From 3e2f9438d3f6236f890c722205d94d72e2ababeb Mon Sep 17 00:00:00 2001
From: tarikceric <46071768+tarikceric@users.noreply.github.com>
Date: Tue, 27 May 2025 10:05:58 -0700
Subject: [PATCH] add ref count test (#846)

* add ref count test

* add coalesce and full outer join

* add recency and unique test
---
 .../tests/reference_tx_count_comparison.sql   | 29 +++++++++++++++++++
 models/gold/defi/defi__ez_dex_swaps.yml       | 17 ++++++++++-
 2 files changed, 45 insertions(+), 1 deletion(-)
 create mode 100644 macros/tests/reference_tx_count_comparison.sql

diff --git a/macros/tests/reference_tx_count_comparison.sql b/macros/tests/reference_tx_count_comparison.sql
new file mode 100644
index 00000000..0d149ca5
--- /dev/null
+++ b/macros/tests/reference_tx_count_comparison.sql
@@ -0,0 +1,29 @@
+{% test reference_tx_count_comparison(model, reference_table, id_column='tx_id') %}
+    {% set failures %}
+        WITH reference_count AS (
+            SELECT
+                COUNT({{ id_column }}) AS reference_count
+            FROM {{ ref(reference_table) }}
+            WHERE block_timestamp BETWEEN current_date - 2
+                                      AND current_timestamp() - INTERVAL '2 HOUR'
+        ),
+        model_count AS (
+            SELECT
+                COUNT({{ id_column }}) AS model_count
+            FROM {{ model }}
+            WHERE block_timestamp BETWEEN current_date - 2
+                                      AND current_timestamp() - INTERVAL '2 HOUR'
+        )
+        SELECT
+            COALESCE(r.reference_count, 0) AS reference_count,
+            COALESCE(m.model_count, 0) AS model_count,
+            reference_count - model_count AS count_difference
+        FROM reference_count r
+        FULL OUTER JOIN model_count m
+        WHERE reference_count <> model_count
+    {% endset %}
+
+    SELECT *
+    FROM ({{ failures }})
+{% endtest %}
+
diff --git a/models/gold/defi/defi__ez_dex_swaps.yml b/models/gold/defi/defi__ez_dex_swaps.yml
index 2d2e6b3b..fcfb5744 100644
--- a/models/gold/defi/defi__ez_dex_swaps.yml
+++ b/models/gold/defi/defi__ez_dex_swaps.yml
@@ -2,9 +2,21 @@ version: 2
 models:
   - name: defi__ez_dex_swaps
     description: A convenience table containing swaps performed on Jupiter (V2, V3), Orca, Raydium, Saber, Bonkswap, Dooar, Phoenix, Meteora, Pumpswap, and Pumpfun swap programs. For Phoenix, we are not capturing swaps where there are separate transactions for placing the order and filling the order. NOTE - Jupiter (V4, V5, V6) aggregator swaps exist in defi.fact_swaps_jupiter_summary, and individual routes are present in defi.fact_swaps_jupiter_inner. This documentation [guide](https://docs.google.com/document/d/1gxU7Q8BNf2w6xsDIczfxxOFMPM0ujQSJTzwGPDkZTVU/edit?tab=t.0) and [video](https://www.loom.com/share/f20d1f54b63342eea6457381be7175a1?sid=589f65ec-505a-446c-b249-19e6a9b9a6fb) explain how Flipside models Solana swaps, why we split the data, and how to use each table effectively.
+    recent_modified_date_filter: &recent_modified_date_filter
+      config:
+        where: modified_timestamp >= current_date - 7
+    tests:
+      - reference_tx_count_comparison:
+          reference_table: 'defi__fact_swaps'
+          id_column: 'tx_id'
     columns:
       - name: BLOCK_TIMESTAMP
         description: "{{ doc('block_timestamp') }}"
+        data_tests:
+          - not_null: *recent_modified_date_filter
+          - dbt_expectations.expect_row_values_to_have_recent_data:
+              datepart: day
+              interval: 2
       - name: BLOCK_ID
         description: "{{ doc('block_id') }}"
       - name: TX_ID
@@ -34,7 +46,10 @@ models:
       - name: _LOG_ID
         description: "Combination of TX_ID and event index"
       - name: EZ_SWAPS_ID
-        description: '{{ doc("pk") }}'   
+        description: '{{ doc("pk") }}'
+        data_tests:
+          - not_null: *recent_modified_date_filter
+          - unique: *recent_modified_date_filter  
       - name: INSERTED_TIMESTAMP
         description: '{{ doc("inserted_timestamp") }}'   
       - name: MODIFIED_TIMESTAMP