dispatch: Add a match ratio metric

Author: balayette

display.py

@@ -99,7 +99,7 @@ def pair_graphs(pairs):
 
         f = Page(f"out/pairs/{idx}.html")
         f.write("<div>")
-        f.write(f"<h1>{pair['directory1']} | {pair['directory2']}</h1>")
+        f.write(f"<h1>{pair['directory1']} | {pair['directory2']} | {pair['match_ratio']}</h1>")
         f.write("</div>")
 
         left = ""
@@ -178,7 +178,7 @@ def index(pairs):
     f.write("<h1>Pairs</h1>")
 
     f.write(
-        "<table><tr><th>Directory 1</th><th>Directory 2</th><th>Matching files</th><th>Details</th></tr>"
+        "<table><tr><th>Directory 1</th><th>Directory 2</th><th>Score</th><th>Details</th></tr>"
     )
 
     for idx, pair in enumerate(pairs):
@@ -189,7 +189,7 @@ def index(pairs):
 
         f.write(f"<td>{pair['directory1']}</td>")
         f.write(f"<td>{pair['directory2']}</td>")
-        f.write(f"<td>{len(pair['matches'])}</td>")
+        f.write(f"<td>{pair['match_ratio']}</td>")
         f.write(f'<td><a href="pairs/{idx}.html">Details</a></td>')
 
         f.write("</tr>")
@@ -205,7 +205,7 @@ def main():
     with open("pairs.json", "r") as f:
         pairs = json.load(f)
 
-    pairs = sorted(pairs, key=lambda pair: -len(pair["matches"]))
+    pairs = sorted(pairs, key=lambda pair: -pair["match_ratio"])
 
     pair_graphs(pairs)
     index(pairs)

src/dispatch.cc

@@ -53,6 +53,7 @@ struct Pair {
   std::vector<Match> matches;
   Directory *directory1;
   Directory *directory2;
+  float match_ratio;
 };
 
 std::regex *glob = nullptr;
@@ -153,6 +154,8 @@ void do_diff(Pair *pair, Directory *d1, Directory *d2) {
   pair->directory1 = d1;
   pair->directory2 = d2;
 
+  unsigned file_count = 0;
+  double totalSim = 0.0;
   for (size_t i = 0; i < d1->sexps.size(); i++) {
     for (size_t j = 0; j < d2->sexps.size(); j++) {
       const std::string &p1 = d1->sexps[i].path;
@@ -176,10 +179,18 @@ void do_diff(Pair *pair, Directory *d1, Directory *d2) {
       if (s < sim)
         continue;
 
+      totalSim += s;
+      file_count++;
+
       pair->matches.emplace_back(&d1->sexps[i], &d2->sexps[j], s,
                                  MappingsVec2(t1, mapping));
     }
   }
+
+  if (file_count == 0)
+    pair->match_ratio = 0.0;
+  else
+    pair->match_ratio = totalSim / file_count;
 }
 
 std::ostream &dump_match(const Match &match, std::ostream &os) {
@@ -219,6 +230,7 @@ void dump_pairs(const std::vector<Pair> &pairs) {
 
     f << "\"directory1\": \"" << pair.directory1->cc_path << "\",";
     f << "\"directory2\": \"" << pair.directory2->cc_path << "\",";
+    f << "\"match_ratio\": " << pair.match_ratio << ",";
     f << "\"matches\": [";
     for (size_t i = 0; i < pair.matches.size(); i++) {
       dump_match(pair.matches[i], f);