Use boost::wregex implementation and manually utf8 decode into it

That way we get proper unicode support in regular expressions as long
as the current locale treats wchar_t as unicode codepoints.

Fixes #638
Fixes #595
Fixes #162

src/file.cc

@@ -394,8 +394,9 @@ CandidateList complete_filename(StringView prefix, const Regex& ignored_regex,
 
     auto filter = [&ignored_regex, check_ignored_regex, include_hidden, only_dir](const dirent& entry, struct stat& st)
     {
-        return (include_hidden or StringView{entry.d_name}.substr(0_byte, 1_byte) != ".") and
-               (not check_ignored_regex or not regex_match(entry.d_name, ignored_regex)) and
+        StringView name{entry.d_name};
+        return (include_hidden or name.substr(0_byte, 1_byte) != ".") and
+               (not check_ignored_regex or not regex_match(name.begin(), name.end(), ignored_regex)) and
                (not only_dir or S_ISDIR(st.st_mode));
     };
     auto files = list_files(dirname, filter);

src/highlighters.cc

@@ -308,13 +308,14 @@ private:
                       buffer.iterator_at(range.end), m_regex,
                       match_flags(is_bol(range.begin),
                                   is_eol(buffer, range.end),
+                                  is_bow(buffer, range.begin),
                                   is_eow(buffer, range.end))};
         RegexIt re_end;
         for (; re_it != re_end; ++re_it)
         {
             for (size_t i = 0; i < m_faces.size(); ++i)
             {
-                auto& sub = (*re_it)[m_faces[i].first];
+                const auto& sub = (*re_it)[m_faces[i].first];
                 matches.push_back({sub.first.coord(), sub.second.coord()});
             }
         }

src/regex.cc

@@ -5,6 +5,12 @@
 namespace Kakoune
 {
 
+using Utf8It = RegexUtf8It<const char*>;
+
+Regex::Regex(StringView re, flag_type flags) try
+    : boost::wregex{Utf8It{re.begin(), re}, Utf8It{re.end(), re}}, m_str(re.str())
+{} catch (std::runtime_error& err) { throw regex_error(err.what()); }
+
 String option_to_string(const Regex& re)
 {
     return re.str();

src/regex.hh

@@ -3,12 +3,9 @@
 
 #include "string.hh"
 #include "exception.hh"
+#include "utf8_iterator.hh"
 
-#ifdef KAK_USE_STDREGEX
-#include <regex>
-#else
 #include <boost/regex.hpp>
-#endif
 
 namespace Kakoune
 {
@@ -20,16 +17,12 @@ struct regex_error : runtime_error
     {}
 };
 
-#ifdef KAK_USE_STDREGEX
 // Regex that keeps track of its string representation
-struct Regex : std::regex
+struct Regex : boost::wregex
 {
     Regex() = default;
 
-    explicit Regex(StringView re, flag_type flags = ECMAScript) try
-        : std::regex(re.begin(), re.end(), flags), m_str(re.str()) {}
-        catch (std::runtime_error& err) { throw regex_error(err.what()); }
-
+    explicit Regex(StringView re, flag_type flags = ECMAScript);
     bool empty() const { return m_str.empty(); }
     bool operator==(const Regex& other) const { return m_str == other.m_str; }
     bool operator!=(const Regex& other) const { return m_str != other.m_str; }
@@ -39,37 +32,97 @@ struct Regex : std::regex
 private:
     String m_str;
 };
-namespace regex_ns = std;
-#else
-struct Regex : boost::regex
+
+template<typename It>
+using RegexUtf8It = utf8::iterator<It, wchar_t, ssize_t>;
+
+namespace RegexConstant = boost::regex_constants;
+
+template<typename Iterator>
+struct MatchResults : boost::match_results<RegexUtf8It<Iterator>>
 {
-    Regex() = default;
+    using ParentType = boost::match_results<RegexUtf8It<Iterator>>;
+    struct SubMatch : std::pair<Iterator, Iterator>
+    {
+        SubMatch() = default;
+        SubMatch(const boost::sub_match<RegexUtf8It<Iterator>>& m)
+            : std::pair<Iterator, Iterator>{m.first.base(), m.second.base()},
+              matched{m.matched}
+        {}
 
