Feat: refactor largest connected subgraph (#20)

* rename macro to largest-connected-subgraph

* make the largest_connected_subgraph macro make a table that is vertex focused rather than a graph

* fix unit tests after refactor

* fix readme error for largest connected subgraph

README.md

@@ -196,7 +196,7 @@ from the integration_tests folder. sqlfluff does not currently support macros, m
   - [graph_is_connected](#graph_is_connected)
 
 **[Macros](#macros)**
-  - [largest_connected_subgraph_identifier](#largest_connected_subgraph_identifier)
+  - [largest_connected_subgraph](#largest_connected_subgraph)
 
 **[Helper Macros](#helper-macros)**
   - [array_agg](#array_agg)
@@ -238,7 +238,7 @@ flowchart
 ```
 
 ## Macros
-### [largest_connected_subgraph_identifier](macros/largest_connected_subgraph_identifier.sql)
+### [largest_connected_subgraph](macros/largest_connected_subgraph.sql)
 
 Arguments:
 - input: the input node (inputted as `ref(...)` or `source(...)`) or CTE (inputted as a string)
@@ -250,7 +250,7 @@ Arguments:
 **Usage:**
 ```sql
 with subgraphs as (
-    {{ dbt_graph_theory.largest_connected_subgraph_identifier(
+    {{ dbt_graph_theory.largest_connected_subgraph(
         input=ref('example_model'),
         edge_id='edge_id_field_name',
         vertex_1='vertex_1_field_name',
@@ -264,7 +264,7 @@ with subgraphs as (
 ```sql
 ...
 subgraphs as (
-    {{ dbt_graph_theory.largest_connected_subgraph_identifier(
+    {{ dbt_graph_theory.largest_connected_subgraph(
         input='example_cte',
         edge_id='different_edge_id_field_name',
         vertex_1='different_vertex_1_field_name',
@@ -290,16 +290,17 @@ flowchart
 
 The following table is returned:
 
-| edge_id | vertex_1 | vertex_2 | subgraph_id |  subgraph_members  |
-|:-------:|:--------:|:--------:|:-----------:|:------------------:|
-|    1    |     A    |     B    |      1      |  ['A', 'B', 'C']   |
-|    2    |     B    |     C    |      1      |  ['A', 'B', 'C']   |
-|    3    |     D    |     E    |      2      |['D', 'E', 'F', 'G']|
-|    4    |     E    |     F    |      2      |['D', 'E', 'F', 'G']|
-|    5    |     D    |     F    |      2      |['D', 'E', 'F', 'G']|
-|    6    |     E    |     G    |      2      |['D', 'E', 'F', 'G']|
-
-subgraph_id is designed to be unique to both the graph and subgraph level. 
+|  vertex  | subgraph_id |  subgraph_members  |
+|:--------:|:-----------:|:------------------:|
+|     A    |      1      |  ['A', 'B', 'C']   |
+|     B    |      1      |  ['A', 'B', 'C']   |
+|     C    |      2      |  ['A', 'B', 'C']   |
+|     D    |      2      |['D', 'E', 'F', 'G']|
+|     E    |      2      |['D', 'E', 'F', 'G']|
+|     F    |      2      |['D', 'E', 'F', 'G']|
+|     G    |      2      |['D', 'E', 'F', 'G']|
+
+subgraph_id is designed to be unique to both the graph and subgraph level. When graph_id is defined, the output is also at a graph_id level.
 
 ## Helper Macros
 Note that the below are designed for internal (ie. dbt-graph-theory) use only. Use them at your own risk!

integration_tests/models/test_largest_connected_subgraph_identifier/test_largest_connected_subgraph_identifier.yml → integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph.yml

@@ -1,41 +1,41 @@
 version: 2
 models:
-  - name: test_largest_conn_subgraph_1_subgraph
+  - name: test_largest_connected_subgraph_1_subgraph
     description: Unit test checking a situation with 1 subgraph and no graph_id.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:
           value: 0
-  - name: test_largest_conn_subgraph_4_subgraph
+  - name: test_largest_connected_subgraph_4_subgraph
     description: Unit test checking a situation with 4 subgraphs and no graph_id.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:
           value: 0
-  - name: test_largest_conn_subgraph_2_subgraph_no_edge
+  - name: test_largest_connected_subgraph_2_subgraph_no_edge
