_as_string='" << lead_as_string << "', par_as_string_nolead='" << par_as_string_nolead << "'");
902 LYXERR(Debug::FIND, "size=" << par_as_string.size() << ", substr='" << str.substr(0, par_as_string.size()) << "'");
903 if (str.substr(0, par_as_string.size()) == par_as_string)
904 return par_as_string.size();
906 size_t pos = str.find(par_as_string_nolead);
907 if (pos != string::npos)
908 return par_as_string.size();
911 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
912 // Try all possible regexp matches,
913 //until one that verifies the braces match test is found
914 regex const *p_regexp = at_begin ? ®exp : ®exp2;
915 sregex_iterator re_it(str.begin(), str.end(), *p_regexp);
916 sregex_iterator re_it_end;
917 for (; re_it != re_it_end; ++re_it) {
918 match_results<string::const_iterator> const & m = *re_it;
919 // Check braces on the segment that matched the entire regexp expression,
920 // plus the last subexpression, if a (.*?) was inserted in the constructor.
921 if (!braces_match(m[0].first, m[0].second, open_braces))
923 // Check braces on segments that matched all (.*?) subexpressions,
924 // except the last "padding" one inserted by lyx.
925 for (size_t i = 1; i < m.size() - 1; ++i)
926 if (!braces_match(m[i].first, m[i].second))
928 // Exclude from the returned match length any length
929 // due to close wildcards added at end of regexp
930 if (close_wildcards == 0)
931 return m[0].second - m[0].first;
933 return m[m.size() - close_wildcards].first - m[0].first;
940 int MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
942 int res = findAux(cur, len, at_begin);
944 "res=" << res << ", at_begin=" << at_begin << ", matchword=" << opt.matchword << ", inTexted=" << cur.inTexted());
945 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
947 Paragraph const & par = cur.paragraph();
948 bool ws_left = cur.pos() > 0 ?
949 par.isWordSeparator(cur.pos() - 1) : true;
950 bool ws_right = cur.pos() + res < par.size() ?
951 par.isWordSeparator(cur.pos() + res) : true;
953 "cur.pos()=" << cur.pos() << ", res=" << res
954 << ", separ: " << ws_left << ", " << ws_right
956 if (ws_left && ws_right)
962 string MatchStringAdv::normalize(docstring const & s) const
965 if (! opt.casesensitive)
966 t = lyx::to_utf8(lowercase(s));
969 // Remove \n at begin
970 while (t.size() > 0 && t[0] == '\n')
973 while (t.size() > 0 && t[t.size() - 1] == '\n')
974 t = t.substr(0, t.size() - 1);
976 // Replace all other \n with spaces
977 while ((pos = t.find("\n")) != string::npos)
978 t.replace(pos, 1, " ");
979 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
980 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
981 while (regex_replace(t, t, "\\\\(emph|textbf|subsubsection|subsection|section|subparagraph|paragraph|part)(\\{\\})+", ""))
982 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
987 docstring stringifyFromCursor(DocIterator const & cur, int len)
989 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
990 if (cur.inTexted()) {
991 Paragraph const & par = cur.paragraph();
992 // TODO what about searching beyond/across paragraph breaks ?
