X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2Flyxfind.cpp;h=c8fc40947e75643cb281589fc41b62b30508858c;hb=3bf1b97ae547aea5d0243e41b2d8af463a6e40c5;hp=3963c9633164c0efe42ff94b8077514cc1d82702;hpb=df82386ce68cd7c94ae388f18a16d30578d1f7d4;p=features.git

diff --git a/src/lyxfind.cpp b/src/lyxfind.cpp
index 3963c96331..c8fc40947e 100644
--- a/src/lyxfind.cpp
+++ b/src/lyxfind.cpp
@@ -14,6 +14,7 @@
  */
 
 #include <config.h>
+#include <iterator>
 
 #include "lyxfind.h"
 
@@ -1110,7 +1111,7 @@ private:
 	 ** @todo Normalization should also expand macros, if the corresponding
 	 ** search option was checked.
 	 **/
-	string normalize(docstring const & s, bool ignore_fomat) const;
+	string convertLF2Space(docstring const & s, bool ignore_fomat) const;
 	// normalized string to search
 	string par_as_string;
 	// regular expression to use for searching
@@ -2423,7 +2424,7 @@ void LatexInfo::buildEntries(bool isPatternString)
 	}
 	// Ignore language if there is math somewhere in pattern-string
 	if (isPatternString) {
-		for (auto s: usedText) {
+		for (auto const & s: usedText) {
 			// Remove entries created in previous search runs
 			keys.erase(s);
 		}
@@ -3668,7 +3669,7 @@ MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
 		previous_single_replace = true;
 	}
 	// When using regexp, braces are hacked already by escape_for_regex()
-	par_as_string = normalize(ds, opt.ignoreformat);
+	par_as_string = convertLF2Space(ds, opt.ignoreformat);
 	open_braces = 0;
 	close_wildcards = 0;
 
@@ -3774,7 +3775,7 @@ MatchStringAdv::MatchStringAdv(lyx::Buffer & buf, FindAndReplaceOptions & opt)
 			}
 			if (opt.matchword) {
 				modifyRegexForMatchWord(par_as_string);
-				opt.matchword = false;
+				// opt.matchword = false;
 			}
 			regexp_str = "(" + lead_as_regexp + ")()" + par_as_string;
 			regexp2_str = "(" + lead_as_regexp + ")(.*?)" + par_as_string;
@@ -3793,7 +3794,7 @@ MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, MatchStrin
 
 	docstring docstr = stringifyFromForSearch(opt, cur, len);
 	string str;
-	str = normalize(docstr, opt.ignoreformat);
+	str = convertLF2Space(docstr, opt.ignoreformat);
 	if (!opt.ignoreformat) {
 		str = correctlanguagesetting(str, false, !opt.ignoreformat);
 		// remove closing '}' and '\n' to allow for use of '$' in regex
@@ -3949,20 +3950,17 @@ MatchResult MatchStringAdv::findAux(DocIterator const & cur, int len, MatchStrin
 MatchResult MatchStringAdv::operator()(DocIterator const & cur, int len, MatchStringAdv::matchType at_begin) const
 {
 	MatchResult mres = findAux(cur, len, at_begin);
-	int res = mres.match_len;
 	LYXERR(Debug::FINDVERBOSE,
-	       "res=" << res << ", at_begin=" << matchTypeAsString(at_begin)
+	       "res=" << mres.match_len << ", at_begin=" << matchTypeAsString(at_begin)
 	       << ", matchAtStart=" << opt.matchAtStart
 	       << ", inTexted=" << cur.inTexted());
-	if (opt.matchAtStart) {
-		if (cur.pos() != 0)
-			mres.match_len = 0;
-		else if (mres.match_prefix > 0)
-			mres.match_len = 0;
-		return mres;
+	if (mres.match_len > 0) {
+		if (opt.matchAtStart) {
+			if (cur.pos() > 0 || mres.match_prefix > 0)
+				mres.match_len = 0;
+		}
 	}
-	else
-		return mres;
+	return mres;
 }
 
