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"
44 #include "support/convert.h"
45 #include "support/debug.h"
46 #include "support/docstream.h"
47 #include "support/gettext.h"
48 #include "support/lassert.h"
49 #include "support/lstrings.h"
51 #include "support/regex.h"
52 #include <boost/next_prior.hpp>
55 using namespace lyx::support;
61 bool parse_bool(docstring & howto)
66 howto = split(howto, var, ' ');
71 class MatchString : public binary_function<Paragraph, pos_type, int>
74 MatchString(docstring const & str, bool cs, bool mw)
75 : str(str), case_sens(cs), whole_words(mw)
78 // returns true if the specified string is at the specified position
79 // del specifies whether deleted strings in ct mode will be considered
80 int operator()(Paragraph const & par, pos_type pos, bool del = true) const
82 return par.find(str, case_sens, whole_words, pos, del);
90 // match whole words only
95 int findForward(DocIterator & cur, MatchString const & match,
98 for (; cur; cur.forwardChar())
100 int len = match(cur.paragraph(), cur.pos(), find_del);
108 int findBackwards(DocIterator & cur, MatchString const & match,
109 bool find_del = true)
113 if (cur.inTexted()) {
114 int len = match(cur.paragraph(), cur.pos(), find_del);
123 bool findChange(DocIterator & cur, bool next)
127 for (; cur; next ? cur.forwardPos() : cur.backwardPos())
128 if (cur.inTexted() && cur.paragraph().isChanged(cur.pos())) {
130 // if we search backwards, take a step forward
131 // to correctly set the anchor
140 bool searchAllowed(docstring const & str)
143 frontend::Alert::error(_("Search error"), _("Search string is empty"));
150 bool findOne(BufferView * bv, docstring const & searchstr,
151 bool case_sens, bool whole, bool forward, bool find_del = true)
153 if (!searchAllowed(searchstr))
156 DocIterator cur = bv->cursor();
158 MatchString const match(searchstr, case_sens, whole);
160 int match_len = forward ? findForward(cur, match, find_del) :
161 findBackwards(cur, match, find_del);
164 bv->putSelectionAt(cur, match_len, !forward);
166 return match_len > 0;
170 int replaceAll(BufferView * bv,
171 docstring const & searchstr, docstring const & replacestr,
172 bool case_sens, bool whole)
174 Buffer & buf = bv->buffer();
176 if (!searchAllowed(searchstr) || buf.isReadonly())
179 DocIterator cur_orig(bv->cursor());
181 MatchString const match(searchstr, case_sens, whole);
184 int const rsize = replacestr.size();
185 int const ssize = searchstr.size();
188 cur.setCursor(doc_iterator_begin(&buf));
189 int match_len = findForward(cur, match, false);
190 while (match_len > 0) {
191 // Backup current cursor position and font.
192 pos_type const pos = cur.pos();
193 Font const font = cur.paragraph().getFontSettings(buf.params(), pos);
195 int striked = ssize - cur.paragraph().eraseChars(pos, pos + match_len,
196 buf.params().trackChanges);
197 cur.paragraph().insert(pos, replacestr, font,
198 Change(buf.params().trackChanges ?
199 Change::INSERTED : Change::UNCHANGED));
200 for (int i = 0; i < rsize + striked; ++i)
203 match_len = findForward(cur, match, false);
206 bv->putSelectionAt(doc_iterator_begin(&buf), 0, false);
208 cur_orig.fixIfBroken();
209 bv->setCursor(cur_orig);
215 // the idea here is that we are going to replace the string that
216 // is selected IF it is the search string.
217 // if there is a selection, but it is not the search string, then
218 // we basically ignore it. (FIXME We ought to replace only within
220 // if there is no selection, then:
221 // (i) if some search string has been provided, then we find it.
222 // (think of how the dialog works when you hit "replace" the
224 // (ii) if no search string has been provided, then we treat the
225 // word the cursor is in as the search string. (why? i have no
226 // idea.) but this only works in text?
228 // returns the number of replacements made (one, if any) and
229 // whether anything at all was done.
230 pair<bool, int> replaceOne(BufferView * bv, docstring searchstr,
231 docstring const & replacestr, bool case_sens,
232 bool whole, bool forward)
234 Cursor & cur = bv->cursor();
235 if (!cur.selection()) {
236 // no selection, non-empty search string: find it
237 if (!searchstr.empty()) {
238 findOne(bv, searchstr, case_sens, whole, forward);
239 return pair<bool, int>(true, 0);
241 // empty search string
244 return pair<bool, int>(false, 0);
245 // select current word and treat it as the search string
246 cur.innerText()->selectWord(cur, WHOLE_WORD);
247 searchstr = cur.selectionAsString(false);
250 // if we still don't have a search string, report the error
252 if (!searchAllowed(searchstr))
253 return pair<bool, int>(false, 0);
255 bool have_selection = cur.selection();
256 docstring const selected = cur.selectionAsString(false);
258 case_sens ? searchstr == selected
259 : compare_no_case(searchstr, selected) == 0;
261 // no selection or current selection is not search word:
262 // just find the search word
263 if (!have_selection || !match) {
264 findOne(bv, searchstr, case_sens, whole, forward);
265 return pair<bool, int>(true, 0);
268 // we're now actually ready to replace. if the buffer is
269 // read-only, we can't, though.