-    explicit Regex(StringView re, flag_type flags = ECMAScript) try
-        : boost::regex(re.begin(), re.end(), flags) {}
-        catch (std::runtime_error& err) { throw regex_error(err.what()); }
+        bool matched = false;
+    };
 
-    String str() const { auto s = boost::regex::str(); return {s.data(), (int)s.length()}; }
+    struct iterator : boost::match_results<RegexUtf8It<Iterator>>::iterator
+    {
+        using ParentType = typename boost::match_results<RegexUtf8It<Iterator>>::iterator;
+        iterator(const ParentType& it) : ParentType(it) {}
+
+        SubMatch operator*() const { return {ParentType::operator*()}; }
+    };
+
+    iterator begin() const { return {ParentType::begin()}; }
+    iterator cbegin() const { return {ParentType::cbegin()}; }
+    iterator end() const { return {ParentType::end()}; }
+    iterator cend() const { return {ParentType::cend()}; }
+
+    SubMatch operator[](size_t s) const { return {ParentType::operator[](s)}; }
 };
-namespace regex_ns = boost;
-#endif
 
 template<typename Iterator>
-using RegexIterator = regex_ns::regex_iterator<Iterator>;
-
-template<typename Iterator>
-using MatchResults = regex_ns::match_results<Iterator>;
-
-namespace RegexConstant = regex_ns::regex_constants;
-
-inline RegexConstant::match_flag_type match_flags(bool bol, bool eol, bool eow)
+struct RegexIterator : boost::regex_iterator<RegexUtf8It<Iterator>>
 {
-    return (bol ? RegexConstant::match_default : RegexConstant::match_not_bol |
-                                                 RegexConstant::match_prev_avail) |
+    using ParentType = boost::regex_iterator<RegexUtf8It<Iterator>>;
+    using Utf8It = RegexUtf8It<Iterator>;
+    using ValueType = MatchResults<Iterator>;
+
+    RegexIterator() = default;
+    RegexIterator(Iterator begin, Iterator end, const Regex& re,
+                  RegexConstant::match_flag_type flags = RegexConstant::match_default)
+        : ParentType{Utf8It{begin, begin, end}, Utf8It{end, begin, end}, re, flags} {}
+
+    const ValueType& operator*() const { return *reinterpret_cast<const ValueType*>(&ParentType::operator*()); }
+    const ValueType* operator->() const { return reinterpret_cast<const ValueType*>(ParentType::operator->()); }
+};
+
+inline RegexConstant::match_flag_type match_flags(bool bol, bool eol, bool bow, bool eow)
+{
+    return (bol ? RegexConstant::match_default : RegexConstant::match_not_bol) |
            (eol ? RegexConstant::match_default : RegexConstant::match_not_eol) |
+           (bow ? RegexConstant::match_default : RegexConstant::match_not_bow) |
            (eow ? RegexConstant::match_default : RegexConstant::match_not_eow);
 }
 
+template<typename It>
+bool regex_match(It begin, It end, const Regex& re)
+{
+    using Utf8It = RegexUtf8It<It>;
+    return boost::regex_match(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, re);
+}
+
+template<typename It>
+bool regex_match(It begin, It end, MatchResults<It>& res, const Regex& re)
+{
+    using Utf8It = RegexUtf8It<It>;
+    return boost::regex_match(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, res, re);
+}
+
+template<typename It>
+bool regex_search(It begin, It end, const Regex& re,
+                  RegexConstant::match_flag_type flags = RegexConstant::match_default)
+{
+    using Utf8It = RegexUtf8It<It>;
+    return boost::regex_search(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, re);
+}
+
+template<typename It>
+bool regex_search(It begin, It end, MatchResults<It>& res, const Regex& re,
+                  RegexConstant::match_flag_type flags = RegexConstant::match_default)
+{
+    using Utf8It = RegexUtf8It<It>;
+    return boost::regex_search(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, res, re);
+}
+
 String option_to_string(const Regex& re);
 void option_from_string(StringView str, Regex& re);
 

