From 02542b1e695caa48c492dd8cc97a079679adbd8a Mon Sep 17 00:00:00 2001
From: Fangrui Song <i@maskray.me>
Date: Sun, 18 Mar 2018 12:17:40 -0700
Subject: [PATCH] Optimize FuzzyMatcher and add tests.

---
 src/fuzzy_match.cc               | 76 +++++++++++++++++++++++++-------
 src/fuzzy_match.h                |  2 +-
 src/messages/workspace_symbol.cc |  5 ++-
 3 files changed, 66 insertions(+), 17 deletions(-)

diff --git a/src/fuzzy_match.cc b/src/fuzzy_match.cc
index db5a5d5f..480b69d2 100644
--- a/src/fuzzy_match.cc
+++ b/src/fuzzy_match.cc
@@ -1,7 +1,11 @@
 #include "fuzzy_match.h"
 
+#include <doctest/doctest.h>
+
 #include <ctype.h>
+#include <stdio.h>
 #include <algorithm>
+#include <vector>
 
 enum CharClass { Other, Lower, Upper };
 enum CharRole { None, Tail, Head };
@@ -49,20 +53,20 @@ int FuzzyMatcher::MissScore(int j, bool last) {
 }
 
 int FuzzyMatcher::MatchScore(int i, int j, bool last) {
-  int s = 40;
+  int s = 0;
   if (pat[i] == text[j]) {
     s++;
     if ((pat_set & 1 << Upper) || i == j)
-      s += 20;
+      s += 10;
   }
   if (pat_role[i] == Head && text_role[j] == Head)
-    s += 50;
+    s += 30;
   if (text_role[j] == Tail && i && !last)
-    s -= 50;
-  if (pat_role[i] == Head && text_role[j] == Tail)
     s -= 30;
+  if (pat_role[i] == Head && text_role[j] == Tail)
+    s -= 10;
   if (i == 0 && text_role[j] == Tail)
-    s -= 70;
+    s -= 40;
   return s;
 }
 
@@ -87,31 +91,73 @@ int FuzzyMatcher::Match(std::string_view text) {
     low_text[i] = (char)::tolower(text[i]);
   CalculateRoles(text, text_role, &text_set);
   dp[0][0][0] = 0;
-  dp[0][0][1] = kMinScore;
+  dp[0][0][1] = kMinScore * 2;
   for (int j = 0; j < n; j++) {
     dp[0][j + 1][0] = dp[0][j][0] + MissScore(j, false);
-    dp[0][j + 1][1] = kMinScore;
+    dp[0][j + 1][1] = kMinScore * 2;
   }
   for (int i = 0; i < int(pat.size()); i++) {
     int(*pre)[2] = dp[i & 1];
     int(*cur)[2] = dp[(i + 1) & 1];
-    cur[0][0] = cur[0][1] = kMinScore;
-    for (int j = 0; j < n; j++) {
+    cur[i][0] = cur[i][1] = kMinScore;
+    for (int j = i; j < n; j++) {
       cur[j + 1][0] = std::max(cur[j][0] + MissScore(j, false),
                                cur[j][1] + MissScore(j, true));
-      if (low_pat[i] != low_text[j])
-        cur[j + 1][1] = kMinScore;
-      else {
+      // For the first char of pattern, apply extra restriction to filter bad
+      // candidates (e.g. |int| in |PRINT|)
+      if (low_pat[i] == low_text[j] &&
+          (i || text_role[j] != Tail || pat[i] == text[j])) {
         cur[j + 1][1] = std::max(pre[j][0] + MatchScore(i, j, false),
                                  pre[j][1] + MatchScore(i, j, true));
-      }
+      } else
+        cur[j + 1][1] = kMinScore * 2;
     }
   }
 
   // Enumerate the end position of the match in str. Each removed trailing
   // character has a penulty.
   int ret = kMinScore;
-  for (int j = 1; j <= n; j++)
+  for (int j = pat.size(); j <= n; j++)
     ret = std::max(ret, dp[pat.size() & 1][j][1] - 3 * (n - j));
   return ret;
 }
+
+TEST_SUITE("fuzzy_match") {
+  bool Ranks(std::string_view pat, std::vector<const char*> texts) {
+    FuzzyMatcher fuzzy(pat);
+    std::vector<int> scores;
+    for (auto text : texts)
+      scores.push_back(fuzzy.Match(text));
+    bool ret = true;
+    for (size_t i = 0; i < texts.size() - 1; i++)
+      if (scores[i] < scores[i + 1]) {
+        ret = false;
+        break;
+      }
+    if (1 || !ret) {
+      for (size_t i = 0; i < texts.size(); i++)
+        printf("%s %d ", texts[i], scores[i]);
+      puts("");
+    }
+    return ret;
+  }
+
+  TEST_CASE("test") {
+    // case
+    Ranks("monad", {"monad", "Monad", "mONAD"});
+    // initials
+    Ranks("ab", {"ab", "aoo_boo", "acb"});
+    Ranks("CC", {"CamelCase", "camelCase", "camelcase"});
+    Ranks("cC", {"camelCase", "CamelCase", "camelcase"});
+    Ranks("Da.Te", {"Data.Text", "Data.Text.Lazy", "Data.Aeson.Encoding.text"});
+    // prefix
+    Ranks("is", {"isIEEE", "inSuf"});
+    // shorter
+    Ranks("ma", {"map", "many", "maximum"});
+    Ranks("print", {"printf", "sprintf"});
+    // score(PRINT) = kMinScore
+    Ranks("int", {"int", "INT", "PRINT"});
+    // score(PRINT) > kMinScore
+    Ranks("Int", {"int", "INT", "PRINT"});
+  }
+}
diff --git a/src/fuzzy_match.h b/src/fuzzy_match.h
index a3b5191c..79599857 100644
--- a/src/fuzzy_match.h
+++ b/src/fuzzy_match.h
@@ -11,7 +11,7 @@ public:
   constexpr static int kMaxText = 200;
   // Negative but far from INT_MIN so that intermediate results are hard to
   // overflow.
-  constexpr static int kMinScore = INT_MIN / 2;
+  constexpr static int kMinScore = INT_MIN / 4;
 
   FuzzyMatcher(std::string_view pattern);
   int Match(std::string_view text);
diff --git a/src/messages/workspace_symbol.cc b/src/messages/workspace_symbol.cc
index f6a00f27..e02b2212 100644
--- a/src/messages/workspace_symbol.cc
+++ b/src/messages/workspace_symbol.cc
@@ -136,7 +136,10 @@ struct WorkspaceSymbolHandler : BaseMessageHandler<Ipc_WorkspaceSymbol> {
       std::sort(permutation.begin(), permutation.end(),
                 std::greater<std::pair<int, int>>());
       out.result.reserve(result_indices.size());
-      for (int i = 0; i < int(result_indices.size()); i++)
+      // Discard awful candidates.
+      for (int i = 0; i < int(result_indices.size()) &&
+                      permutation[i].first > FuzzyMatcher::kMinScore;
+           i++)
         out.result.push_back(
             std::move(unsorted_results[permutation[i].second]));
     } else {