Refactor region highlighter, extract code to helper classes

src/highlighters.cc

@@ -667,59 +667,11 @@ HighlighterAndId reference_factory(HighlighterParameters params)
                             });
 }
 
-struct RegionHighlighter
+namespace RegionHighlight
 {
-public:
-    RegionHighlighter(Regex begin, Regex end, Regex recurse = Regex{})
-        : m_begin(std::move(begin)),
-          m_end(std::move(end)),
-          m_recurse(std::move(recurse))
-    {
-        if (m_begin.empty() or m_end.empty())
-            throw runtime_error("invalid regex for region highlighter");
-    }
 
-    void operator()(HierachicalHighlighter::GroupMap groups, const Context& context,
+struct Match
-                    HighlightFlags flags, DisplayBuffer& display_buffer)
+{
-    {
-        if (flags != HighlightFlags::Highlight)
-            return;
-
-        auto it = groups.find("content");
-        if (it == groups.end())
-            return;
-
-        auto range = display_buffer.range();
-        const auto& buffer = context.buffer();
-        auto& regions = update_cache_ifn(buffer);
-        auto begin = std::lower_bound(regions.begin(), regions.end(), range.first,
-                                      [](const Region& r, const ByteCoord& c) { return r.end < c; });
-        auto end = std::lower_bound(begin, regions.end(), range.second,
-                                    [](const Region& r, const ByteCoord& c) { return r.begin < c; });
-        auto correct = [&](const ByteCoord& c) -> ByteCoord {
-            if (buffer[c.line].length() == c.column)
-                return {c.line+1, 0};
-            return c;
-        };
-        for (; begin != end; ++begin)
-            apply_highlighter(context, flags, display_buffer,
-                              correct(begin->begin), correct(begin->end),
-                              it->second);
-    }
-private:
-    Regex m_begin;
-    Regex m_end;
-    Regex m_recurse;
-
-    struct Region
-    {
-        ByteCoord begin;
-        ByteCoord end;
-    };
-    using RegionList = std::vector<Region>;
-
-    struct Match
-    {
     size_t timestamp;
     LineCount line;
     ByteCount begin;
@@ -727,93 +679,11 @@ private:
 
     ByteCoord begin_coord() const { return { line, begin }; }
     ByteCoord end_coord() const { return { line, end }; }
-    };
+};
-    using MatchList = std::vector<Match>;
+using MatchList = std::vector<Match>;
 
-    struct Cache
+void find_matches(const Buffer& buffer, MatchList& matches, const Regex& regex)
-    {
+{
-        size_t timestamp = 0;
-        MatchList begin_matches;
-        MatchList end_matches;
-        MatchList recurse_matches;
-        RegionList regions;
-    };
-    BufferSideCache<Cache> m_cache;
-
-    const RegionList& update_cache_ifn(const Buffer& buffer)
-    {
-        Cache& cache = m_cache.get(buffer);
-        const size_t buf_timestamp = buffer.timestamp();
-        if (cache.timestamp == buf_timestamp)
-            return cache.regions;
-
-        if (cache.timestamp == 0)
-        {
-            find_matches(buffer, cache.begin_matches, m_begin);
-            find_matches(buffer, cache.end_matches, m_end);
-            if (not m_recurse.empty())
-                find_matches(buffer, cache.recurse_matches, m_recurse);
-        }
-        else
-        {
-            auto modifs = compute_line_modifications(buffer, cache.timestamp);
-            update_matches(buffer, modifs, cache.begin_matches, m_begin);
-            update_matches(buffer, modifs, cache.end_matches, m_end);
-            if (not m_recurse.empty())
-                update_matches(buffer, modifs, cache.recurse_matches, m_recurse);
-        }
-
-        auto compare_matches_end = [](const Match& lhs, const Match& rhs) {
-            return lhs.end_coord() < rhs.end_coord();
-        };
-
-        cache.regions.clear();
-        int recurse_level = 0;
-        auto ref_it = cache.begin_matches.begin();
-        for (auto beg_it = cache.begin_matches.begin(),
-                  end_it = cache.end_matches.begin(),
-                  rec_it = cache.recurse_matches.begin();
-             beg_it != cache.begin_matches.end(); )
-        {
-            end_it = std::upper_bound(end_it, cache.end_matches.end(),
-                                      *ref_it, compare_matches_end);
-            rec_it = std::upper_bound(rec_it, cache.recurse_matches.end(),
-                                      *ref_it, compare_matches_end);
-
-            if (end_it == cache.end_matches.end())
-            {
-                cache.regions.push_back({ {beg_it->line, beg_it->begin},
-                                          buffer.end_coord() });
-                break;
-            }
-
-            while (rec_it != cache.recurse_matches.end() and
-                   rec_it->end_coord() < end_it->begin_coord())
-            {
-                ++recurse_level;
-                ++rec_it;
-            }
-
-            if (recurse_level == 0)
-            {
-                cache.regions.push_back({ beg_it->begin_coord(),
-                                          end_it->end_coord() });
-                beg_it = std::upper_bound(beg_it, cache.begin_matches.end(),
-                                          *end_it, compare_matches_end);
-                ref_it = beg_it;
-            }
-            else
-            {
-                --recurse_level;
-                ref_it = end_it;
-            }
-        }
-        cache.timestamp = buf_timestamp;
-        return cache.regions;
-    }
-
-    void find_matches(const Buffer& buffer, MatchList& matches, const Regex& regex)
-    {
     const size_t buf_timestamp = buffer.timestamp();
     for (auto line = 0_line, end = buffer.line_count(); line < end; ++line)
     {
@@ -825,10 +695,11 @@ private:
             matches.push_back({ buf_timestamp, line, b, e });
         }
     }
-    }
+}
 
-    void update_matches(const Buffer& buffer, memoryview<LineModification> modifs, MatchList& matches, const Regex& regex)
+void update_matches(const Buffer& buffer, memoryview<LineModification> modifs,
-    {
+                    MatchList& matches, const Regex& regex)
+{
     const size_t buf_timestamp = buffer.timestamp();
     // remove out of date matches and update line for others
     auto ins_pos = matches.begin();
@@ -883,6 +754,175 @@ private:
                        [](const Match& lhs, const Match& rhs) {
                            return lhs.begin_coord() < rhs.begin_coord();
                        });
+}
+
+struct RegionMatches
+{
+    MatchList begin_matches;
+    MatchList end_matches;
+    MatchList recurse_matches;
+
+    static bool compare_to_end(ByteCoord lhs, const Match& rhs)
+    {
+        return lhs < rhs.end_coord();
+    }
+
+    MatchList::const_iterator find_next_begin(ByteCoord pos) const
+    {
+        return std::upper_bound(begin_matches.begin(), begin_matches.end(),
+                                pos, compare_to_end);
+    }
+
+    MatchList::const_iterator find_matching_end(MatchList::const_iterator beg_it) const
+    {
+        auto end_it = end_matches.begin();
+        auto rec_it = recurse_matches.begin();
+        auto ref_pos = beg_it->end_coord();
+        int recurse_level = 0;
+        while (true)
+        {
+            end_it = std::upper_bound(end_it, end_matches.end(),
+                                      ref_pos, compare_to_end);
+            rec_it = std::upper_bound(rec_it, recurse_matches.end(),
+                                      ref_pos, compare_to_end);
+
+            if (end_it == end_matches.end())
+                return end_it;
+
+            while (rec_it != recurse_matches.end() and
+                   rec_it->end_coord() < end_it->begin_coord())
+            {
+                ++recurse_level;
+                ++rec_it;
+            }
+
+            if (recurse_level == 0)
+                return end_it;
+
+            --recurse_level;
+            ref_pos = end_it->end_coord();
+        }
+    }
+};
+
+struct RegionDesc
+{
+    Regex m_begin;
+    Regex m_end;
+    Regex m_recurse;
+
+    RegionMatches find_matches(const Buffer& buffer) const
+    {
+        RegionMatches res;
+        RegionHighlight::find_matches(buffer, res.begin_matches, m_begin);
+        RegionHighlight::find_matches(buffer, res.end_matches, m_end);
+        if (not m_recurse.empty())
+            RegionHighlight::find_matches(buffer, res.recurse_matches, m_recurse);
+        return res;
+    }
+
+    void update_matches(const Buffer& buffer,
+                        memoryview<LineModification> modifs,
+                        RegionMatches& matches) const
+    {
+        RegionHighlight::update_matches(buffer, modifs, matches.begin_matches, m_begin);
+        RegionHighlight::update_matches(buffer, modifs, matches.end_matches, m_end);
+        if (not m_recurse.empty())
+            RegionHighlight::update_matches(buffer, modifs, matches.recurse_matches, m_recurse);
+    }
+};
+
+struct RegionHighlighter
+{
+public:
+    RegionHighlighter(Regex begin, Regex end, Regex recurse = Regex{})
+        : m_region{ std::move(begin), std::move(end), std::move(recurse) }
+    {
+        if (m_region.m_begin.empty() or m_region.m_end.empty())
+            throw runtime_error("invalid regex for region highlighter");
+    }
+
+    void operator()(HierachicalHighlighter::GroupMap groups, const Context& context,
+                    HighlightFlags flags, DisplayBuffer& display_buffer)
+    {
+        if (flags != HighlightFlags::Highlight)
+            return;
+
+        auto it = groups.find("content");
+        if (it == groups.end())
+            return;
+
+        auto range = display_buffer.range();
+        const auto& buffer = context.buffer();
+        auto& regions = update_cache_ifn(buffer);
+        auto begin = std::lower_bound(regions.begin(), regions.end(), range.first,
+                                      [](const Region& r, ByteCoord c) { return r.end < c; });
+        auto end = std::lower_bound(begin, regions.end(), range.second,
+                                    [](const Region& r, ByteCoord c) { return r.begin < c; });
+        auto correct = [&](ByteCoord c) -> ByteCoord {
+            if (buffer[c.line].length() == c.column)
+                return {c.line+1, 0};
+            return c;
+        };
+        for (; begin != end; ++begin)
+            apply_highlighter(context, flags, display_buffer,
+                              correct(begin->begin), correct(begin->end),
+                              it->second);
+    }
+private:
+    RegionDesc m_region;
+
+    struct Region
+    {
+        ByteCoord begin;
+        ByteCoord end;
+    };
+    using RegionList = std::vector<Region>;
+
+    struct Cache
+    {
+        size_t timestamp = 0;
+        RegionMatches matches;
+        RegionList regions;
+    };
+    BufferSideCache<Cache> m_cache;
+
+    const RegionList& update_cache_ifn(const Buffer& buffer)
+    {
+        Cache& cache = m_cache.get(buffer);
+        const size_t buf_timestamp = buffer.timestamp();
+        if (cache.timestamp == buf_timestamp)
+            return cache.regions;
+
+        if (cache.timestamp == 0)
+            cache.matches = m_region.find_matches(buffer);
+        else
+        {
+            auto modifs = compute_line_modifications(buffer, cache.timestamp);
+            m_region.update_matches(buffer, modifs, cache.matches);
+        }
+
+        cache.regions.clear();
+        for (auto beg_it = cache.matches.begin_matches.cbegin();
+             beg_it != cache.matches.begin_matches.end(); )
+        {
+            auto end_it = cache.matches.find_matching_end(beg_it);
+
+            if (end_it == cache.matches.end_matches.end())
+            {
+                cache.regions.push_back({ {beg_it->line, beg_it->begin},
+                                          buffer.end_coord() });
+                break;
+            }
+            else
+            {
+                cache.regions.push_back({ beg_it->begin_coord(),
+                                          end_it->end_coord() });
+                beg_it = cache.matches.find_next_begin(end_it->end_coord());
+            }
+        }
+        cache.timestamp = buf_timestamp;
+        return cache.regions;
     }
 };
 
@@ -911,6 +951,8 @@ HighlighterAndId region_factory(HighlighterParameters params)
     }
 }
 
+}
+
 void register_highlighters()
 {
     HighlighterRegistry& registry = HighlighterRegistry::instance();
@@ -925,7 +967,7 @@ void register_highlighters()
     registry.register_func("group", highlighter_group_factory);
     registry.register_func("flag_lines", flag_lines_factory);
     registry.register_func("ref", reference_factory);
-    registry.register_func("region", region_factory);
+    registry.register_func("region", RegionHighlight::region_factory);
 }
 
 }