270 if (bv->buffer().isReadonly())
271 return pair<bool, int>(false, 0);
273 cap::replaceSelectionWithString(cur, replacestr, forward);
275 cur.pos() += replacestr.length();
276 LASSERT(cur.pos() <= cur.lastpos(), /* */);
278 cur.pos() -= replacestr.length();
279 LASSERT(cur.pos() >= 0, /* */);
281 findOne(bv, searchstr, case_sens, whole, forward, false);
283 return pair<bool, int>(true, 1);
289 docstring const find2string(docstring const & search,
290 bool casesensitive, bool matchword, bool forward)
294 << int(casesensitive) << ' '
295 << int(matchword) << ' '
301 docstring const replace2string(docstring const & replace,
302 docstring const & search, bool casesensitive, bool matchword,
303 bool all, bool forward)
306 ss << replace << '\n'
308 << int(casesensitive) << ' '
309 << int(matchword) << ' '
316 bool lyxfind(BufferView * bv, FuncRequest const & ev)
318 if (!bv || ev.action() != LFUN_WORD_FIND)
321 //lyxerr << "find called, cmd: " << ev << endl;
323 // data is of the form
325 // <casesensitive> <matchword> <forward>"
327 docstring howto = split(ev.argument(), search, '\n');
329 bool casesensitive = parse_bool(howto);
330 bool matchword = parse_bool(howto);
331 bool forward = parse_bool(howto);
333 return findOne(bv, search, casesensitive, matchword, forward);
337 bool lyxreplace(BufferView * bv,
338 FuncRequest const & ev, bool has_deleted)
340 if (!bv || ev.action() != LFUN_WORD_REPLACE)
343 // data is of the form
346 // <casesensitive> <matchword> <all> <forward>"
349 docstring howto = split(ev.argument(), rplc, '\n');
350 howto = split(howto, search, '\n');
352 bool casesensitive = parse_bool(howto);
353 bool matchword = parse_bool(howto);
354 bool all = parse_bool(howto);
355 bool forward = parse_bool(howto);
357 int replace_count = 0;
362 replace_count = replaceAll(bv, search, rplc, casesensitive, matchword);
363 update = replace_count > 0;
366 replaceOne(bv, search, rplc, casesensitive, matchword, forward);
368 replace_count = rv.second;
371 Buffer const & buf = bv->buffer();
373 // emit message signal.
374 buf.message(_("String not found!"));
376 if (replace_count == 0) {
377 buf.message(_("String found."));
378 } else if (replace_count == 1) {
379 buf.message(_("String has been replaced."));
381 docstring const str =
382 bformat(_("%1$d strings have been replaced."), replace_count);
387 // if we have deleted characters, we do not replace at all, but
388 // rather search for the next occurence
389 if (findOne(bv, search, casesensitive, matchword, forward))
392 bv->message(_("String not found!"));
398 bool findNextChange(BufferView * bv)
400 return findChange(bv, true);
404 bool findPreviousChange(BufferView * bv)
406 return findChange(bv, false);
410 bool findChange(BufferView * bv, bool next)
412 if (bv->cursor().selection()) {
413 // set the cursor at the beginning or at the end of the selection
414 // before searching. Otherwise, the current change will be found.
415 if (next != (bv->cursor().top() > bv->cursor().normalAnchor()))
416 bv->cursor().setCursorToAnchor();
419 DocIterator cur = bv->cursor();
421 // Are we within a change ? Then first search forward (backward),
422 // clear the selection and search the other way around (see the end
423 // of this function). This will avoid changes to be selected half.
424 bool search_both_sides = false;
425 DocIterator tmpcur = cur;
427 while (tmpcur.inMathed())
429 Change change_next_pos
430 = tmpcur.paragraph().lookupChange(tmpcur.pos());
431 if (change_next_pos.changed() && cur.inMathed()) {
433 search_both_sides = true;
434 } else if (tmpcur.pos() > 0 && tmpcur.inTexted()) {
435 Change change_prev_pos
436 = tmpcur.paragraph().lookupChange(tmpcur.pos() - 1);
437 if (change_next_pos.isSimilarTo(change_prev_pos))
438 search_both_sides = true;
441 if (!findChange(cur, next))
444 bv->cursor().setCursor(cur);
445 bv->cursor().resetAnchor();
448 // take a step into the change
451 Change orig_change = cur.paragraph().lookupChange(cur.pos());
453 CursorSlice & tip = cur.top();
455 for (; !tip.at_end(); tip.forwardPos()) {
456 Change change = tip.paragraph().lookupChange(tip.pos());
457 if (!change.isSimilarTo(orig_change))
461 for (; !tip.at_begin();) {
463 Change change = tip.paragraph().lookupChange(tip.pos());
464 if (!change.isSimilarTo(orig_change)) {
465 // take a step forward to correctly set the selection
472 // Now put cursor to end of selection:
473 bv->cursor().setCursor(cur);
474 bv->cursor().setSelection();
476 if (search_both_sides) {
477 bv->cursor().setSelection(false);
478 findChange(bv, !next);
486 typedef vector<pair<string, string> > Escapes;
488 /// A map of symbols and their escaped equivalent needed within a regex.
489 Escapes const & get_regexp_escapes()
491 static Escapes escape_map;
492 if (escape_map.empty()) {
493 escape_map.push_back(pair<string, string>("\\", "\\\\"));
494 escape_map.push_back(pair<string, string>("^", "\\^"));
495 escape_map.push_back(pair<string, string>("$", "\\$"));
496 escape_map.push_back(pair<string, string>("{", "\\{"));
497 escape_map.push_back(pair<string, string>("}", "\\}"));
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>(".", "\\."));
509 /// A map of lyx escaped strings and their unescaped equivalent.