src/selectors.cc

@@ -615,20 +615,21 @@ void select_all_matches(SelectionList& selections, const Regex& regex, unsigned
         auto sel_end = utf8::next(buffer.iterator_at(sel.max()), buffer.end());
         const auto flags = match_flags(is_bol(sel_beg.coord()),
                                        is_eol(buffer, sel_end.coord()),
+                                       is_bow(buffer, sel_beg.coord()),
                                        is_eow(buffer, sel_end.coord()));
         RegexIt re_it(sel_beg, sel_end, regex, flags);
         RegexIt re_end;
 
         for (; re_it != re_end; ++re_it)
         {
-            auto begin = ensure_char_start(buffer, (*re_it)[capture].first);
+            auto begin = (*re_it)[capture].first;
             if (begin == sel_end)
                 continue;
-            auto end = ensure_char_start(buffer, (*re_it)[capture].second);
+            auto end = (*re_it)[capture].second;
 
             CaptureList captures;
             captures.reserve(mark_count);
-            for (auto& match : *re_it)
+            for (const auto& match : *re_it)
                 captures.push_back(buffer.string(match.first.coord(),
                                                  match.second.coord()));
 
@@ -661,6 +662,7 @@ void split_selections(SelectionList& selections, const Regex& regex, unsigned ca
         auto sel_end = utf8::next(buffer.iterator_at(sel.max()), buffer.end());
         const auto flags = match_flags(is_bol(begin.coord()),
                                        is_eol(buffer, sel_end.coord()),
+                                       is_bow(buffer, begin.coord()),
                                        is_eow(buffer, sel_end.coord()));
 
         RegexIt re_it(begin, sel_end, regex, flags);
@@ -674,11 +676,10 @@ void split_selections(SelectionList& selections, const Regex& regex, unsigned ca
 
             if (end != buf_begin)
             {
-                end = ensure_char_start(buffer, end);
                 auto sel_end = (begin == end) ? end : utf8::previous(end, begin);
                 result.push_back(keep_direction({ begin.coord(), sel_end.coord() }, sel));
             }
-            begin = ensure_char_start(buffer, (*re_it)[capture].second);
+            begin = (*re_it)[capture].second;
         }
         if (begin.coord() <= sel.max())
             result.push_back(keep_direction({ begin.coord(), sel.max() }, sel));

src/selectors.hh

@@ -226,7 +226,8 @@ inline bool find_last_match(const Buffer& buffer, const BufferIterator& pos,
     const bool is_pos_eow = is_eow(buffer, pos.coord());
     auto begin = buffer.begin();
     while (begin != pos and regex_search(begin, pos, matches, regex,
-                                         match_flags(is_bol(begin.coord()), is_pos_eol, is_pos_eow)))
+                                         match_flags(is_bol(begin.coord()), is_pos_eol,
+                                                     is_bow(buffer, begin.coord()), is_pos_eow)))
     {
         begin = utf8::next(matches[0].first, pos);
         if (res.empty() or matches[0].second > res[0].second)
@@ -244,7 +245,8 @@ bool find_match_in_buffer(const Buffer& buffer, const BufferIterator pos,
     if (direction == Forward)
     {
         if (regex_search(pos, buffer.end(), matches, ex,
-                         match_flags(is_bol(pos.coord()), true, true)))
+                         match_flags(is_bol(pos.coord()), true,
+                                     is_bow(buffer, pos.coord()), true)))
             return true;
         wrapped = true;
         return regex_search(buffer.begin(), buffer.end(), matches, ex);
@@ -276,9 +278,9 @@ Selection find_next_match(const Buffer& buffer, const Selection& sel, const Rege
     auto pos = direction == Forward ? utf8::next(begin, buffer.end()) : begin;
     if ((found = find_match_in_buffer<direction>(buffer, pos, matches, regex, wrapped)))
     {
-        begin = ensure_char_start(buffer, matches[0].first);
-        end = ensure_char_start(buffer, matches[0].second);
-        for (auto& match : matches)
+        begin = matches[0].first;
+        end = matches[0].second;
+        for (const auto& match : matches)
             captures.push_back(buffer.string(match.first.coord(),
                                              match.second.coord()));
     }