993 // TODO Try adding a AS_STR_INSERTS as last arg
994 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
995 int(par.size()) : cur.pos() + len;
996 OutputParams runparams(&cur.buffer()->params().encoding());
998 runparams.nice = true;
999 runparams.flavor = OutputParams::LATEX;
1000 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
1001 // No side effect of file copying and image conversion
1002 runparams.dryrun = true;
1003 LYXERR(Debug::FIND, "Stringifying with cur: "
1004 << cur << ", from pos: " << cur.pos() << ", end: " << end);
1005 return par.stringify(cur.pos(), end, AS_STR_INSETS, runparams);
1006 } else if (cur.inMathed()) {
1008 CursorSlice cs = cur.top();
1009 MathData md = cs.cell();
1010 MathData::const_iterator it_end =
1011 ( ( len == -1 || cs.pos() + len > int(md.size()) )
1012 ? md.end() : md.begin() + cs.pos() + len );
1013 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
1014 s = s + asString(*it);
1015 LYXERR(Debug::FIND, "Stringified math: '" << s << "'");
1018 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
1023 /** Computes the LaTeX export of buf starting from cur and ending len positions
1024 * after cur, if len is positive, or at the paragraph or innermost inset end
1027 docstring latexifyFromCursor(DocIterator const & cur, int len)
1029 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
1030 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
1031 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
1032 Buffer const & buf = *cur.buffer();
1033 LASSERT(buf.params().isLatex(), /* */);
1036 odocstringstream ods;
1037 otexstream os(ods, texrow);
1038 OutputParams runparams(&buf.params().encoding());
1039 runparams.nice = false;
1040 runparams.flavor = OutputParams::LATEX;
1041 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
1042 // No side effect of file copying and image conversion
1043 runparams.dryrun = true;
1045 if (cur.inTexted()) {
1046 // @TODO what about searching beyond/across paragraph breaks ?
1047 pos_type endpos = cur.paragraph().size();
1048 if (len != -1 && endpos > cur.pos() + len)
1049 endpos = cur.pos() + len;
1050 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
1051 string(), cur.pos(), endpos);
1052 LYXERR(Debug::FIND, "Latexified text: '" << lyx::to_utf8(ods.str()) << "'");
1053 } else if (cur.inMathed()) {
1054 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
1055 for (int s = cur.depth() - 1; s >= 0; --s) {
1056 CursorSlice const & cs = cur[s];
1057 if (cs.asInsetMath() && cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
1058 WriteStream ws(ods);
1059 cs.asInsetMath()->asHullInset()->header_write(ws);
1064 CursorSlice const & cs = cur.top();
1065 MathData md = cs.cell();
1066 MathData::const_iterator it_end = ( ( len == -1 || cs.pos() + len > int(md.size()) )
1067 ? md.end() : md.begin() + cs.pos() + len );
1068 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
1069 ods << asString(*it);
1071 // Retrieve the math environment type, and add '$' or '$]'
1072 // or others (\end{equation}) accordingly
1073 for (int s = cur.depth() - 1; s >= 0; --s) {
1074 CursorSlice const & cs = cur[s];
1075 InsetMath * inset = cs.asInsetMath();
1076 if (inset && inset->asHullInset()) {
1077 WriteStream ws(ods);
1078 inset->asHullInset()->footer_write(ws);
1082 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
1084 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
1090 /** Finalize an advanced find operation, advancing the cursor to the innermost
1091 ** position that matches, plus computing the length of the matching text to
1094 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
1096 // Search the foremost position that matches (avoids find of entire math
1097 // inset when match at start of it)
1099 DocIterator old_cur(cur.buffer());
1101 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
1105 } while (cur && cur.depth() > d && match(cur) > 0);
1107 LASSERT(match(cur) > 0, /* */);
1108 LYXERR(Debug::FIND, "Ok");
1110 // Compute the match length
1112 if (cur.pos() + len > cur.lastpos())
1114 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
1115 while (cur.pos() + len <= cur.lastpos() && match(cur, len) == 0) {
1117 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
1119 // Length of matched text (different from len param)
1120 int old_len = match(cur, len);
1122 // Greedy behaviour while matching regexps
1123 while ((new_len = match(cur, len + 1)) > old_len) {
1126 LYXERR(Debug::FIND, "verifying match with len = " << len);
1133 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
1138 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
1139 int match_len = match(cur, -1, false);
1140 LYXERR(Debug::FIND, "match_len: " << match_len);
1142 for (; cur; cur.forwardPos()) {
1143 LYXERR(Debug::FIND, "Advancing cur: " << cur);
1144 int match_len = match(cur);
1145 LYXERR(Debug::FIND, "match_len: " << match_len);
1147 // Sometimes in finalize we understand it wasn't a match
1148 // and we need to continue the outest loop
1149 int len = findAdvFinalize(cur, match);
1157 if (cur.pit() < cur.lastpit()) {
1158 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
1161 // This should exit nested insets, if any, or otherwise undefine the currsor.
1162 cur.pos() = cur.lastpos();
1163 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
1171 /// Find the most backward consecutive match within same paragraph while searching backwards.
1172 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
1174 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
1175 DocIterator tmp_cur = cur;
1176 int len = findAdvFinalize(tmp_cur, match);
1177 Inset & inset = cur.inset();
1178 for (; cur != cur_begin; cur.backwardPos()) {
1179 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
1180 DocIterator new_cur = cur;
1181 new_cur.backwardPos();
1182 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur))
1184 int new_len = findAdvFinalize(new_cur, match);
1189 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
1195 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match) {
1198 // Backup of original position
1199 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
1200 if (cur == cur_begin)
1203 DocIterator cur_orig(cur);
1205 bool pit_changed = false;
1206 found_match = false;
1209 found_match = match(cur, -1, false);
1213 cur.pos() = cur.lastpos();
1215 cur.pos() = cur_orig.pos();
1216 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
1217 DocIterator cur_prev_iter;
1219 found_match = match(cur);
1220 LYXERR(Debug::FIND, "findBackAdv3: found_match="
1221 << found_match << ", cur: " << cur);
1223 return findMostBackwards(cur, match);
1225 // Stop if begin of document reached
1226 if (cur == cur_begin)
1228 cur_prev_iter = cur;
1232 if (cur == cur_begin)
1244 } // anonym namespace
1247 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
1248 DocIterator const & cur, int len)
1250 LASSERT(cur.pos() >= 0 && cur.pos() <= cur.lastpos(), /* */);
1251 if (!opt.ignoreformat)
1252 return latexifyFromCursor(cur, len);
1254 return stringifyFromCursor(cur, len);
1258 FindAndReplaceOptions::FindAndReplaceOptions(docstring const & find_buf_name, bool casesensitive,
1259 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
1260 docstring const & repl_buf_name, bool keep_case,
1262 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
1263 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
1264 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope)
1272 /** Check if 'len' letters following cursor are all non-lowercase */
1273 static bool allNonLowercase(DocIterator const & cur, int len) {
1274 pos_type end_pos = cur.pos() + len;
1275 for (pos_type pos = cur.pos(); pos != end_pos; ++pos)
1276 if (isLowerCase(cur.paragraph().getChar(pos)))
1282 /** Check if first letter is upper case and second one is lower case */
1283 static bool firstUppercase(DocIterator const & cur) {
1285 if (cur.pos() >= cur.lastpos() - 1) {
1286 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
1289 ch1 = cur.paragraph().getChar(cur.pos());
1290 ch2 = cur.paragraph().getChar(cur.pos()+1);
1291 bool result = isUpperCase(ch1) && isLowerCase(ch2);
1292 LYXERR(Debug::FIND, "firstUppercase(): "
1293 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
1294 << ch2 << "(" << char(ch2) << ")"
1295 << ", result=" << result << ", cur=" << cur);
1300 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
1302 ** \fixme What to do with possible further paragraphs in replace buffer ?