510 Escapes const & get_lyx_unescapes() {
511 static Escapes escape_map;
512 if (escape_map.empty()) {
513 escape_map.push_back(pair<string, string>("{*}", "*"));
514 escape_map.push_back(pair<string, string>("{[}", "["));
515 escape_map.push_back(pair<string, string>("\\$", "$"));
516 escape_map.push_back(pair<string, string>("\\backslash{}", "\\"));
517 escape_map.push_back(pair<string, string>("\\backslash", "\\"));
518 escape_map.push_back(pair<string, string>("\\sim ", "~"));
519 escape_map.push_back(pair<string, string>("\\^", "^"));
524 /** @todo Probably the maps need to be migrated to regexps, in order to distinguish if
525 ** the found occurrence were escaped.
527 string apply_escapes(string s, Escapes const & escape_map)
529 LYXERR(Debug::FIND, "Escaping: '" << s << "'");
530 Escapes::const_iterator it;
531 for (it = escape_map.begin(); it != escape_map.end(); ++it) {
532 // LYXERR(Debug::FIND, "Escaping " << it->first << " as " << it->second);
533 unsigned int pos = 0;
534 while (pos < s.length() && (pos = s.find(it->first, pos)) < s.length()) {
535 s.replace(pos, it->first.length(), it->second);
536 // LYXERR(Debug::FIND, "After escape: " << s);
537 pos += it->second.length();
538 // LYXERR(Debug::FIND, "pos: " << pos);
541 LYXERR(Debug::FIND, "Escaped : '" << s << "'");
545 /** Return the position of the closing brace matching the open one at s[pos],
546 ** or s.size() if not found.
548 size_t find_matching_brace(string const & s, size_t pos)
550 LASSERT(s[pos] == '{', /* */);
552 for (++pos; pos < s.size(); ++pos) {
555 else if (s[pos] == '{')
557 else if (s[pos] == '}') {
559 if (open_braces == 0)
566 /// Within \regexp{} apply get_regex_escapes(), while outside apply get_lyx_unescapes().
567 string escape_for_regex(string s)
570 while (pos < s.size()) {
571 size_t new_pos = s.find("\\regexp{{{", pos);
572 if (new_pos == string::npos)
574 LYXERR(Debug::FIND, "new_pos: " << new_pos);
575 string t = apply_escapes(s.substr(pos, new_pos - pos), get_lyx_unescapes());
576 LYXERR(Debug::FIND, "t : " << t);
577 t = apply_escapes(t, get_regexp_escapes());
578 LYXERR(Debug::FIND, "t : " << t);
579 s.replace(pos, new_pos - pos, t);
580 new_pos = pos + t.size();
581 LYXERR(Debug::FIND, "Regexp after escaping: " << s);
582 LYXERR(Debug::FIND, "new_pos: " << new_pos);
583 if (new_pos == s.size())
585 size_t end_pos = s.find("}}}", new_pos + 10); // find_matching_brace(s, new_pos + 7);
586 LYXERR(Debug::FIND, "end_pos: " << end_pos);
587 t = apply_escapes(s.substr(new_pos + 10, end_pos - (new_pos + 10)), get_lyx_unescapes());
588 LYXERR(Debug::FIND, "t : " << t);
589 if (end_pos == s.size()) {
590 s.replace(new_pos, end_pos - new_pos, t);
592 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
595 s.replace(new_pos, end_pos + 3 - new_pos, t);
596 LYXERR(Debug::FIND, "Regexp after \\regexp{} removal: " << s);
597 pos = new_pos + t.size();
598 LYXERR(Debug::FIND, "pos: " << pos);
603 /// Wrapper for lyx::regex_replace with simpler interface
604 bool regex_replace(string const & s, string & t, string const & searchstr,
605 string const & replacestr)
607 lyx::regex e(searchstr);
609 ostream_iterator<char, char> it(oss);
610 lyx::regex_replace(it, s.begin(), s.end(), e, replacestr);
611 // tolerate t and s be references to the same variable
612 bool rv = (s != oss.str());
617 /** Checks if supplied string segment is well-formed from the standpoint of matching open-closed braces.
619 ** Verify that closed braces exactly match open braces. This avoids that, for example,
620 ** \frac{.*}{x} matches \frac{x+\frac{y}{x}}{z} with .* being 'x+\frac{y'.
623 ** Number of open braces that must remain open at the end for the verification to succeed.
625 bool braces_match(string::const_iterator const & beg,
626 string::const_iterator const & end,
630 string::const_iterator it = beg;
631 LYXERR(Debug::FIND, "Checking " << unmatched << " unmatched braces in '" << string(beg, end) << "'");
632 for (; it != end; ++it) {
633 // Skip escaped braces in the count
638 } else if (*it == '{') {
640 } else if (*it == '}') {
641 if (open_pars == 0) {
642 LYXERR(Debug::FIND, "Found unmatched closed brace");
648 if (open_pars != unmatched) {
649 LYXERR(Debug::FIND, "Found " << open_pars
650 << " instead of " << unmatched
651 << " unmatched open braces at the end of count");
654 LYXERR(Debug::FIND, "Braces match as expected");
658 /** The class performing a match between a position in the document and the FindAdvOptions.
660 class MatchStringAdv {
662 MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt);
664 /** Tests if text starting at the supplied position matches with the one provided to the MatchStringAdv
665 ** constructor as opt.search, under the opt.* options settings.
668 ** If set, then match is searched only against beginning of text starting at cur.
669 ** If unset, then match is searched anywhere in text starting at cur.
672 ** The length of the matching text, or zero if no match was found.
674 int operator()(DocIterator const & cur, int len = -1, bool at_begin = true) const;
679 /// first buffer on which search was started
680 lyx::Buffer * const p_first_buf;
682 FindAndReplaceOptions const & opt;
685 /// Auxiliary find method (does not account for opt.matchword)
686 int findAux(DocIterator const & cur, int len = -1, bool at_begin = true) const;
688 /** Normalize a stringified or latexified LyX paragraph.
692 ** <li>if search is not casesensitive, then lowercase the string;
693 ** <li>remove any newline at begin or end of the string;
694 ** <li>replace any newline in the middle of the string with a simple space;
695 ** <li>remove stale empty styles and environments, like \emph{} and \textbf{}.
698 ** @todo Normalization should also expand macros, if the corresponding
699 ** search option was checked.
701 string normalize(docstring const & s) const;
702 // normalized string to search
703 string par_as_string;
704 // regular expression to use for searching
706 // same as regexp, but prefixed with a ".*"
708 // unmatched open braces in the search string/regexp
710 // number of (.*?) subexpressions added at end of search regexp for closing
711 // environments, math mode, styles, etc...
713 // Are we searching with regular expressions ?
718 static docstring buffer_to_latex(Buffer & buffer)
720 OutputParams runparams(&buffer.params().encoding());
722 odocstringstream ods;
723 otexstream os(ods, texrow);
724 runparams.nice = true;
725 runparams.flavor = OutputParams::LATEX;
726 runparams.linelen = 80; //lyxrc.plaintext_linelen;
727 // No side effect of file copying and image conversion
728 runparams.dryrun = true;
729 pit_type const endpit = buffer.paragraphs().size();
730 for (pit_type pit = 0; pit != endpit; ++pit) {
731 TeXOnePar(buffer, buffer.text(), pit, os, runparams);
732 LYXERR(Debug::FIND, "searchString up to here: " << ods.str());
738 static docstring stringifySearchBuffer(Buffer & buffer, FindAndReplaceOptions const & opt) {
740 if (!opt.ignoreformat) {
741 str = buffer_to_latex(buffer);
743 ParIterator it = buffer.par_iterator_begin();
744 ParIterator end = buffer.par_iterator_end();
745 OutputParams runparams(&buffer.params().encoding());
747 runparams.nice = true;
748 runparams.flavor = OutputParams::LATEX;
749 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
750 runparams.dryrun = true;
751 for (; it != end; ++it) {
752 LYXERR(Debug::FIND, "Adding to search string: '"
753 << it->asString(false)
756 it->stringify(pos_type(0), it->size(),
757 AS_STR_INSETS, runparams);
764 /// Return separation pos between the leading material and the rest
765 static size_t identifyLeading(string const & s) {
767 // @TODO Support \item[text]
768 while (regex_replace(t, t, "\\\\(emph|textbf|subsubsection|subsection|section|subparagraph|paragraph|part)\\{", "")
769 || regex_replace(t, t, "^\\$", "")
770 || regex_replace(t, t, "^\\\\\\[ ", "")
771 || regex_replace(t, t, "^\\\\item ", ""))
772 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: " << t);
777 MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions const & opt)
778 : p_buf(&buf), p_first_buf(&buf), opt(opt)
780 Buffer & find_buf = *theBufferList().getBuffer(FileName(to_utf8(opt.find_buf_name)), true);
781 par_as_string = normalize(stringifySearchBuffer(find_buf, opt));
785 use_regexp = !opt.ignoreformat || par_as_string.find("\\regexp") != std::string::npos;
788 // Remove trailing closure of math, macros and environments, so to catch parts of them.
790 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
791 if (regex_replace(par_as_string, par_as_string, "(.*)[[:blank:]]\\'", "$1"))
793 if (regex_replace(par_as_string, par_as_string, "(.*[^\\\\]) ?\\$\\'", "$1"))
795 // @todo need to account for open square braces as well ?
796 if (regex_replace(par_as_string, par_as_string, "(.*[^\\\\]) ?\\\\\\]\\'", "$1"))
798 if (regex_replace(par_as_string, par_as_string, "(.*[^\\\\]) ?\\\\end\\{[a-zA-Z_]*\\}\\'", "$1"))
800 if (regex_replace(par_as_string, par_as_string, "(.*[^\\\\]) ?\\}\\'", "$1")) {
806 LYXERR(Debug::FIND, "Open braces: " << open_braces);
807 LYXERR(Debug::FIND, "Built MatchStringAdv object: par_as_string = '" << par_as_string << "'");
809 size_t lead_size = identifyLeading(par_as_string);
810 string lead_as_regexp;
812 lead_as_regexp = escape_for_regex(par_as_string.substr(0, lead_size));
813 par_as_string = par_as_string.substr(lead_size, par_as_string.size() - lead_size);
814 LYXERR(Debug::FIND, "lead_as_regexp is '" << lead_as_regexp << "'");
815 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
817 par_as_string = escape_for_regex(par_as_string);
818 // Insert (.*?) before trailing closure of math, macros and environments, so to catch parts of them.
