Filter prefixes

sssom/util.py

@@ -149,6 +149,28 @@ def clean_prefix_map(self) -> None:
             self.df = filter_out_prefixes(self.df, missing_prefixes)
         self.prefix_map = new_prefixes
 
+    def remove_mappings(self, msdf: "MappingSetDataFrame"):
+        """Remove mappings in right msdf from left msdf.
+
+        :param msdf: MappingSetDataframe object to be removed from primary msdf object.
+        """
+        self.df = (
+            pd.merge(
+                self.df,
+                msdf.df,
+                on=KEY_FEATURES,
+                how="outer",
+                suffixes=("", "_2"),
+                indicator=True,
+            )
+            .query("_merge == 'left_only'")
+            .drop("_merge", axis=1)
+            .reset_index(drop=True)
+        )
+
+        self.df = self.df[self.df.columns.drop(list(self.df.filter(regex=r"_2")))]
+        self.clean_prefix_map()
+
 
 @dataclass
 class EntityPair:
@@ -990,27 +1012,54 @@ def get_prefixes_used_in_table(df: pd.DataFrame) -> List[str]:
     return list(set(prefixes))
 
 
-def filter_out_prefixes(df: pd.DataFrame, filter_prefixes: List[str]) -> pd.DataFrame:
+def filter_out_prefixes(
+    df: pd.DataFrame, filter_prefixes: List[str], features: list = KEY_FEATURES
+) -> pd.DataFrame:
     """Filter any row where a CURIE in one of the key column uses one of the given prefixes.
 
     :param df: Pandas DataFrame
     :param filter_prefixes: List of prefixes
+    :param features: List of dataframe column names dataframe to consider
     :return: Pandas Dataframe
     """
     filter_prefix_set = set(filter_prefixes)
     rows = []
 
     for _, row in df.iterrows():
-        # Get list of CURIEs from the 3 columns (KEY_FEATURES) for the row.
-        prefixes = {get_prefix_from_curie(curie) for curie in row[KEY_FEATURES]}
-        # Confirm if none of the 3 CURIEs in the list above appear in the filter_prefixes list.
+        prefixes = {get_prefix_from_curie(curie) for curie in row[features]}
+        # Confirm if none of the CURIEs in the list above appear in the filter_prefixes list.
         # If TRUE, append row.
         if not any(prefix in prefixes for prefix in filter_prefix_set):
             rows.append(row)
     if rows:
         return pd.DataFrame(rows)
     else:
-        return pd.DataFrame(columns=KEY_FEATURES)
+        return pd.DataFrame(columns=features)
+
+
+def filter_prefixes(
+    df: pd.DataFrame, filter_prefixes: List[str], features: list = KEY_FEATURES
+) -> pd.DataFrame:
+    """Filter any row where a CURIE in one of the key column uses one of the given prefixes.
+
+    :param df: Pandas DataFrame
+    :param filter_prefixes: List of prefixes
+    :param features: List of dataframe column names dataframe to consider
+    :return: Pandas Dataframe
+    """
+    filter_prefix_set = set(filter_prefixes)
+    rows = []
+
+    for _, row in df.iterrows():
+        prefixes = {get_prefix_from_curie(curie) for curie in row[features]}
+        # Confirm if all of the CURIEs in the list above appear in the filter_prefixes list.
+        # If TRUE, append row.
+        if all(prefix in filter_prefix_set for prefix in prefixes):
+            rows.append(row)
+    if rows:
+        return pd.DataFrame(rows)
+    else:
+        return pd.DataFrame(columns=features)
 
 
 # TODO this is not used anywhere

tests/test_utils.py

@@ -1,13 +1,21 @@
 """Test for merging MappingSetDataFrames."""
 import unittest
 
+from sssom.constants import OBJECT_ID, SUBJECT_ID
 from sssom.io import extract_iri
+from sssom.parsers import parse_sssom_table
+from sssom.util import MappingSetDataFrame, filter_out_prefixes, filter_prefixes
 from tests.constants import data_dir
 
 
 class TestIO(unittest.TestCase):
     """A test case for merging msdfs."""
 
+    def setUp(self) -> None:
+        """Set up."""
+        self.msdf = parse_sssom_table(f"{data_dir}/basic.tsv")
+        self.features = [SUBJECT_ID, OBJECT_ID]
+
     def test_broken_predicate_list(self):
         """Test merging of multiple msdfs."""
         pred_filter_list = ["skos:relatedMatch", f"{data_dir}/predicate_list3.txt"]
@@ -18,3 +26,31 @@ def test_broken_predicate_list(self):
             if p_iri:
                 iri_list.extend(p_iri)
         self.assertEqual(3, len(iri_list))
+
+    def test_filter_prefixes(self):
+        """Test filtering MSDF.df by prefixes provided."""
+        prefix_filter_list = ["x", "y"]
+        filtered_df = filter_prefixes(self.msdf.df, prefix_filter_list, self.features)
+        self.assertEqual(len(filtered_df), 40)
+
+    def test_filter_out_prefixes(self):
+        """Test filtering MSDF.df by prefixes provided."""
+        prefix_filter_list = ["x", "y"]
+        filtered_df = filter_out_prefixes(
+            self.msdf.df, prefix_filter_list, self.features
+        )
+        self.assertEqual(len(filtered_df), 5)
+
+    def test_remove_mappings(self):
+        """Test remove mappings."""
+        prefix_filter_list = ["x", "y"]
+        filtered_df = filter_out_prefixes(
+            self.msdf.df, prefix_filter_list, self.features
+        )
+        new_msdf = MappingSetDataFrame(
+            df=filtered_df, prefix_map=self.msdf.prefix_map, metadata=self.msdf.metadata
+        )
+        original_length = len(self.msdf.df)
+        self.msdf.remove_mappings(new_msdf)
+        # len(self.msdf.df) = 141 and len(new_msdf.df) = 5
+        self.assertEqual(len(self.msdf.df), original_length - len(new_msdf.df))