[2.8] Fix Max Frequency Misscalculation - [MOD-8158] (#5560)

GuyAv46 · web-flow · commit a6864e3f7dff · 2025-01-27T13:38:54.000Z
Fix Max Frequency Misscalculation - [MOD-8158] (#5553) * fix unrelated test * add a failing test * fix issue * revert whitespace change from test_vecsim.py * revert whitespace change in test_issues.py (cherry picked from commit b398d93)
diff --git a/src/document.c b/src/document.c
@@ -676,7 +676,6 @@ FIELD_PREPROCESSOR(vectorPreprocessor) {
                            fs->vectorOpts.expBlobSize);
     return -1;
   }
-  aCtx->fwIdx->maxFreq++;
   return 0;
 }
 
diff --git a/tests/pytests/test_issues.py b/tests/pytests/test_issues.py
@@ -1300,6 +1300,13 @@ def test_mod_6783(env:Env):
       res = env.cmd('FT.SEARCH', 'idx', '*', 'SORTBY', f'f{i}', 'NOCONTENT')
       env.assertEqual(res, expected[i], message=f'Failed on field f{i} with {n_sortables} sortables')
 
+@skip(cluster=True)
+def test_mod_8589(env:Env):
+  env.expect('FT.CREATE', 'idx', 'SCHEMA', 't', 'TEXT', 'v', 'VECTOR', 'FLAT', '6', 'TYPE', 'FLOAT32', 'DIM', '2', 'DISTANCE_METRIC', 'L2').ok()
+  env.cmd('HSET', 'doc1', 'v', '????????', 't', 'foo bar foo') # Max frequency is 2 (foo)
+  docinfo = to_dict(env.cmd(debug_cmd(), 'DOCINFO', 'idx', 'doc1'))
+  env.assertEqual(docinfo['max_freq'], 2)
+
 @skip(cluster=True)
 def test_mod_8568(env:Env):
   env.expect('FT.CREATE', 'idx', 'SCHEMA', 'g', 'GEO').ok()
diff --git a/tests/pytests/test_vecsim.py b/tests/pytests/test_vecsim.py
@@ -786,12 +786,12 @@ def test_hybrid_query_batches_mode_with_tags():
 
     for data_type in VECSIM_DATA_TYPES:
         conn.execute_command('FT.CREATE', 'idx', 'SCHEMA', 'v', 'VECTOR', 'HNSW', '8', 'TYPE', data_type,
-                            'DIM', dim, 'DISTANCE_METRIC', 'L2', 'EF_RUNTIME', 100, 'tags', 'TAG')
+                            'DIM', dim, 'DISTANCE_METRIC', 'L2', 'EF_RUNTIME', 100, 'tags', 'TAG', 'text', 'TEXT')
 
         p = conn.pipeline(transaction=False)
         for i in range(1, index_size+1):
             vector = create_np_array_typed([i]*dim, data_type)
-            p.execute_command('HSET', i, 'v', vector.tobytes(), 'tags', 'hybrid')
+            p.execute_command('HSET', i, 'v', vector.tobytes(), 'tags', 'hybrid', 'text', 'text')
         p.execute()
 
         query_data = create_np_array_typed([index_size/2]*dim, data_type)

Original file line number	Diff line number	Diff line change
`@@ -676,7 +676,6 @@ FIELD_PREPROCESSOR(vectorPreprocessor) {`
`676`	`676`	`fs->vectorOpts.expBlobSize);`
`677`	`677`	`return -1;`
`678`	`678`	`}`
`679`		`- aCtx->fwIdx->maxFreq++;`
`680`	`679`	`return 0;`
`681`	`680`	`}`
`682`	`681`