819 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
821 // Insert .* before trailing '\$' ('$' has been escaped by escape_for_regex)
822 regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\$)\\'", "$1(.*?)$2")
823 // Insert .* before trailing '\\\]' ('\]' has been escaped by escape_for_regex)
824 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\\\\\\\\\])\\'", "$1(.*?)$2")
825 // Insert .* before trailing '\\end\{...}' ('\end{...}' has been escaped by escape_for_regex)
826 || regex_replace(par_as_string, par_as_string,
827 "(.*[^\\\\])(\\\\\\\\end\\\\\\{[a-zA-Z_]*\\\\\\})\\'", "$1(.*?)$2")
828 // Insert .* before trailing '\}' ('}' has been escaped by escape_for_regex)
829 || regex_replace(par_as_string, par_as_string, "(.*[^\\\\])(\\\\\\})\\'", "$1(.*?)$2")
833 LYXERR(Debug::FIND, "par_as_string now is '" << par_as_string << "'");
834 LYXERR(Debug::FIND, "Open braces: " << open_braces);
835 LYXERR(Debug::FIND, "Close .*? : " << close_wildcards);
836 LYXERR(Debug::FIND, "Replaced text (to be used as regex): " << par_as_string);
837 // If entered regexp must match at begin of searched string buffer
838 string regexp_str = string("\\`") + lead_as_regexp + par_as_string;
839 LYXERR(Debug::FIND, "Setting regexp to : " << regexp_str << endl);
840 regexp = lyx::regex(regexp_str);
842 // If entered regexp may match wherever in searched string buffer
843 string regexp2_str = string("\\`.*") + lead_as_regexp + ".*" + par_as_string;
844 LYXERR(Debug::FIND, "Setting regexp2 to: " << regexp2_str << endl);
845 regexp2 = lyx::regex(regexp2_str);
850 int MatchStringAdv::findAux(DocIterator const & cur, int len, bool at_begin) const
852 docstring docstr = stringifyFromForSearch(opt, cur, len);
853 LYXERR(Debug::FIND, "Matching against '" << lyx::to_utf8(docstr) << "'");
854 string str = normalize(docstr);
855 LYXERR(Debug::FIND, "After normalization: '" << str << "'");
857 LYXERR(Debug::FIND, "Searching in normal mode: par_as_string='" << par_as_string << "', str='" << str << "'");
859 LYXERR(Debug::FIND, "size=" << par_as_string.size() << ", substr='" << str.substr(0, par_as_string.size()) << "'");
860 if (str.substr(0, par_as_string.size()) == par_as_string)
861 return par_as_string.size();
863 string t = par_as_string;
864 while (regex_replace(t, t, "\\\\(emph|textbf|subsubsection|subsection|section|subparagraph|paragraph|part)\\{", "")
865 || regex_replace(t, t, "^\\$", "")
866 || regex_replace(t, t, "^\\\\\\[ ", ""))
867 LYXERR(Debug::FIND, " after removing leading $, \\[ , \\emph{, \\textbf{, etc.: " << t);
868 size_t pos = str.find(t);
869 if (pos != string::npos)
870 return par_as_string.size();
873 LYXERR(Debug::FIND, "Searching in regexp mode: at_begin=" << at_begin);
874 // Try all possible regexp matches,
875 //until one that verifies the braces match test is found
876 regex const *p_regexp = at_begin ? ®exp : ®exp2;
877 sregex_iterator re_it(str.begin(), str.end(), *p_regexp);
878 sregex_iterator re_it_end;
879 for (; re_it != re_it_end; ++re_it) {
880 match_results<string::const_iterator> const & m = *re_it;
881 // Check braces on the segment that matched the entire regexp expression,
882 // plus the last subexpression, if a (.*?) was inserted in the constructor.
883 if (!braces_match(m[0].first, m[0].second, open_braces))
885 // Check braces on segments that matched all (.*?) subexpressions.
886 for (size_t i = 1; i < m.size(); ++i)
887 if (!braces_match(m[i].first, m[i].second))
889 // Exclude from the returned match length any length
890 // due to close wildcards added at end of regexp
891 if (close_wildcards == 0)
892 return m[0].second - m[0].first;
894 return m[m.size() - close_wildcards].first - m[0].first;
901 int MatchStringAdv::operator()(DocIterator const & cur, int len, bool at_begin) const
903 int res = findAux(cur, len, at_begin);
905 "res=" << res << ", at_begin=" << at_begin << ", matchword=" << opt.matchword << ", inTexted=" << cur.inTexted());
906 if (res == 0 || !at_begin || !opt.matchword || !cur.inTexted())
908 Paragraph const & par = cur.paragraph();
909 bool ws_left = cur.pos() > 0 ?
910 par.isWordSeparator(cur.pos() - 1) : true;
911 bool ws_right = cur.pos() + res < par.size() ?
912 par.isWordSeparator(cur.pos() + res) : true;
914 "cur.pos()=" << cur.pos() << ", res=" << res
915 << ", separ: " << ws_left << ", " << ws_right
917 if (ws_left && ws_right)
923 string MatchStringAdv::normalize(docstring const & s) const
926 if (! opt.casesensitive)
927 t = lyx::to_utf8(lowercase(s));
930 // Remove \n at begin
931 while (t.size() > 0 && t[0] == '\n')
934 while (t.size() > 0 && t[t.size() - 1] == '\n')
935 t = t.substr(0, t.size() - 1);
937 // Replace all other \n with spaces
938 while ((pos = t.find("\n")) != string::npos)
939 t.replace(pos, 1, " ");
940 // Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
941 LYXERR(Debug::FIND, "Removing stale empty \\emph{}, \\textbf{}, \\*section{} macros from: " << t);
942 while (regex_replace(t, t, "\\\\(emph|textbf|subsubsection|subsection|section|subparagraph|paragraph|part)(\\{\\})+", ""))
943 LYXERR(Debug::FIND, " further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
948 docstring stringifyFromCursor(DocIterator const & cur, int len)
950 LYXERR(Debug::FIND, "Stringifying with len=" << len << " from cursor at pos: " << cur);
951 if (cur.inTexted()) {
952 Paragraph const & par = cur.paragraph();
953 // TODO what about searching beyond/across paragraph breaks ?