test/regression/638-highlight-codepoint-with-bracket/cmd

@@ -0,0 +1 @@
+w<c-l>:q<ret>

test/regression/638-highlight-codepoint-with-bracket/display

@@ -0,0 +1,5 @@
+{ "jsonrpc": "2.0", "method": "draw", "params": [[[{ "face": { "fg": "red", "bg": "default", "attributes": [] }, "contents": "“" }, { "face": { "fg": "white", "bg": "blue", "attributes": [] }, "contents": "We" }, { "face": { "fg": "black", "bg": "white", "attributes": [] }, "contents": " " }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "ought to scrape this planet clean of every living thing on it," }, { "face": { "fg": "red", "bg": "default", "attributes": [] }, "contents": "”" }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "\u000a" }]], { "fg": "default", "bg": "default", "attributes": [] }, { "fg": "blue", "bg": "default", "attributes": [] }] }
+{ "jsonrpc": "2.0", "method": "menu_hide", "params": [] }
+{ "jsonrpc": "2.0", "method": "info_hide", "params": [] }
+{ "jsonrpc": "2.0", "method": "draw_status", "params": [[], [{ "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "out 1:4 " }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": " " }, { "face": { "fg": "blue", "bg": "default", "attributes": [] }, "contents": "1 sel" }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": " - unnamed0@[kak-test-regression-638-highlight-codepoint-with-bracket]" }], { "fg": "cyan", "bg": "default", "attributes": [] }] }
+{ "jsonrpc": "2.0", "method": "refresh", "params": [true] }

test/regression/638-highlight-codepoint-with-bracket/in

@@ -0,0 +1 @@
+“We ought to scrape this planet clean of every living thing on it,”

test/regression/638-highlight-codepoint-with-bracket/out

@@ -0,0 +1 @@
+“We ought to scrape this planet clean of every living thing on it,”

test/regression/638-highlight-codepoint-with-bracket/patch

@@ -0,0 +1,11 @@
+--- a/display	2016-05-10 09:21:59.272300947 +0100
++++ b/display	2016-05-10 09:37:27.866341923 +0100
+@@ -1,5 +1,5 @@
+-{ "jsonrpc": "2.0", "method": "draw", "params": [[[{ "face": { "fg": "black", "bg": "white", "attributes": [] }, "contents": "“" }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "We ought to scrape this planet clean of every living thing on it," }, { "face": { "fg": "red", "bg": "default", "attributes": [] }, "contents": "”" }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "\u000a" }]], { "fg": "default", "bg": "default", "attributes": [] }, { "fg": "blue", "bg": "default", "attributes": [] }] }
++{ "jsonrpc": "2.0", "method": "draw", "params": [[[{ "face": { "fg": "red", "bg": "default", "attributes": [] }, "contents": "“" }, { "face": { "fg": "white", "bg": "blue", "attributes": [] }, "contents": "We" }, { "face": { "fg": "black", "bg": "white", "attributes": [] }, "contents": " " }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "ought to scrape this planet clean of every living thing on it," }, { "face": { "fg": "red", "bg": "default", "attributes": [] }, "contents": "”" }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "\u000a" }]], { "fg": "default", "bg": "default", "attributes": [] }, { "fg": "blue", "bg": "default", "attributes": [] }] }
+ { "jsonrpc": "2.0", "method": "menu_hide", "params": [] }
+ { "jsonrpc": "2.0", "method": "info_hide", "params": [] }
+-{ "jsonrpc": "2.0", "method": "draw_status", "params": [[], [{ "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "out 1:1 " }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": " " }, { "face": { "fg": "blue", "bg": "default", "attributes": [] }, "contents": "1 sel" }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": " - unnamed0@[kak-test-regression-638-highlight-codepoint-with-bracket]" }], { "fg": "cyan", "bg": "default", "attributes": [] }] }
++{ "jsonrpc": "2.0", "method": "draw_status", "params": [[], [{ "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": "out 1:4 " }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": " " }, { "face": { "fg": "blue", "bg": "default", "attributes": [] }, "contents": "1 sel" }, { "face": { "fg": "default", "bg": "default", "attributes": [] }, "contents": " - unnamed0@[kak-test-regression-638-highlight-codepoint-with-bracket]" }], { "fg": "cyan", "bg": "default", "attributes": [] }] }
+ { "jsonrpc": "2.0", "method": "refresh", "params": [true] }
+

test/regression/638-highlight-codepoint-with-bracket/rc

@@ -0,0 +1 @@
+addhl regex '[“”]' 0:red