Rewrite line modifications, hopefully with easier to comprehend code

5 files changed, 130 insertions, 69 deletions

diff --git a/src/highlighters.cc b/src/highlighters.cc
index 1f879b0b..20f09181 100644
--- a/src/highlighters.cc
+++ b/src/highlighters.cc
@@ -749,7 +749,7 @@ void update_matches(const Buffer& buffer, ArrayView<LineModification> modifs,
         if (not erase and modif_it != modifs.begin())
         {
             auto& prev = *(modif_it-1);
-            erase = it->line <= prev.old_line + prev.num_removed;
+            erase = it->line < prev.old_line + prev.num_removed;
             it->line += prev.diff();
         }
         erase = erase or (it->line >= buffer.line_count());
@@ -774,7 +774,7 @@ void update_matches(const Buffer& buffer, ArrayView<LineModification> modifs,
     for (auto& modif : modifs)
     {
         for (auto line = modif.new_line;
-             line < modif.new_line + modif.num_added+1 and
+             line < modif.new_line + modif.num_added and
              line < buffer.line_count(); ++line)
         {
             auto l = buffer[line];
diff --git a/src/line_modification.cc b/src/line_modification.cc
index 96249856..1943350f 100644
--- a/src/line_modification.cc
+++ b/src/line_modification.cc
@@ -5,41 +5,33 @@
 namespace Kakoune
 {
 
-namespace
+static LineModification make_line_modif(const Buffer::Change& change)
 {
-
-struct LineChange
-{
-    LineChange(const Buffer::Change& change)
+    LineModification res = {};
+    res.old_line = change.begin.line;
+    res.new_line = change.begin.line;
+    if (change.type == Buffer::Change::Insert)
     {
-        ByteCoord begin = change.begin;
-        ByteCoord end = change.end;
-        if (change.type == Buffer::Change::Insert)
-        {
-            if (change.at_end and begin != ByteCoord{0,0})
-            {
-                kak_assert(begin.column == 0);
-                --begin.line;
-            }
-            pos = begin.line;
-            num = end.line - begin.line;
-        }
-        else
-        {
-            if (change.at_end and begin != ByteCoord{0,0})
-            {
-                kak_assert(begin.column == 0);
-                --begin.line;
-            }
-            pos = begin.line;
-            num = begin.line - end.line;
-        }
+        res.num_added = change.end.line - change.begin.line;
+         // inserted a new line at buffer end but end coord is on same line
+        if (change.at_end and change.end.column != 0)
+            ++res.num_added;
     }
-
-    LineCount pos;
-    LineCount num;
-};
-
+    else
+    {
+        res.num_removed = change.end.line - change.begin.line;
+        // removed last line, but end coord is on same line
+        if (change.at_end and change.end.column != 0)
+            ++res.num_removed;
+    }
+    // modified a line
+    if (not change.at_end and
+        (change.begin.column != 0 or change.end.column != 0))
+    {
+        ++res.num_removed;
+        ++res.num_added;
+    }
+    return res;
 }
 
 Vector<LineModification> compute_line_modifications(const Buffer& buffer, size_t timestamp)
@@ -47,60 +39,51 @@ Vector<LineModification> compute_line_modifications(const Buffer& buffer, size_t
     Vector<LineModification> res;
     for (auto& buf_change : buffer.changes_since(timestamp))
     {
-        const LineChange change(buf_change);
+        auto change = make_line_modif(buf_change);
 
-        auto pos = std::upper_bound(res.begin(), res.end(), change.pos,
+        auto pos = std::upper_bound(res.begin(), res.end(), change.new_line,
                                     [](const LineCount& l, const LineModification& c)
                                     { return l < c.new_line; });
 
         if (pos != res.begin())
         {
             auto& prev = *(pos-1);
-            if (change.pos <= prev.new_line + prev.num_added)
+            if (change.new_line <= prev.new_line + prev.num_added)
+            {
                 --pos;
+                LineCount removed_from_previously_added_by_pos = clamp(pos->new_line + pos->num_added - change.new_line, 0_line, std::min(pos->num_added, change.num_removed));
+                pos->num_removed += change.num_removed - removed_from_previously_added_by_pos;
+                pos->num_added += change.num_added - removed_from_previously_added_by_pos;
+            }
             else
-                pos = res.insert(pos, {change.pos - prev.diff(), change.pos, 0, 0});
+            {
+                change.old_line -= prev.diff();
+                pos = res.insert(pos, change);
+            }
         }
         else
-            pos = res.insert(pos, {change.pos, change.pos, 0, 0});
+            pos = res.insert(pos, change);
 
         auto& modif = *pos;
         auto next = pos + 1;
-        if (change.num > 0)
+        if (buf_change.type == Buffer::Change::Erase)
         {
-            modif.num_added += change.num;
-            for (auto it = next; it != res.end(); ++it)
-                it->new_line += change.num;
-        }
-        if (change.num < 0)
-        {
-            const LineCount num_removed = -change.num;
-
-            auto delend = std::upper_bound(next, res.end(), change.pos + num_removed,
+            auto delend = std::upper_bound(next, res.end(), change.new_line + change.num_removed,
                                            [](const LineCount& l, const LineModification& c)
                                            { return l < c.new_line; });
 
             for (auto it = next; it != delend; ++it)
             {
-                LineCount removed_from_it = (change.pos + num_removed - it->new_line);
-                modif.num_removed += it->num_removed - std::min(removed_from_it, it->num_added);
-                modif.num_added += std::max(0_line, it->num_added - removed_from_it);
+                LineCount removed_from_previously_added_by_it = std::min(it->num_added, change.new_line + change.num_removed - it->new_line);
+                modif.num_removed += it->num_removed - removed_from_previously_added_by_it;
+                modif.num_added += it->num_added - removed_from_previously_added_by_it;
             }
             next = res.erase(next, delend);
-
-            const LineCount num_added_after_pos =
-                modif.new_line + modif.num_added - change.pos;
-            const LineCount num_removed_from_added =
-                std::min(num_removed, num_added_after_pos);
-
-            kak_assert(modif.num_added >= num_removed_from_added);
-
-            modif.num_added -= num_removed_from_added;
-            modif.num_removed += num_removed - num_removed_from_added;
-
-            for (auto it = next; it != res.end(); ++it)
-                it->new_line -= num_removed;
         }
+
+        auto diff = buf_change.end.line - buf_change.begin.line;
+        for (auto it = next; it != res.end(); ++it)
+            it->new_line += diff;
     }
     return res;
 }
diff --git a/src/line_modification.hh b/src/line_modification.hh
index d56a65d4..7c21503c 100644
--- a/src/line_modification.hh
+++ b/src/line_modification.hh
@@ -14,8 +14,8 @@ struct LineModification
 {
     LineCount old_line; // line position in the old buffer
     LineCount new_line; // new line position
-    LineCount num_removed; // number of lines removed after this one
-    LineCount num_added; // number of lines added after this one
+    LineCount num_removed; // number of lines removed (including this one)
+    LineCount num_added; // number of lines added (including this one)
 
     LineCount diff() const { return new_line - old_line + num_added - num_removed; }
 };
diff --git a/src/unit_tests.cc b/src/unit_tests.cc
index 504c9246..ec9bb64c 100644
--- a/src/unit_tests.cc
+++ b/src/unit_tests.cc
@@ -3,6 +3,9 @@
 #include "keys.hh"
 #include "selectors.hh"
 #include "word_db.hh"
+#include "line_modification.hh"
+
+#include <tuple>
 
 using namespace Kakoune;
 
@@ -151,6 +154,80 @@ void test_keys()
     kak_assert(keys == parsed_keys);
 }
 
+bool operator==(const LineModification& lhs, const LineModification& rhs)
+{
+    return std::tie(lhs.old_line, lhs.new_line, lhs.num_removed, lhs.num_added) ==
+           std::tie(rhs.old_line, rhs.new_line, rhs.num_removed, rhs.num_added);
+}
+
+void test_line_modifications()
+{
+    {
+        Buffer buffer("test", Buffer::Flags::None, { "line 1\n"_ss, "line 2\n"_ss });
+        auto ts = buffer.timestamp();
+        buffer.erase(buffer.iterator_at({1, 0}), buffer.iterator_at({2, 0}));
+
+        auto modifs = compute_line_modifications(buffer, ts);
+        kak_assert(modifs.size() == 1 && modifs[0] == LineModification{ 1 COMMA 1 COMMA 1 COMMA 0 });
+    }
+
+    {
+        Buffer buffer("test", Buffer::Flags::None, { "line 1\n"_ss, "line 2\n"_ss });
+        auto ts = buffer.timestamp();
+        buffer.insert(buffer.iterator_at({1, 7}), "line 3");
+
+        auto modifs = compute_line_modifications(buffer, ts);
+        kak_assert(modifs.size() == 1 && modifs[0] == LineModification{ 2 COMMA 2 COMMA 0 COMMA 1 });
+    }
+
+    {
+        Buffer buffer("test", Buffer::Flags::None,
+                      { "line 1\n"_ss, "line 2\n"_ss, "line 3\n"_ss });
+
+        auto ts = buffer.timestamp();
+        buffer.insert(buffer.iterator_at({1, 4}), "hoho\nhehe");
+        buffer.erase(buffer.iterator_at({0, 0}), buffer.iterator_at({1, 0}));
+
+        auto modifs = compute_line_modifications(buffer, ts);
+        kak_assert(modifs.size() == 1 && modifs[0] == LineModification{ 0 COMMA 0 COMMA 2 COMMA 2 });
+    }
+
+    {
+        Buffer buffer("test", Buffer::Flags::None,
+                      { "line 1\n"_ss, "line 2\n"_ss, "line 3\n"_ss, "line 4\n"_ss });
+
+        WordDB word_db(buffer);
+        word_db.find_matching("", prefix_match);
+
+        auto ts = buffer.timestamp();
+        buffer.erase(buffer.iterator_at({0,0}), buffer.iterator_at({3,0}));
+        buffer.insert(buffer.iterator_at({1,0}), "newline 1\nnewline 2\nnewline 3\n");
+        buffer.erase(buffer.iterator_at({0,0}), buffer.iterator_at({1,0}));
+        {
+            auto modifs = compute_line_modifications(buffer, ts);
+            kak_assert(modifs.size() == 1 && modifs[0] == LineModification{ 0 COMMA 0 COMMA 4 COMMA 3 });
+        }
+        buffer.insert(buffer.iterator_at({3,0}), "newline 4\n");
+
+        {
+            auto modifs = compute_line_modifications(buffer, ts);
+            kak_assert(modifs.size() == 1 && modifs[0] == LineModification{ 0 COMMA 0 COMMA 4 COMMA 4 });
+        }
+
+        word_db.find_matching("", prefix_match);
+    }
+
+    {
+        Buffer buffer("test", Buffer::Flags::None, { "line 1\n"_ss });
+        auto ts = buffer.timestamp();
+        buffer.insert(buffer.iterator_at({0,0}), "n");
+        buffer.insert(buffer.iterator_at({0,1}), "e");
+        buffer.insert(buffer.iterator_at({0,2}), "w");
+        auto modifs = compute_line_modifications(buffer, ts);
+        kak_assert(modifs.size() == 1 && modifs[0] == LineModification{ 0 COMMA 0 COMMA 1 COMMA 1 });
+    }
+}
+
 void run_unit_tests()
 {
     test_utf8();
@@ -159,4 +236,5 @@ void run_unit_tests()
     test_buffer();
     test_undo_group_optimizer();
     test_word_db();
+    test_line_modifications();
 }
diff --git a/src/word_db.cc b/src/word_db.cc
index 6b0eeb0a..2786396b 100644
--- a/src/word_db.cc
+++ b/src/word_db.cc
@@ -108,13 +108,13 @@ void WordDB::update_db()
 
         kak_assert((int)new_lines.size() == (int)modif.new_line);
 
-        while (old_line <= modif.old_line + modif.num_removed)
+        while (old_line < modif.old_line + modif.num_removed)
         {
             kak_assert(old_line < m_lines.size());
             remove_words(get_words(SharedString{m_lines[(int)old_line++]}));
         }
 
-        for (auto l = 0_line; l <= modif.num_added; ++l)
+        for (auto l = 0_line; l < modif.num_added; ++l)
         {
             if (modif.new_line + l >= buffer.line_count())
                 break;