Add tests for model save load and add support for none data

icat/anchorlist.py

@@ -652,10 +652,15 @@ def highlight_regex(self) -> str:
         return kw_regex
 
     def build_tfidf_features(self):
+        # TODO: this eventually needs to use the anchorlist cache, and also
+        # this may need to change when tfidf just becomes a sim function instead
+        # of a dedicated anchor
         if self.model is None:
             raise RuntimeError(
                 "The anchorlist has no associated model to get a dataset from."
             )
+        if self.model.data.active_data is None:
+            return
 
         self.tfidf_vectorizer = TfidfVectorizer(stop_words="english")
         # TODO: coupling: accessing active_data through model

icat/histograms.py

@@ -100,6 +100,8 @@ def _set_layout(self):
     # ============================================================
 
     def refresh_data(self, data: DataManager):
+        if data.active_data is None:
+            return
         if data.prediction_col in data.active_data.columns:
             self.hist_local.set_data(
                 data.active_data.loc[

icat/instance.py

@@ -153,6 +153,9 @@ def populate(self, index: int):
         """Fill or update all of the fields for the given index. This
         should be called anytime the model updates, or when the user
         clicks/requests to view a new instance."""
+        if self.data.active_data is None:
+            return
+
         self.index = index
         self.index_display.children = [str(index)]
         row = self.data.active_data.iloc[index]

icat/view.py

@@ -153,6 +153,9 @@ def _send_anchorlist_anchor_modification_to_viz(
     def _serialize_data_to_dicts(self) -> dict:
         feature_names = self.model.feature_names()
 
+        if self.model.data.active_data is None:
+            return []
+
         # first we need to see if we are filtering data based on prediction range
         # we reference the sample indices a lot, so just compute this once and reference
         # throughout.

tests/conftest.py

@@ -64,5 +64,5 @@ def fun_df():
 def data_file_loc():
     shutil.rmtree("test/exampledata", ignore_errors=True)
     os.makedirs("test/exampledata", exist_ok=True)
-    yield
+    yield "test/exampledata/thing"
     shutil.rmtree("test/exampledata", ignore_errors=True)

tests/test_anchorlist.py

@@ -169,7 +169,9 @@ def test_highlight_regex(kw_set1, kw_set2, expected_result):
     """Special characters should be escaped, esp *"""
 
 
-def test_save_load(data_file_loc, fun_df):
+def test_save_load_anchorlist(data_file_loc, fun_df):
+    """When we save an anchorlist and then reload it, all the anchors should reload into
+    the same spot with the same parameters."""
     model = Model(fun_df, "text")
     a1 = DictionaryAnchor(anchor_name="thing1")
     a1.keywords = ["hello", "there"]
@@ -183,10 +185,10 @@ def test_save_load(data_file_loc, fun_df):
     model.add_anchor(a1)
     model.add_anchor(a2)
 
-    model.anchor_list.save("test/exampledata/thing")
+    model.anchor_list.save(data_file_loc)
 
     model2 = Model(fun_df, "text")
-    model2.anchor_list.load("test/exampledata/thing")
+    model2.anchor_list.load(data_file_loc)
     assert model2.anchor_list.cache["test_cache"] == 13
 
     a21 = model2.anchor_list.anchors[0]

tests/test_anchors.py

@@ -380,10 +380,10 @@ def test_save_load_similarity_anchor(data_file_loc):
     a1.text_col = "my_text"
     a1.reference_texts = ["I am a powerful potato"]
     a1.reference_short = ["I am a"]
-    a1.save("test/exampledata/test")
+    a1.save(data_file_loc)
 
     a2 = SimilarityFunctionAnchor()
-    a2.load("test/exampledata/test")
+    a2.load(data_file_loc)
     assert a2.anchor_name == "I am an anchor"
     assert a2.weight == 1.2
     assert not a2.in_view

tests/test_model.py

@@ -280,3 +280,19 @@ def test_changing_anchor_name_twice_before_model_trained_modifies_data(
     model.anchor_list.anchors[0]._anchor_name_input.fire_event("blur", "testing123")
     assert "_testing123" in model.data.active_data.columns
     assert "_testing" not in model.data.active_data.columns
+
+
+def test_save_load_model(data_file_loc, fun_df, dummy_anchor):
+    """Saving a model and then reloading it should load in all the same data and anchors."""
+
+    model = Model(fun_df, text_col="text")
+    model.anchor_list.add_anchor(dummy_anchor)
+    model.data.apply_label(0, 1)
+    model.save(data_file_loc)
+
+    model2 = Model(None, text_col="")
+    model2.load(data_file_loc)
+    assert len(model2.data.active_data) == 12
+    assert model2.text_col == "text"
+    assert len(model2.anchor_list.anchors) == 1
+    assert len(model2.training_data) == 1