1304 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case) {
1305 ParagraphList::iterator pit = buffer.paragraphs().begin();
1306 pos_type right = pos_type(1);
1307 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
1308 right = pit->size() + 1;
1309 pit->changeCase(buffer.params(), right, right, others_case);
1314 static void findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
1316 Cursor & cur = bv->cursor();
1317 if (opt.repl_buf_name == docstring())
1320 DocIterator sel_beg = cur.selectionBegin();
1321 DocIterator sel_end = cur.selectionEnd();
1322 if (&sel_beg.inset() != &sel_end.inset()
1323 || sel_beg.pit() != sel_end.pit())
1325 int sel_len = sel_end.pos() - sel_beg.pos();
1326 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
1327 << ", sel_len: " << sel_len << endl);
1330 LASSERT(sel_len > 0, /**/);
1332 if (!matchAdv(sel_beg, sel_len))
1335 // Build a copy of the replace buffer, adapted to the KeepCase option
1336 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
1338 repl_buffer_orig.write(oss);
1339 string lyx = oss.str();
1340 Buffer repl_buffer("", false);
1341 repl_buffer.setUnnamed(true);
1342 LASSERT(repl_buffer.readString(lyx), /**/);
1343 if (opt.keep_case && sel_len >= 2) {
1344 if (cur.inTexted()) {
1345 if (firstUppercase(cur))
1346 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
1347 else if (allNonLowercase(cur, sel_len))
1348 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
1351 cap::cutSelection(cur, false, false);
1352 if (!cur.inMathed()) {
1353 repl_buffer.changeLanguage(
1354 repl_buffer.language(),
1355 cur.getFont().language());
1356 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
1357 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
1358 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
1359 repl_buffer.params().documentClassPtr(),
1360 bv->buffer().errorList("Paste"));
1361 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
1362 sel_len = repl_buffer.paragraphs().begin()->size();
1365 odocstringstream ods;
1366 otexstream os(ods, texrow);
1367 OutputParams runparams(&repl_buffer.params().encoding());
1368 runparams.nice = false;
1369 runparams.flavor = OutputParams::LATEX;
1370 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
1371 runparams.dryrun = true;
1372 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
1373 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
1374 docstring repl_latex = ods.str();
1375 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
1377 regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
1378 regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
1379 repl_latex = from_utf8(s);
1380 LYXERR(Debug::FIND, "Replacing by insert()ing latex: '" << repl_latex << "' cur=" << cur << " with depth=" << cur.depth());
1381 MathData ar(cur.buffer());
1382 asArray(repl_latex, ar, Parse::NORMAL);
1384 sel_len = ar.size();
1385 LYXERR(Debug::FIND, "After insert() cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
1387 if (cur.pos() >= sel_len)
1388 cur.pos() -= sel_len;
1391 LYXERR(Debug::FIND, "After pos adj cur=" << cur << " with depth: " << cur.depth() << " and len: " << sel_len);
1392 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
1393 bv->processUpdateFlags(Update::Force);
1394 bv->buffer().updatePreviews();
1398 /// Perform a FindAdv operation.
1399 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
1405 MatchStringAdv matchAdv(bv->buffer(), opt);
1406 findAdvReplace(bv, opt, matchAdv);
1409 match_len = findForwardAdv(cur, matchAdv);
1411 match_len = findBackwardsAdv(cur, matchAdv);
1413 // This may only be raised by lyx::regex()
1414 bv->message(_("Invalid regular expression!"));
1418 if (match_len == 0) {
1419 bv->message(_("Match not found!"));
1423 bv->message(_("Match found!"));
1425 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
1426 bv->putSelectionAt(cur, match_len, !opt.forward);
1432 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
1434 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
1435 << opt.casesensitive << ' '
1436 << opt.matchword << ' '
1437 << opt.forward << ' '
1438 << opt.expandmacros << ' '
1439 << opt.ignoreformat << ' '
1440 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
1441 << opt.keep_case << ' '
1444 LYXERR(Debug::FIND, "built: " << os.str());
1450 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
1452 LYXERR(Debug::FIND, "parsing");
1456 while (line != "EOSS") {
1460 if (is.eof()) // Tolerate malformed request
1464 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
1465 opt.find_buf_name = from_utf8(s);
1466 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat;
1467 is.get(); // Waste space before replace string
1470 while (line != "EOSS") {
1474 if (is.eof()) // Tolerate malformed request
1478 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
1479 opt.repl_buf_name = from_utf8(s);
1480 is >> opt.keep_case;
1483 opt.scope = FindAndReplaceOptions::SearchScope(i);
1484 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
1485 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case);