 #if 0
@@ -3990,8 +3988,7 @@ static bool simple_replace(string &t, string from, string to)
 }
 #endif
 
-#if 1
-static string convertLF2Space(docstring const &s, bool ignore_format)
+string MatchStringAdv::convertLF2Space(docstring const &s, bool ignore_format) const
 {
 	// Using original docstring to handle '\n'
 
@@ -4058,77 +4055,6 @@ static string convertLF2Space(docstring const &s, bool ignore_format)
 	return(t.str());
 }
 
-#else
-static string convertLF2Space(docstring const & s, bool ignore_format)
-{
-	// Using utf8-converted string to handle '\n'
-
-	string t;
-	t = lyx::to_utf8(s);
-	// Remove \n at begin
-	while (!t.empty() && t[0] == '\n')
-		t = t.substr(1);
-	// Remove \n* at end
-	while (!t.empty() && t[t.size() - 1] == '\n') {
-		t = t.substr(0, t.size() - 1);
-	}
-	size_t pos;
-	// Handle all other '\n'
-	while ((pos = t.find("\n")) != string::npos) {
-		if (pos > 1 && t[pos-1] == '\\' && t[pos-2] == '\\' ) {
-			// Handle '\\\n'
-			if (isPrintableNonspace(t[pos+1]) && ((pos < 3) || isPrintableNonspace(t[pos-3]))) {
-				t.replace(pos-2, 3, " ");
-			}
-			else {
-				// Already a space there
-				t.replace(pos-2, 3, "");
-			}
-		}
-		else {
-			if (!isAlnumASCII(t[pos+1]) || !isAlnumASCII(t[pos-1])) {
-				// '\n' adjacent to non-alpha-numerics, discard
-				t.replace(pos, 1, "");
-			}
-			else {
-				// Replace all other \n with spaces
-				t.replace(pos, 1, " ");
-			}
-			if (!ignore_format) {
-				size_t count = 0;
-				while ((pos > count + 1) && (t[pos - 1 -count] == '%')) {
-					count++;
-				}
-				if (count > 0) {
-					t.replace(pos - count, count, "");
-				}
-			}
-		}
-	}
-	return(t);
-
-}
-#endif
-
-string MatchStringAdv::normalize(docstring const & s, bool ignore_format) const
-{
-	string t = convertLF2Space(s, ignore_format);
-
-	// The following replaces are not appropriate in non-format-search mode
-	if (!ignore_format) {
-		// Remove stale empty \emph{}, \textbf{} and similar blocks from latexify
-		// Kornel: Added textsl, textsf, textit, texttt and noun
-		// + allow to seach for colored text too
-		LYXERR(Debug::FINDVERBOSE, "Removing stale empty macros from: " << t);
-		while (regex_replace(t, t, "\\\\(emph|noun|text(bf|sl|sf|it|tt)|(u|uu)line|(s|x)out|uwave)(\\{(\\{\\})?\\})+", ""))
-			LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\emph{}, \\textbf{} macros from: " << t);
-		while (regex_replace(t, t, "\\\\((sub)?(((sub)?section)|paragraph)|part)\\*?(\\{(\\{\\})?\\})+", ""))
-			LYXERR(Debug::FINDVERBOSE, "  further removing stale empty \\section{}, \\part{}, \\paragraph{} macros from: " << t);
-		while (regex_replace(t, t, "\\\\(foreignlanguage|textcolor|item)\\{[a-z]+\\}(\\{(\\{\\})?\\})+", ""));
-	}
-	return t;
-}
-
 docstring stringifyFromCursor(DocIterator const & cur, int len)
 {
 	LYXERR(Debug::FINDVERBOSE, "Stringifying with len=" << len << " from cursor at pos: " << cur);
@@ -4513,7 +4439,7 @@ int findForwardAdv(DocIterator & cur, MatchStringAdv & match)
 					default:
 						// Todo@
 						// Handle not like MatchResult::newIsTooFar
-						LYXERR0( "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
+						LYXERR(Debug::FINDVERBOSE, "Probably too far: Increment = " << increment << " match_prefix = " << mres.match_prefix);
 						firstInvalid--;
 						increment = increment*3/4;
 						cur = old_cur;
@@ -4566,7 +4492,8 @@ MatchResult findMostBackwards(DocIterator & cur, MatchStringAdv const & match, M
 		LYXERR(Debug::FINDVERBOSE, "findMostBackwards(): cur=" << cur);
 		DocIterator new_cur = cur;
 		new_cur.backwardPos();
-		if (new_cur == cur || &new_cur.inset() != &inset || !match(new_cur, -1, MatchStringAdv::MatchFromStart).match_len)
+		if (new_cur == cur || &new_cur.inset() != &inset
+		    || match(new_cur, -1, MatchStringAdv::MatchFromStart).match_len <= 0)
 			break;
 		MatchResult new_mr = findAdvFinalize(new_cur, match, expected);
 		if (new_mr.match_len == mr.match_len)
@@ -4764,7 +4691,7 @@ static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, Ma
 		return 0;
 	LASSERT(sel_len > 0, return 0);
 
