Cu 8699qbr8e Multiprocessing empty batches (#35)

* CU-8699qbr8e: Fix issue with multiprocessing in case of empty batches

* CU-8699qbr8e: Add a few simple tests to make sure empty and small datasets can be multiprocessed

Author: mart-r

medcat-v2/medcat/cat.py

@@ -265,6 +265,9 @@ def _mp_one_batch_per_process(
                     executor.submit(self._mp_worker_func, batch))
             except StopIteration:
                 break
+        if not futures:
+            # NOTE: if there wasn't any data, we didn't process anything
+            return
         # Main process works on next batch while workers are busy
         main_batch: Optional[list[tuple[str, str, bool]]]
         try:
@@ -282,6 +285,10 @@ def _mp_one_batch_per_process(
         # so we're going to wait for them to finish, yield their results,
         # and subsequently submit the next batch to keep them busy
         for _ in range(external_processes):
+            if not futures:
+                # NOTE: if there's no futures then there can't be
+                #       anything to batch
+                break
             # Wait for any future to complete
             done_future = next(as_completed(futures))
             futures.remove(done_future)

medcat-v2/tests/test_cat.py

@@ -418,13 +418,29 @@ def test_can_get_multiple_entities(self):
             "The dog is sitting outside the house."
         ]
         ents = list(self.cat.get_entities_multi_texts(texts))
+        self.assert_ents(ents, texts)
+
+    def assert_ents(self, ents: list[tuple], texts: list[str]):
         self.assertEqual(len(ents), len(texts))
         # NOTE: text IDs are integers starting from 0
         exp_ids = set(str(i) for i in range(len(texts)))
         for ent_id_str, ent in ents:
             with self.subTest(f"Entity: {ent_id_str} [{ent}]"):
                 self.assertIn(ent_id_str, exp_ids)
 
+    def test_can_multiprocess_empty(self):
+        texts = []
+        ents = list(self.cat.get_entities_multi_texts(texts, n_process=3))
+        self.assert_ents(ents, texts)
+
+    def test_can_get_multiprocess(self):
+        texts = [
+            "The fittest most fit of chronic kidney failure",
+            "The dog is sitting outside the house."
+        ]
+        ents = list(self.cat.get_entities_multi_texts(texts, n_process=3))
+        self.assert_ents(ents, texts)
+
 
 class CATWithDocAddonTests(CATIncludingTests):
     EXAMPLE_TEXT = "Example text to tokenize"