954 // TODO Try adding a AS_STR_INSERTS as last arg
955 pos_type end = ( len == -1 || cur.pos() + len > int(par.size()) ) ?
956 int(par.size()) : cur.pos() + len;
957 OutputParams runparams(&cur.buffer()->params().encoding());
959 runparams.nice = true;
960 runparams.flavor = OutputParams::LATEX;
961 runparams.linelen = 100000; //lyxrc.plaintext_linelen;
962 // No side effect of file copying and image conversion
963 runparams.dryrun = true;
964 LYXERR(Debug::FIND, "Stringifying with cur: "
965 << cur << ", from pos: " << cur.pos() << ", end: " << end);
966 return par.stringify(cur.pos(), end, AS_STR_INSETS, runparams);
967 } else if (cur.inMathed()) {
969 CursorSlice cs = cur.top();
970 MathData md = cs.cell();
971 MathData::const_iterator it_end =
972 ( ( len == -1 || cs.pos() + len > int(md.size()) )
973 ? md.end() : md.begin() + cs.pos() + len );
974 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
978 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
983 /** Computes the LaTeX export of buf starting from cur and ending len positions
984 * after cur, if len is positive, or at the paragraph or innermost inset end
987 docstring latexifyFromCursor(DocIterator const & cur, int len)
989 LYXERR(Debug::FIND, "Latexifying with len=" << len << " from cursor at pos: " << cur);
990 LYXERR(Debug::FIND, " with cur.lastpost=" << cur.lastpos() << ", cur.lastrow="
991 << cur.lastrow() << ", cur.lastcol=" << cur.lastcol());
992 Buffer const & buf = *cur.buffer();
993 LASSERT(buf.isLatex(), /* */);
996 odocstringstream ods;
997 otexstream os(ods, texrow);
998 OutputParams runparams(&buf.params().encoding());
999 runparams.nice = false;
1000 runparams.flavor = OutputParams::LATEX;
1001 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
1002 // No side effect of file copying and image conversion
1003 runparams.dryrun = true;
1005 if (cur.inTexted()) {
1006 // @TODO what about searching beyond/across paragraph breaks ?
1007 pos_type endpos = cur.paragraph().size();
1008 if (len != -1 && endpos > cur.pos() + len)
1009 endpos = cur.pos() + len;
1010 TeXOnePar(buf, *cur.innerText(), cur.pit(), os, runparams,
1011 string(), cur.pos(), endpos);
1012 LYXERR(Debug::FIND, "Latexified text: '" << lyx::to_utf8(ods.str()) << "'");
1013 } else if (cur.inMathed()) {
1014 // Retrieve the math environment type, and add '$' or '$[' or others (\begin{equation}) accordingly
1015 for (int s = cur.depth() - 1; s >= 0; --s) {
1016 CursorSlice const & cs = cur[s];
1017 if (cs.asInsetMath() && cs.asInsetMath() && cs.asInsetMath()->asHullInset()) {
1018 WriteStream ws(ods);
1019 cs.asInsetMath()->asHullInset()->header_write(ws);
1024 CursorSlice const & cs = cur.top();
1025 MathData md = cs.cell();
1026 MathData::const_iterator it_end = ( ( len == -1 || cs.pos() + len > int(md.size()) )
1027 ? md.end() : md.begin() + cs.pos() + len );
1028 for (MathData::const_iterator it = md.begin() + cs.pos(); it != it_end; ++it)
1031 // Retrieve the math environment type, and add '$' or '$]'
1032 // or others (\end{equation}) accordingly
1033 for (int s = cur.depth() - 1; s >= 0; --s) {
1034 CursorSlice const & cs = cur[s];
1035 InsetMath * inset = cs.asInsetMath();
1036 if (inset && inset->asHullInset()) {
1037 WriteStream ws(ods);
1038 inset->asHullInset()->footer_write(ws);
1042 LYXERR(Debug::FIND, "Latexified math: '" << lyx::to_utf8(ods.str()) << "'");
1044 LYXERR(Debug::FIND, "Don't know how to stringify from here: " << cur);
1050 /** Finalize an advanced find operation, advancing the cursor to the innermost
1051 ** position that matches, plus computing the length of the matching text to
1054 int findAdvFinalize(DocIterator & cur, MatchStringAdv const & match)
1056 // Search the foremost position that matches (avoids find of entire math
1057 // inset when match at start of it)
1059 DocIterator old_cur(cur.buffer());
1061 LYXERR(Debug::FIND, "Forwarding one step (searching for innermost match)");
1065 } while (cur && cur.depth() > d && match(cur) > 0);
1067 LASSERT(match(cur) > 0, /* */);
1068 LYXERR(Debug::FIND, "Ok");
1070 // Compute the match length
1072 if (cur.pos() + len > cur.lastpos())
1074 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
1075 while (cur.pos() + len <= cur.lastpos() && match(cur, len) == 0) {
1077 LYXERR(Debug::FIND, "verifying unmatch with len = " << len);
1079 // Length of matched text (different from len param)
1080 int old_len = match(cur, len);
1082 // Greedy behaviour while matching regexps
1083 while ((new_len = match(cur, len + 1)) > old_len) {
1086 LYXERR(Debug::FIND, "verifying match with len = " << len);
1093 int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
1098 LYXERR(Debug::FIND, "findForwardAdv() cur: " << cur);
1099 if (match(cur, -1, false)) {
1100 for (; cur; cur.forwardPos()) {
1101 LYXERR(Debug::FIND, "Advancing cur: " << cur);
1103 // Sometimes in finalize we understand it wasn't a match
1104 // and we need to continue the outest loop
1105 int len = findAdvFinalize(cur, match);
1113 if (cur.pit() < cur.lastpit()) {
1114 LYXERR(Debug::FIND, "Advancing par: cur=" << cur);
1117 // This should exit nested insets, if any, or otherwise undefine the currsor.