-	if (!matchAdv(sel_beg, sel_len, MatchStringAdv::MatchFromStart).match_len)
+	if (matchAdv(sel_beg, sel_len, MatchStringAdv::MatchFromStart).match_len <= 0)
 		return 0;
 
 	// Build a copy of the replace buffer, adapted to the KeepCase option
@@ -4832,6 +4759,10 @@ static int findAdvReplace(BufferView * bv, FindAndReplaceOptions const & opt, Ma
 	return 1;
 }
 
+static bool isWordChar(char_type c)
+{
+	return isLetterChar(c) || isNumberChar(c);
+}
 
 /// Perform a FindAdv operation.
 bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
@@ -4857,8 +4788,52 @@ bool findAdv(BufferView * bv, FindAndReplaceOptions & opt)
 			bv->putSelectionAt(bv->cursor().selectionBegin(), length, !opt.forward);
 		num_replaced += findAdvReplace(bv, opt, matchAdv);
 		cur = bv->cursor();
-		if (opt.forward)
+		if (opt.forward) {
+			if (opt.matchword && cur.pos() > 0) {  // Skip word-characters if we are in the mid of a word
+				if (cur.inTexted()) {
+					Paragraph const & par = cur.paragraph();
+					int len_limit, new_pos;
+					if (cur.lastpos() < par.size())
+						len_limit = cur.lastpos();
+					else
+						len_limit = par.size();
+					for (new_pos = cur.pos() - 1; new_pos < len_limit; new_pos++) {
+						if (!isWordChar(par.getChar(new_pos)))
+							break;
+					}
+					if (new_pos > cur.pos())
+						cur.pos() = new_pos;
+				}
+				else if (cur.inMathed()) {
+					// Check if 'cur.pos()-1' and 'cur.pos()' both point to a letter,
+					// I am not sure, we should consider the selection
+					bool sel = bv->cursor().selection();
+					if (!sel && cur.pos() < cur.lastpos()) {
+						CursorSlice const & cs = cur.top();
+						MathData md = cs.cell();
+						int len = -1;
+						MathData::const_iterator it_end = md.end();
+						MathData md2;
+						// Start the check with one character before actual cursor position
+						for (MathData::const_iterator it = md.begin() + cs.pos() - 1;
+						    it != it_end; ++it)
+							md2.push_back(*it);
+						docstring inp = asString(md2);
+						for (len = 0; (unsigned) len < inp.size() && len + cur.pos() <= cur.lastpos(); len++) {
+							if (!isWordChar(inp[len]))
+								break;
+						}
+						// len == 0 means previous char was a word separator
+						// len == 1       search starts with a word separator
+						// len == 2 ...   we have to skip len -1 chars
+						if (len > 1)
+							cur.pos() = cur.pos() + len - 1;
+					}
+				}
+				opt.matchword = false;
+			}
 			pos_len = findForwardAdv(cur, matchAdv);
+		}
 		else
 			pos_len = findBackwardsAdv(cur, matchAdv);
 	} catch (exception & ex) {