     description: Unit test checking a situation with 1 subgraph, no connecting edges and no graph_id.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:
           value: 0
-  - name: test_largest_conn_subgraph_3_subgraph_no_edge
+  - name: test_largest_connected_subgraph_3_subgraph_no_edge
     description: Unit test checking a situation with 3 subgraphs, no connecting edges and no graph_id.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:
           value: 0
-  - name: test_largest_conn_subgraph_graph_id
+  - name: test_largest_connected_subgraph_graph_id
     description: Unit test checking a situation with 2 graph ids, one with two subgraphs and one with 1 subgraph.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:
           value: 0
-  - name: test_largest_conn_subgraph_graph_id_3_subgraph
+  - name: test_largest_connected_subgraph_graph_id_3_subgraph
     description: Unit test checking a situation with 2 graph ids, with three subgraphs each.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:
           value: 0
-  - name: test_largest_conn_subgraph_no_data
+  - name: test_largest_connected_subgraph_no_data
     description: Unit test checking a situation no data.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:
           value: 0
-  - name: test_largest_conn_subgraph_no_data_graph_id
+  - name: test_largest_connected_subgraph_no_data_graph_id
     description: Unit test checking a situation no data but with graph_id defined.
     tests:
       - dbt_expectations.expect_table_row_count_to_equal:

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_1_subgraph.sql

@@ -0,0 +1,22 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_1_subgraph_data')
+    ) }}
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        ('A', '1', array['A', 'B', 'C', 'D', 'E']),
+        ('B', '1', array['A', 'B', 'C', 'D', 'E']),
+        ('C', '1', array['A', 'B', 'C', 'D', 'E']),
+        ('D', '1', array['A', 'B', 'C', 'D', 'E']),
+        ('E', '1', array['A', 'B', 'C', 'D', 'E'])
+    ) as v (vertex, subgraph_id, subgraph_members)
+)
+
+select * from {{ cte_difference(
+    'computed',
+    'subgraph_members',
+    fields=["vertex", "subgraph_id", "subgraph_members"]
+) }}

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_2_subgraph_no_edge.sql

@@ -0,0 +1,29 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_2_subgraph_no_edge_data')
+    ) }}
+),
+
+-- recast because vertex_2 is all null in seed data, interpreted as int dtype
+recast_computed as (
+    select
+        vertex::text as vertex,
+        subgraph_id,
+        subgraph_members
+    from
+        computed
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        ('A', '1', array['A']),
+        ('B', '2', array['B'])
+    ) as v (vertex, subgraph_id, subgraph_members)
+)
+
+select * from {{ cte_difference(
+    'recast_computed',
+    'subgraph_members',
+    fields=["vertex", "subgraph_id", "subgraph_members"]
+) }}

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_3_subgraph_no_edge.sql

@@ -0,0 +1,20 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_3_subgraph_no_edge_data')
+    ) }}
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        ('A', '1', array['A']),
+        ('B', '2', array['B']),
+        ('C', '3', array['C'])
+    ) as v (vertex, subgraph_id, subgraph_members)
+)
+
+select * from {{ cte_difference(
+    'computed',
+    'subgraph_members',
+    fields=["vertex", "subgraph_id", "subgraph_members"]
+) }}

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_4_subgraph.sql

@@ -0,0 +1,26 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_4_subgraph_data')
+    ) }}
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        ('A', '1', array['A', 'B', 'C', 'D']),
+        ('B', '1', array['A', 'B', 'C', 'D']),
+        ('C', '1', array['A', 'B', 'C', 'D']),
+        ('D', '1', array['A', 'B', 'C', 'D']),
