CU-8693az82g Remove cdb tests side effects (#380)

* 8693az82g: Add method to CDBMaker to reset the CDB

* 8693az82g: Add test in CDB tests to ensure a new CDB is used for each test

* 8693az82g: Reset CDB in CDB tests before each test to avoid side effects

medcat/cdb_maker.py

@@ -49,6 +49,14 @@ def __init__(self, config: Config, cdb: Optional[CDB] = None) -> None:
                              name='skip_and_punct',
                              additional_fields=['is_punct'])
 
+    def reset_cdb(self) -> None:
+        """This will re-create a new internal CDB based on the same config.
+
+        This will be necessary if/when you're wishing to call `prepare_csvs`
+        multiple times on the same object `CDBMaker` instance.
+        """
+        self.cdb = CDB(config=self.config)
+
     def prepare_csvs(self,
                      csv_paths: Union[pd.DataFrame, List[str]],
                      sep: str = ',',
@@ -59,6 +67,12 @@ def prepare_csvs(self,
                      only_existing_cuis: bool = False, **kwargs) -> CDB:
         r"""Compile one or multiple CSVs into a CDB.
 
+        Note: This class/method generally uses the same instance of the CDB.
+              So if you're using the same CDBMaker and calling `prepare_csvs`
+              multiple times, you are likely to get leakage from prior calls
+              into new ones.
+              To reset the CDB, call `reset_cdb`.
+
         Args:
             csv_paths (Union[pd.DataFrame, List[str]]):
                 An array of paths to the csv files that should be processed. Can also be an array of pd.DataFrames

tests/test_cdb.py

@@ -22,11 +22,21 @@ def setUp(self) -> None:
         cdb_2_csv = os.path.join(os.path.dirname(os.path.realpath(__file__)), "..", "examples", "cdb_2.csv")
         self.tmp_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "tmp")
         os.makedirs(self.tmp_dir, exist_ok=True)
+        # resetting the CDB because otherwise the CDBMaker
+        # will refer to and modify the same instance of the CDB
+        # and this can (and does!) create side effects
+        CDBTests.cdb_maker.reset_cdb()
         self.undertest = CDBTests.cdb_maker.prepare_csvs([cdb_csv, cdb_2_csv], full_build=True)
 
     def tearDown(self) -> None:
         shutil.rmtree(self.tmp_dir)
 
+    def test_setup_changes_cdb(self):
+        id1 = id(self.undertest)
+        self.setUp()
+        id2 = id(self.undertest)
+        self.assertNotEqual(id1, id2)
+
     def test_name2cuis(self):
         self.assertEqual({
             'second~csv': ['C0000239'],