From 7045e11aa03c9425a0522f48c2631ddef22a4ccc Mon Sep 17 00:00:00 2001
From: Tobie Tusing <tobie.tusing@kalderos.com>
Date: Fri, 7 Apr 2023 09:09:58 -0700
Subject: [PATCH] Remove trivial nodes before building subdag (#7194)

* remove trial nodes before building subdag

* add changie

* Update graph.py

remove comment

* further optimize by sorting node search by degree

* change degree to product of in and out degree
---
 .../Under the Hood-20230319-172824.yaml       |  6 ++++++
 core/dbt/graph/graph.py                       | 21 ++++++++++++++++++-
 2 files changed, 26 insertions(+), 1 deletion(-)
 create mode 100644 .changes/unreleased/Under the Hood-20230319-172824.yaml

diff --git a/.changes/unreleased/Under the Hood-20230319-172824.yaml b/.changes/unreleased/Under the Hood-20230319-172824.yaml
new file mode 100644
index 00000000000..06722f57ae6
--- /dev/null
+++ b/.changes/unreleased/Under the Hood-20230319-172824.yaml	
@@ -0,0 +1,6 @@
+kind: Under the Hood
+body: Improves build times for common selections by improving subgraph calculation
+time: 2023-03-19T17:28:24.321555-07:00
+custom:
+  Author: '@ttusing'
+  Issue: "7195"
diff --git a/core/dbt/graph/graph.py b/core/dbt/graph/graph.py
index a5b5e91456e..29f24cae734 100644
--- a/core/dbt/graph/graph.py
+++ b/core/dbt/graph/graph.py
@@ -74,7 +74,26 @@ def get_subset_graph(self, selected: Iterable[UniqueId]) -> "Graph":
         new_graph = self.graph.copy()
         include_nodes = set(selected)
 
-        for node in self:
+        still_removing = True
+        while still_removing:
+            nodes_to_remove = list(
+                node
+                for node in new_graph
+                if node not in include_nodes
+                and (new_graph.in_degree(node) * new_graph.out_degree(node)) == 0
+            )
+            if len(nodes_to_remove) == 0:
+                still_removing = False
+            else:
+                new_graph.remove_nodes_from(nodes_to_remove)
+
+        # sort remaining nodes by degree
+        remaining_nodes = list(new_graph.nodes())
+        remaining_nodes.sort(
+            key=lambda node: new_graph.in_degree(node) * new_graph.out_degree(node)
+        )
+
+        for node in remaining_nodes:
             if node not in include_nodes:
                 source_nodes = [x for x, _ in new_graph.in_edges(node)]
                 target_nodes = [x for _, x in new_graph.out_edges(node)]