Add file uri to _extra part of variant objects

(for linking purposes)

bento_variant_service/search.py

@@ -54,6 +54,9 @@ def search_worker_prime(
         try:
             variant = next(possible_matches)
 
+            # TODO: Do we use as_chord_representation or as_augmented_chord_representation here?
+            #  Maybe not augmented, since we won't allow querying augmented stuff.
+
             match = rest_of_query is None or check_ast_against_data_structure(
                 rest_of_query, variant.as_chord_representation(), VARIANT_SCHEMA)
             found = found or match
@@ -62,7 +65,7 @@ def search_worker_prime(
                 break
 
             if match:  # implicitly internal_data is True here as well
-                matches.append(variant.as_chord_representation())
+                matches.append(variant.as_augmented_chord_representation())
 
         except StopIteration:
             break

bento_variant_service/tables/vcf/table.py

@@ -152,6 +152,7 @@ def _variants(
                         start_pos=int(row[1]),
                         ref_bases=row[3],
                         alt_bases=tuple(row[4].split(",")),
+                        file_uri=vcf.original_index_uri,
                     )
 
                     variant.calls = tuple(VCFVariantTable._variant_calls(variant, vcf.sample_ids, row,

bento_variant_service/variants/models.py

@@ -14,14 +14,16 @@ class Variant:
     """
 
     def __init__(self, assembly_id: str, chromosome: str, ref_bases: str, alt_bases: Tuple[str, ...], start_pos: int,
-                 calls: Tuple["Call"] = ()):
+                 calls: Tuple["Call"] = (), file_uri: Optional[str] = None):
         self.assembly_id: str = assembly_id  # Assembly ID for context
         self.chromosome: str = chromosome  # Chromosome where the variant occurs
         self.ref_bases: str = ref_bases  # Reference bases
         self.alt_bases: Tuple[str] = alt_bases  # Alternate bases  TODO: Structural variants
         self.start_pos: int = start_pos  # Starting position on the chromosome w/r/t the reference, 0-indexed
         self.calls: Tuple["Call"] = calls  # Variant calls, per sample  TODO: Make this a dict?
 
+        self.file_uri: Optional[str] = file_uri  # File URI, "
+
     @property
     def end_pos(self) -> int:
         """
@@ -40,6 +42,14 @@ def as_chord_representation(self):
             "calls": [c.as_chord_representation() for c in self.calls],
         }
 
+    def as_augmented_chord_representation(self):
+        return {
+            **self.as_chord_representation(),
+            # _ prefix is context dependent -> immune from equality, used by Bento in weird contexts
+            "_extra": {
+                "file_uri": self.file_uri,
+            },
+        }
 
     def __eq__(self, other):
         # Use and shortcutting to return False early if the other instance isn't a Variant

bento_variant_service/variants/schemas.py

@@ -193,9 +193,31 @@
                 "required": False,
                 "type": "unlimited",
                 "order": 6,
-            }
-        }
-    }
+            },
+        },
+        "_extra": {
+            "type": "object",
+            "properties": {
+                "file_uri": {
+                    "type": "string",
+                    "search": {
+                        "operations": [op.SEARCH_OP_EQ],
+                        "queryable": "internal",
+                        "canNegate": True,
+                        "required": False,
+                        "type": "single",  # single / unlimited
+                        "order": 0,
+                    },
+                },
+            },
+            "search": {
+                "queryable": "internal",
+                "required": False,
+                "type": "unlimited",
+                "order": 8,
+            },
+        },
+    },
 }
 
 VARIANT_TABLE_METADATA_SCHEMA = {

tests/test_search.py

@@ -170,8 +170,11 @@ def test_chord_variant_search(app, client, table_manager):
             assert "results" in data
             assert "fixed_id" in data["results"]
             assert len(data["results"]["fixed_id"]["matches"]) == 3
-            assert json.dumps(data["results"]["fixed_id"]["matches"][0], sort_keys=True) == \
-                json.dumps(VARIANT_1.as_chord_representation(), sort_keys=True)
+            first_match = data["results"]["fixed_id"]["matches"][0]
+            assert json.dumps(first_match, sort_keys=True) == \
+                json.dumps(VARIANT_1.as_augmented_chord_representation(), sort_keys=True)
+            assert json.dumps(dict((k, v) for k, v in first_match.items() if not k.startswith("_")),
+                              sort_keys=True) == json.dumps(VARIANT_1.as_chord_representation(), sort_keys=True)
 
             # Test private table search
 
@@ -194,8 +197,8 @@ def test_chord_variant_search(app, client, table_manager):
             assert "results" in data
 
             assert len(data["results"]) == 3
-            assert json.dumps(data["results"][0], sort_keys=True) == json.dumps(VARIANT_1.as_chord_representation(),
-                                                                                sort_keys=True)
+            assert json.dumps(data["results"][0], sort_keys=True) == \
+                json.dumps(VARIANT_1.as_augmented_chord_representation(), sort_keys=True)
 
             for q, r in TEST_PRIVATE_QUERIES:
                 qj = {"query": q}

tests/test_vcf_file.py

@@ -7,6 +7,7 @@
 def test_vcf_file():
     file = VCFFile(VCF_ONE_VAR_FILE_URI)
     assert file.path == os.path.realpath(VCF_ONE_VAR_FILE_PATH)
+    assert file.original_uri == VCF_ONE_VAR_FILE_URI
     assert file.index_path is None
     assert file.assembly_id == "GRCh37"
     assert len(file.sample_ids) == 835