1118 cur.pos() = cur.lastpos();
1119 LYXERR(Debug::FIND, "Advancing pos: cur=" << cur);
1127 /// Find the most backward consecutive match within same paragraph while searching backwards.
1128 int findMostBackwards(DocIterator & cur, MatchStringAdv const & match)
1130 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
1131 DocIterator tmp_cur = cur;
1132 int len = findAdvFinalize(tmp_cur, match);
1133 Inset & inset = cur.inset();
1134 for (; cur != cur_begin; cur.backwardPos()) {
1135 LYXERR(Debug::FIND, "findMostBackwards(): cur=" << cur);
1136 DocIterator new_cur = cur;
1137 new_cur.backwardPos();
1138 if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur))
1140 int new_len = findAdvFinalize(new_cur, match);
1145 LYXERR(Debug::FIND, "findMostBackwards(): exiting with cur=" << cur);
1151 int findBackwardsAdv(DocIterator & cur, MatchStringAdv & match) {
1154 // Backup of original position
1155 DocIterator cur_begin = doc_iterator_begin(cur.buffer());
1156 if (cur == cur_begin)
1159 DocIterator cur_orig(cur);
1161 bool pit_changed = false;
1162 found_match = false;
1165 found_match = match(cur, -1, false);
1169 cur.pos() = cur.lastpos();
1171 cur.pos() = cur_orig.pos();
1172 LYXERR(Debug::FIND, "findBackAdv2: cur: " << cur);
1173 DocIterator cur_prev_iter;
1175 found_match = match(cur);
1176 LYXERR(Debug::FIND, "findBackAdv3: found_match="
1177 << found_match << ", cur: " << cur);
1179 return findMostBackwards(cur, match);
1181 // Stop if begin of document reached
1182 if (cur == cur_begin)
1184 cur_prev_iter = cur;
1188 if (cur == cur_begin)
1200 } // anonym namespace
1203 docstring stringifyFromForSearch(FindAndReplaceOptions const & opt,
1204 DocIterator const & cur, int len)
1206 LASSERT(cur.pos() >= 0 && cur.pos() <= cur.lastpos(), /* */);
1207 if (!opt.ignoreformat)
1208 return latexifyFromCursor(cur, len);
1210 return stringifyFromCursor(cur, len);
1214 FindAndReplaceOptions::FindAndReplaceOptions(docstring const & find_buf_name, bool casesensitive,
1215 bool matchword, bool forward, bool expandmacros, bool ignoreformat,
1216 docstring const & repl_buf_name, bool keep_case,
1218 : find_buf_name(find_buf_name), casesensitive(casesensitive), matchword(matchword),
1219 forward(forward), expandmacros(expandmacros), ignoreformat(ignoreformat),
1220 repl_buf_name(repl_buf_name), keep_case(keep_case), scope(scope)
1228 /** Check if 'len' letters following cursor are all non-lowercase */
1229 static bool allNonLowercase(DocIterator const & cur, int len) {
1230 pos_type end_pos = cur.pos() + len;
1231 for (pos_type pos = cur.pos(); pos != end_pos; ++pos)
1232 if (isLowerCase(cur.paragraph().getChar(pos)))
1238 /** Check if first letter is upper case and second one is lower case */
1239 static bool firstUppercase(DocIterator const & cur) {
1241 if (cur.pos() >= cur.lastpos() - 1) {
1242 LYXERR(Debug::FIND, "No upper-case at cur: " << cur);
1245 ch1 = cur.paragraph().getChar(cur.pos());
1246 ch2 = cur.paragraph().getChar(cur.pos()+1);
1247 bool result = isUpperCase(ch1) && isLowerCase(ch2);
1248 LYXERR(Debug::FIND, "firstUppercase(): "
1249 << "ch1=" << ch1 << "(" << char(ch1) << "), ch2="
1250 << ch2 << "(" << char(ch2) << ")"
1251 << ", result=" << result << ", cur=" << cur);
1256 /** Make first letter of supplied buffer upper-case, and the rest lower-case.
1258 ** \fixme What to do with possible further paragraphs in replace buffer ?
1260 static void changeFirstCase(Buffer & buffer, TextCase first_case, TextCase others_case) {
1261 ParagraphList::iterator pit = buffer.paragraphs().begin();
1262 pos_type right = pos_type(1);
1263 pit->changeCase(buffer.params(), pos_type(0), right, first_case);
1264 right = pit->size() + 1;
1265 pit->changeCase(buffer.params(), right, right, others_case);
1270 static void findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, MatchStringAdv & matchAdv)
1272 Cursor & cur = bv->cursor();
1273 if (opt.repl_buf_name == docstring())
1276 DocIterator sel_beg = cur.selectionBegin();
1277 DocIterator sel_end = cur.selectionEnd();
1278 if (&sel_beg.inset() != &sel_end.inset()
1279 || sel_beg.pit() != sel_end.pit())
1281 int sel_len = sel_end.pos() - sel_beg.pos();
1282 LYXERR(Debug::FIND, "sel_beg: " << sel_beg << ", sel_end: " << sel_end
1283 << ", sel_len: " << sel_len << endl);
1286 LASSERT(sel_len > 0, /**/);
1288 if (!matchAdv(sel_beg, sel_len))
1291 // Build a copy of the replace buffer, adapted to the KeepCase option
1292 Buffer & repl_buffer_orig = *theBufferList().getBuffer(FileName(to_utf8(opt.repl_buf_name)), true);
1294 repl_buffer_orig.write(oss);
1295 string lyx = oss.str();
1296 Buffer repl_buffer("", false);
1297 repl_buffer.setUnnamed(true);
1298 LASSERT(repl_buffer.readString(lyx), /**/);
1299 if (opt.keep_case && sel_len >= 2) {
1300 if (cur.inTexted()) {
1301 if (firstUppercase(cur))
1302 changeFirstCase(repl_buffer, text_uppercase, text_lowercase);
1303 else if (allNonLowercase(cur, sel_len))
1304 changeFirstCase(repl_buffer, text_uppercase, text_uppercase);
1307 cap::cutSelection(cur, false, false);
1308 if (!cur.inMathed()) {
1309 repl_buffer.changeLanguage(
1310 repl_buffer.language(),
1311 cur.getFont().language());
1312 LYXERR(Debug::FIND, "Replacing by pasteParagraphList()ing repl_buffer");
1313 LYXERR(Debug::FIND, "Before pasteParagraphList() cur=" << cur << endl);
1314 cap::pasteParagraphList(cur, repl_buffer.paragraphs(),
1315 repl_buffer.params().documentClassPtr(),
1316 bv->buffer().errorList("Paste"));
1317 LYXERR(Debug::FIND, "After pasteParagraphList() cur=" << cur << endl);
1318 sel_len = repl_buffer.paragraphs().begin()->size();
1321 odocstringstream ods;
1322 otexstream os(ods, texrow);
1323 OutputParams runparams(&repl_buffer.params().encoding());
1324 runparams.nice = false;
1325 runparams.flavor = OutputParams::LATEX;
1326 runparams.linelen = 8000; //lyxrc.plaintext_linelen;
1327 runparams.dryrun = true;
1328 TeXOnePar(repl_buffer, repl_buffer.text(), 0, os, runparams);
1329 //repl_buffer.getSourceCode(ods, 0, repl_buffer.paragraphs().size(), false);
1330 docstring repl_latex = ods.str();
1331 LYXERR(Debug::FIND, "Latexified replace_buffer: '" << repl_latex << "'");
1333 regex_replace(to_utf8(repl_latex), s, "\\$(.*)\\$", "$1");
1334 regex_replace(s, s, "\\\\\\[(.*)\\\\\\]", "$1");
1335 repl_latex = from_utf8(s);
1336 LYXERR(Debug::FIND, "Replacing by niceInsert()ing latex: '" << repl_latex << "'");
1337 sel_len = cur.niceInsert(repl_latex);
1339 cur.pos() -= sel_len;
1342 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << sel_len);
1343 bv->putSelectionAt(DocIterator(cur), sel_len, !opt.forward);
1344 bv->processUpdateFlags(Update::Force);
1348 /// Perform a FindAdv operation.
1349 bool findAdv(BufferView * bv, FindAndReplaceOptions const & opt)
1355 MatchStringAdv matchAdv(bv->buffer(), opt);
1356 findAdvReplace(bv, opt, matchAdv);
1359 match_len = findForwardAdv(cur, matchAdv);
1361 match_len = findBackwardsAdv(cur, matchAdv);
1363 // This may only be raised by lyx::regex()
1364 bv->message(_("Invalid regular expression!"));
1368 if (match_len == 0) {
1369 bv->message(_("Match not found!"));
1373 bv->message(_("Match found!"));
1375 LYXERR(Debug::FIND, "Putting selection at cur=" << cur << " with len: " << match_len);
1376 bv->putSelectionAt(cur, match_len, !opt.forward);
1382 ostringstream & operator<<(ostringstream & os, FindAndReplaceOptions const & opt)
1384 os << to_utf8(opt.find_buf_name) << "\nEOSS\n"
1385 << opt.casesensitive << ' '
1386 << opt.matchword << ' '
1387 << opt.forward << ' '
1388 << opt.expandmacros << ' '
1389 << opt.ignoreformat << ' '
1390 << to_utf8(opt.repl_buf_name) << "\nEOSS\n"
1391 << opt.keep_case << ' '
1394 LYXERR(Debug::FIND, "built: " << os.str());
1400 istringstream & operator>>(istringstream & is, FindAndReplaceOptions & opt)
1402 LYXERR(Debug::FIND, "parsing");
1406 while (line != "EOSS") {
1410 if (is.eof()) // Tolerate malformed request
1414 LYXERR(Debug::FIND, "file_buf_name: '" << s << "'");
1415 opt.find_buf_name = from_utf8(s);
1416 is >> opt.casesensitive >> opt.matchword >> opt.forward >> opt.expandmacros >> opt.ignoreformat;
1417 is.get(); // Waste space before replace string
1420 while (line != "EOSS") {
1424 if (is.eof()) // Tolerate malformed request
1428 LYXERR(Debug::FIND, "repl_buf_name: '" << s << "'");
1429 opt.repl_buf_name = from_utf8(s);
1430 is >> opt.keep_case;
1433 opt.scope = FindAndReplaceOptions::SearchScope(i);
1434 LYXERR(Debug::FIND, "parsed: " << opt.casesensitive << ' ' << opt.matchword << ' ' << opt.forward << ' '
1435 << opt.expandmacros << ' ' << opt.ignoreformat << ' ' << opt.keep_case);