feat: add table neighbors (#552)

bd_api/apps/api/v1/models.py

@@ -472,7 +472,6 @@ def has_picture(self):
     has_picture.short_description = "Has Picture"
     has_picture.boolean = True
 
-    @property
     def get_graphql_has_picture(self):
         """Get the has_picture property for graphql"""
         return self.has_picture()
@@ -485,7 +484,6 @@ def full_slug(self):
 
         return f"{self.slug}"
 
-    @property
     def get_graphql_full_slug(self):
         """Get the full slug or Organization for graphql"""
         return self.full_slug
@@ -573,7 +571,6 @@ def full_slug(self):
             return f"{self.organization.area.slug}_{self.organization.slug}_{self.slug}"
         return f"{self.organization.slug}_{self.slug}"
 
-    @property
     def get_graphql_full_slug(self):
         """Get the full slug or Dataset for graphql"""
         return self.full_slug
@@ -697,7 +694,6 @@ def coverage(self):
 
         return coverage_str
 
-    @property
     def get_graphql_coverage(self):
         """
         Returns the temporal coverage of the dataset in the format
@@ -726,7 +722,6 @@ def full_coverage(self) -> str:
         ]
         return json.dumps(full_coverage_dict)
 
-    @property
     def get_graphql_full_coverage(self):
         return self.full_coverage
 
@@ -735,7 +730,6 @@ def contains_tables(self):
         """Returns true if there are tables in the dataset"""
         return len(self.tables.all()) > 0
 
-    @property
     def get_graphql_contains_tables(self):
         """Returns true if there are tables in the dataset for graphql"""
         return self.contains_tables
@@ -757,7 +751,6 @@ def contains_closed_data(self):
 
         return closed_data
 
-    @property
     def get_graphql_contains_closed_data(self):
         """Returns true if there are tables or columns with closed coverages for graphql"""
         return self.contains_closed_data
@@ -775,7 +768,6 @@ def contains_open_data(self):
 
         return open_data
 
-    @property
     def get_graphql_contains_open_data(self):
         """Returns true if there are tables or columns with open coverages for graphql"""
         return self.contains_open_data
@@ -786,7 +778,6 @@ def contains_closed_tables(self):
         closed_tables = self.tables.all().filter(is_closed=True)
         return len(closed_tables) > 0
 
-    @property
     def get_graphql_contains_closed_tables(self):
         """Returns true if there are tables with closed coverages for graphql (DEPRECATED)"""
         return self.contains_closed_tables
@@ -797,7 +788,6 @@ def contains_open_tables(self):
         open_tables = self.tables.all().filter(is_closed=False)
         return len(open_tables) > 0
 
-    @property
     def get_graphql_contains_open_tables(self):
         """Returns true if there are tables with open coverages for graphql (DEPRECATED)"""
         return self.contains_open_tables
@@ -807,7 +797,6 @@ def contains_raw_data_sources(self):
         """Returns true if there are raw data sources in the dataset"""
         return len(self.raw_data_sources.all()) > 0
 
-    @property
     def get_graphql_contains_raw_data_sources(self):
         """Returns true if there are raw data sources in the dataset for graphql"""
         return self.contains_raw_data_sources
@@ -817,7 +806,6 @@ def contains_information_requests(self):
         """Returns true if there are information requests in the dataset"""
         return len(self.information_requests.all()) > 0
 
-    @property
     def get_graphql_contains_information_requests(self):
         """Returns true if there are information requests in the dataset for graphql"""
         return self.contains_information_requests
@@ -981,6 +969,9 @@ def gbq_slug(self):
             table = cloud_table.gcp_table_id
             return f"basedosdados.{dataset}.{table}"
 
+    def get_graphql_gbq_slug(self):
+        return self.gbq_slug
+
     @property
     def gcs_slug(self):
         """Get the slug used in Google Cloud Storage"""
@@ -989,11 +980,6 @@ def gcs_slug(self):
             table = cloud_table.gcp_table_id
             return f"staging/{dataset}/{table}"
 
-    @property
-    def get_graphql_gbq_slug(self):
-        return self.gbq_slug
-
-    @property
     def get_graphql_gcs_slug(self):
         return self.gcs_slug
 
@@ -1003,7 +989,6 @@ def partitions(self):
         partitions_list = [p.name for p in self.columns.all().filter(is_partition=True)]
         return ", ".join(partitions_list)
 
-    @property
     def get_graphql_partitions(self):
         """Returns a list of columns used to partition the table"""
         return self.partitions
@@ -1022,7 +1007,6 @@ def contains_closed_data(self):
 
         return closed_data
 
-    @property
     def get_graphql_contains_closed_data(self):
         """Returns true if there are columns with closed coverages to be used in graphql"""
         return self.contains_closed_data
@@ -1105,10 +1089,34 @@ def full_coverage(self) -> str:
 
         return json.dumps(full_coverage)
 
-    @property
     def get_graphql_full_coverage(self):
         return self.full_coverage
 
+    @property
+    def neighbors(self):
+        """Similiar tables and columns
+        - Tables and columns with similar directories
+        - Tables and columns with similar coverages or tags (WIP)
+        """
+        all_neighbors = []
+        for column in self.columns.all():
+            for neighbor in column.neighbors:
+                all_neighbors.append((neighbor.table, neighbor))
+        return all_neighbors
+
+    def get_graphql_neighbors(self) -> list[dict]:
+        all_neighbors = []
+        for table, column in self.neighbors:
+            all_neighbors.append(
+                {
+                    "table_id": str(table.id),
+                    "table_name": table.name,
+                    "column_id": str(column.id),
+                    "column_name": column.name,
+                }
+            )
+        return all_neighbors
+
     def clean(self):
         """
         Clean method for Table model
@@ -1298,10 +1306,34 @@ def full_coverage(self) -> str:
 
         return json.dumps(column_full_coverage)
 
-    @property
     def get_graphql_full_coverage(self):
         return self.full_coverage
 
+    @property
+    def neighbors(self):
+        """Similiar tables and columns
+        - Columns with similar directories
+        - Columns with similar coverages or tags (WIP)
+        """
+        if not self.directory_primary_key:
+            return []
+        return (
+            Column.objects.filter(directory_primary_key=self.directory_primary_key)
+            .exclude(id=self.id)
+            .all()
+        )
+
+    def get_graphql_neighbors(self) -> list[dict]:
+        all_neighbors = []
+        for column in self.neighbors:
+            all_neighbors.append(
+                {
+                    "column_id": str(column.id),
+                    "column_name": column.name,
+                }
+            )
+        return all_neighbors
+
 
 class ColumnOriginalName(BaseModel):
     """Model definition for ColumnOriginalName."""

bd_api/custom/graphql_auto.py

@@ -29,6 +29,7 @@
     Schema,
     String,
 )
+from graphene.types.generic import GenericScalar
 from graphene.types.utils import yank_fields_from_attrs
 from graphene_django import DjangoObjectType
 from graphene_django.converter import convert_django_field
@@ -391,6 +392,8 @@ def match_type(model, attr):
                 return partial(List, of_type=Int)
             if "str" in get_type(model, attr):
                 return partial(List, of_type=String)
+        if "dict" in get_type(model, attr):
+            return GenericScalar
         return String
 
     def build_custom_attrs(model, attrs):