From 38cdbb77e0127ca5a8cf4bb10f78533dfb75e9f9 Mon Sep 17 00:00:00 2001
From: popcorny <celu@infuseai.io>
Date: Wed, 12 Jun 2024 16:33:29 +0800
Subject: [PATCH] The recce summary would raise exception if node is excluded
 from the lineage object

Signed-off-by: popcorny <celu@infuseai.io>
---
 recce/adapter/dbt_adapter/__init__.py |  4 ++--
 recce/summary.py                      | 12 ++++++++++--
 2 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/recce/adapter/dbt_adapter/__init__.py b/recce/adapter/dbt_adapter/__init__.py
index d975e701..3278a304 100644
--- a/recce/adapter/dbt_adapter/__init__.py
+++ b/recce/adapter/dbt_adapter/__init__.py
@@ -392,8 +392,6 @@ def get_lineage(self, base: Optional[bool] = False):
 
         manifest_dict = manifest.to_dict()
 
-        parent_map = {k: v for k, v in manifest_dict['parent_map'].items() if not k.startswith('test.')}
-
         nodes = {}
 
         for node in manifest_dict['nodes'].values():
@@ -492,6 +490,8 @@ def get_lineage(self, base: Optional[bool] = False):
                     'config': semantic_models['config'],
                 }
 
+        nodeIds = nodes.keys()
+        parent_map = {k: v for k, v in manifest_dict['parent_map'].items() if k in nodeIds}
         return dict(
             parent_map=parent_map,
             nodes=nodes,
diff --git a/recce/summary.py b/recce/summary.py
index fb17dc8a..d92fcde8 100644
--- a/recce/summary.py
+++ b/recce/summary.py
@@ -1,3 +1,4 @@
+import sys
 from typing import List, Dict, Set, Union, Type, Optional
 from uuid import UUID
 
@@ -18,6 +19,11 @@
 REMOVE_COLOR = '#ff067e'
 
 
+def _warn(msg):
+    # print to stderr
+    print(f"warning: {msg}", file=sys.stderr)
+
+
 class Node:
     id: str
     name: str
@@ -219,9 +225,11 @@ def create_node(self, node_id: str, node_data: dict, data_from: str = 'base'):
 
     def create_edge(self, parent_id: str, child_id: str, edge_from: str = 'base'):
         if parent_id not in self.nodes:
-            raise ValueError(f'Parent node {parent_id} not found in graph')
+            _warn(f'Parent node {parent_id} not found in graph')
+            return
         if child_id not in self.nodes:
-            raise ValueError(f'Child node {child_id} not found in graph')
+            _warn(f'Child node {child_id} not found in graph')
+            return
 
         edge_id = f'{parent_id}-->{child_id}'
         if edge_id in self.edges: