ranked match: prefer real word boundaries to camelCase boundaries

This enables the test case from
mawww#4653 (comment)
and a few others from that PR.

src/ranked_match.cc

@@ -36,17 +36,23 @@ bool matches(UsedLetters query, UsedLetters letters)
 
 using Utf8It = utf8::iterator<const char*>;
 
-static int count_word_boundaries_match(StringView candidate, StringView query)
+struct WordBoundaries {
+    int count;
+    int weighted_count;
+};
+
+static WordBoundaries count_word_boundaries_match(StringView candidate, StringView query)
 {
     int count = 0;
+    int weighted_count = 0;
     Utf8It query_it{query.begin(), query};
     Codepoint prev = 0;
     for (Utf8It it{candidate.begin(), candidate}; it != candidate.end(); ++it)
     {
         const Codepoint c = *it;
-        const bool is_word_boundary = prev == 0 or
-                                      (!iswalnum((wchar_t)prev) and iswalnum((wchar_t)c)) or
-                                      (iswlower((wchar_t)prev) and iswupper((wchar_t)c));
+        const int is_word_boundary = 2 * (prev == 0 or
+                                         (!iswalnum((wchar_t)prev) and iswalnum((wchar_t)c))) +
+                                     1 * (iswlower((wchar_t)prev) and iswupper((wchar_t)c));
         prev = c;
 
         if (not is_word_boundary)
@@ -59,14 +65,15 @@ static int count_word_boundaries_match(StringView candidate, StringView query)
             if (qc == (iswlower((wchar_t)qc) ? lc  : c))
             {
                 ++count;
+                weighted_count += is_word_boundary;
                 query_it = qit+1;
                 break;
             }
         }
         if (query_it == query.end())
             break;
     }
-    return count;
+    return WordBoundaries{count, weighted_count};
 }
 
 static bool smartcase_eq(Codepoint candidate, Codepoint query)
@@ -163,8 +170,9 @@ RankedMatch::RankedMatch(StringView candidate, StringView query, TestFunc func)
         }
     }
 
-    m_word_boundary_match_count = count_word_boundaries_match(candidate, query);
-    if (m_word_boundary_match_count == query.length())
+    auto word_boundaries = count_word_boundaries_match(candidate, query);
+    m_word_boundary_match_count = word_boundaries.weighted_count;
+    if (word_boundaries.count == query.length())
         m_flags |= Flags::OnlyWordBoundary;
 }
 
@@ -198,7 +206,7 @@ bool RankedMatch::operator<(const RankedMatch& other) const
 
     // If we are SingleWord, we dont want to take word boundaries from other
     // words into account.
-    if (not (m_flags & (Flags::Prefix | Flags::SingleWord)) and
+    if (not (m_flags & Flags::Prefix) and
         m_word_boundary_match_count != other.m_word_boundary_match_count)
         return m_word_boundary_match_count > other.m_word_boundary_match_count;
 
@@ -252,11 +260,11 @@ UnitTest test_ranked_match{[] {
         return RankedMatch{better, query} < RankedMatch{worse, query};
     };
 
-    kak_assert(count_word_boundaries_match("run_all_tests", "rat") == 3);
-    kak_assert(count_word_boundaries_match("run_all_tests", "at") == 2);
-    kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "wm") == 2);
-    kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "cobm") == 3);
-    kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "cWBM") == 4);
+    kak_assert(count_word_boundaries_match("run_all_tests", "rat").count == 3);
+    kak_assert(count_word_boundaries_match("run_all_tests", "at").count == 2);
+    kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "wm").count == 2);
+    kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "cobm").count == 3);
+    kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "cWBM").count == 4);
     kak_assert(preferred("so", "source", "source_data"));
     kak_assert(not preferred("so", "source_data", "source"));
     kak_assert(not preferred("so", "source", "source"));
@@ -286,6 +294,9 @@ UnitTest test_ranked_match{[] {
     kak_assert(preferred("codegen", "clang/test/CodeGen/asm.c", "clang/test/ASTMerge/codegen-body/test.c"));
     kak_assert(preferred("cho", "tchou kanaky", "tachou kanay")); // Prefer the leftmost match.
     kak_assert(preferred("clangd", "clang-tools-extra/clangd/README.md", "clang/docs/conf.py"));
+    kak_assert(preferred("luaremote", "src/script/LuaRemote.cpp", "tests/TestLuaRemote.cpp"));
+    kak_assert(preferred("rm.cc", "src/ranked_match.cc", "test/README.asciidoc"));
+    kak_assert(preferred("fooo", "foo.o", "fo.o.o"));
 }};
 
 UnitTest test_used_letters{[]()