+        ('E', '2', array['E', 'F']),
+        ('F', '2', array['E', 'F']),
+        ('G', '3', array['G']),
+        ('H', '4', array['H', 'I']),
+        ('I', '4', array['H', 'I'])
+    ) as v (vertex, subgraph_id, subgraph_members)
+)
+
+select * from {{ cte_difference(
+    'computed',
+    'subgraph_members',
+    fields=["vertex", "subgraph_id", "subgraph_members"]
+) }}

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_graph_id.sql

@@ -0,0 +1,27 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_graph_id_data'),
+        graph_id='graph_id'
+    ) }}
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        (1, 'A', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'B', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'C', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'D', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'E', '1__2', array['E']),
+        (2, 'A', '2__1', array['A', 'B', 'C', 'D']),
+        (2, 'B', '2__1', array['A', 'B', 'C', 'D']),
+        (2, 'C', '2__1', array['A', 'B', 'C', 'D']),
+        (2, 'D', '2__1', array['A', 'B', 'C', 'D'])
+    ) as v (graph_id, vertex, subgraph_id, subgraph_members)
+)
+
+select * from {{ cte_difference(
+    'computed',
+    'subgraph_members',
+    fields=["graph_id", "vertex", "subgraph_id", "subgraph_members"]
+) }}

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_graph_id_3_subgraph.sql

@@ -0,0 +1,36 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_graph_id_3_subgraph_data'),
+        graph_id='graph_id'
+    ) }}
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        (1, 'A', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'B', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'C', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'D', '1__1', array['A', 'B', 'C', 'D']),
+        (1, 'E', '1__2', array['E', 'F', 'G', 'H']),
+        (1, 'F', '1__2', array['E', 'F', 'G', 'H']),
+        (1, 'G', '1__2', array['E', 'F', 'G', 'H']),
+        (1, 'H', '1__2', array['E', 'F', 'G', 'H']),
+        (1, 'I', '1__3', array['I', 'J']),
+        (1, 'J', '1__3', array['I', 'J']),
+        (2, 'A', '2__1', array['A', 'B', 'C']),
+        (2, 'B', '2__1', array['A', 'B', 'C']),
+        (2, 'C', '2__1', array['A', 'B', 'C']),
+        (2, 'D', '2__2', array['D']),
+        (2, 'E', '2__3', array['E', 'F', 'G', 'H']),
+        (2, 'F', '2__3', array['E', 'F', 'G', 'H']),
+        (2, 'G', '2__3', array['E', 'F', 'G', 'H']),
+        (2, 'H', '2__3', array['E', 'F', 'G', 'H'])
+    ) as v (graph_id, vertex, subgraph_id, subgraph_members)
+)
+
+select * from {{ cte_difference(
+    'computed',
+    'subgraph_members',
+    fields=["graph_id", "vertex", "subgraph_id", "subgraph_members"]
+) }}

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_no_data.sql

@@ -0,0 +1,19 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_no_data_data')
+    ) }}
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        (null::text, null::text, array[null])
+    ) as v (vertex, subgraph_id, subgraph_members)
+    where false
+)
+
+select * from {{ cte_difference(
+    'computed',
+    'subgraph_members',
+    fields=["vertex", "subgraph_id", "subgraph_members"]
+) }}

integration_tests/models/test_largest_connected_subgraph/test_largest_connected_subgraph_no_data_graph_id.sql

@@ -0,0 +1,20 @@
+with computed as (
+    {{ dbt_graph_theory.largest_connected_largest_connected_subgraph(
+        input=ref('test_largest_connected_subgraph_no_data_graph_id_data'),
+        graph_id='graph_id'
+    ) }}
+),
+
+subgraph_members as (
+    select v.* from (
+        values
+        (null::integer, null::text, null::text, array[null])
+    ) as v (graph_id, vertex, subgraph_id, subgraph_members)
+    where false
+)
+
+select * from {{ cte_difference(
+    'computed',
+    'subgraph_members',
+    fields=["graph_id", "vertex", "subgraph_id", "subgraph_members"]
+) }}