progress bar shows pairs

blingenf · Apr 6, 2024 · 86bdfce · 86bdfce
1 parent 7522fde
commit 86bdfce
Showing 1 changed file with 20 additions and 24 deletions.
diff --git a/copydetect/detector.py b/copydetect/detector.py
@@ -420,32 +420,28 @@ def _comparison_loop(self):
 
         test_indices = {f: i for i, f in enumerate(self.test_files)}
         ref_indices = {f: i for i, f in enumerate(self.ref_files)}
-        comparison_by_file = defaultdict(list)
-        for test, ref in self.get_comparison_pairs():
-            comparison_by_file[test].append(ref)
-
-        for test_f, references in tqdm(comparison_by_file.items(),
-                                       bar_format='   {l_bar}{bar}{r_bar}',
-                                       disable=self.conf.silent
-                                       ):
+
+        for test_f, ref_f in tqdm(self.get_comparison_pairs(),
+                                  bar_format='   {l_bar}{bar}{r_bar}',
+                                  disable=self.conf.silent
+                                  ):
             i = test_indices[test_f]
-            for ref_f in references:
-                j = ref_indices[ref_f]
+            j = ref_indices[ref_f]
 
-                if (ref_f, test_f) in comparisons:
-                    ref_idx, test_idx = comparisons[(ref_f, test_f)]
-                    overlap = self.token_overlap_matrix[ref_idx, test_idx]
-                    sim2, sim1 = self.similarity_matrix[ref_idx, test_idx]
-                else:
-                    overlap, (sim1, sim2), (slices1, slices2) = compare_files(
-                        self.file_data[test_f], self.file_data[ref_f]
-                    )
-                    comparisons[(test_f, ref_f)] = (i, j)
-                    if slices1.shape[0] != 0:
-                        self.slice_matrix[(test_f, ref_f)] = [slices1, slices2]
-
-                self.similarity_matrix[i, j] = np.array([sim1, sim2])
-                self.token_overlap_matrix[i, j] = overlap
+            if (ref_f, test_f) in comparisons:
+                ref_idx, test_idx = comparisons[(ref_f, test_f)]
+                overlap = self.token_overlap_matrix[ref_idx, test_idx]
+                sim2, sim1 = self.similarity_matrix[ref_idx, test_idx]
+            else:
+                overlap, (sim1, sim2), (slices1, slices2) = compare_files(
+                    self.file_data[test_f], self.file_data[ref_f]
+                )
+                comparisons[(test_f, ref_f)] = (i, j)
+                if slices1.shape[0] != 0:
+                    self.slice_matrix[(test_f, ref_f)] = [slices1, slices2]
+
+            self.similarity_matrix[i, j] = np.array([sim1, sim2])
+            self.token_overlap_matrix[i, j] = overlap
 
         if not self.conf.silent:
             print(f"{time.time()-start_time:6.2f}: Code comparison completed")