fix: reconcile sqlfluff rules

* Merging the sqlfluff config from pyproject.toml into .sqlfluff

.sqlfluff

@@ -3,10 +3,10 @@ dialect = bigquery
 templater = dbt
 runaway_limit = 10
 max_line_length = 80
-indent_unit = space
 
 [sqlfluff:indentation]
-tab_space_size = 4
+indent_unit = space
+tab_space_size = 2
 
 [sqlfluff:layout:type:comma]
 spacing_before = touch
@@ -35,4 +35,13 @@ capitalisation_policy = lower
 
 [sqlfluff:rules:ambiguous.column_references]  # Number in group by
 group_by_and_order_by_style = consistent
-#group_by_and_order_by_style = explicit
+#group_by_and_order_by_style = explicit
+
+[sqlfluff.templater.jinja]
+load_macros_from_path = "warehouse/dbt/macros/"
+apply_dbt_builtins = true
+library_path = "warehouse/common/dbtlintmock"
+
+[sqlfluff.templater.dbt]
+project_dir = "."
+target = "playground"

package.json

@@ -26,6 +26,7 @@
     "dev:frontend": "turbo run dev --filter=@opensource-observer/frontend --parallel",
     "format:staged": "lint-staged",
     "lint": "turbo run lint --concurrency=100%",
+    "lint:sql": "poetry run sqlfluff lint",
     "serve": "yarn serve build",
     "test": "turbo run test --concurrency=1",
     "test:integration": "(docker compose -f docker/compose.yaml down || true) && docker compose -f docker/compose.yaml run --rm test",

pyproject.toml

@@ -39,7 +39,6 @@ lz4 = "^4.3.3"
 arrow = "^1.3.0"
 polars = "^0.20.23"
 
-
 [tool.poetry.scripts]
 bq2cloudsql = 'bq2cloudsql.script:run'
 oso_lets_go = 'oso_lets_go.wizard:run'
@@ -55,24 +54,5 @@ dagster-webserver = "^1.7.2"
 requires = ["poetry-core"]
 build-backend = "poetry.core.masonry.api"
 
-[tool.sqlfluff]
-max_line_length = 80
-
-[tool.sqlfluff.core]
-templater = "dbt"
-
-[tool.sqlfluff.indentation]
-indent_unit = "space"
-tab_space_size = 2
-
-[tool.sqlfluff.templater.jinja]
-load_macros_from_path = "warehouse/dbt/macros/"
-apply_dbt_builtins = true
-library_path = "warehouse/common/dbtlintmock"
-
-[tool.sqlfluff.templater.dbt]
-project_dir = "."
-target = "playground"
-
 [tool.dagster]
 module_name = "oso_dagster.definitions"

warehouse/dbt/models/intermediate/directory/stg_ossd__artifacts_by_project.sql

@@ -26,7 +26,9 @@ all_npm as (
         LOWER(JSON_VALUE(npm.url)) like 'https://npmjs.com/package/%'
         then SUBSTR(LOWER(JSON_VALUE(npm.url)), 28)
       when
-        LOWER(JSON_VALUE(npm.url)) like 'https://www.npmjs.com/package/%'
+        LOWER(
+          JSON_VALUE(npm.url)
+        ) like 'https://www.npmjs.com/package/%'
         then SUBSTR(LOWER(JSON_VALUE(npm.url)), 31)
     end as artifact_name,
     LOWER(JSON_VALUE(npm.url)) as artifact_url,

warehouse/dbt/models/staging/ethereum/stg_ethereum__deployers.sql

@@ -25,9 +25,9 @@ where
   and receipt_status = 1
   and receipt_contract_address is not null
 {% if is_incremental() %}
-        AND block_timestamp >= (
-            SELECT MAX(block_timestamp)
-            FROM {{ this }}
-        )
-        AND block_timestamp < TIMESTAMP_TRUNC(CURRENT_TIMESTAMP(), DAY)
-    {% endif %}
+    AND block_timestamp >= (
+      SELECT MAX(block_timestamp)
+      FROM {{ this }}
+    )
+    AND block_timestamp < TIMESTAMP_TRUNC(CURRENT_TIMESTAMP(), DAY)
+  {% endif %}

warehouse/dbt/models/staging/github/stg_github__events.sql

@@ -29,22 +29,23 @@ where
 
   {# If this is production then make sure it's incremental #}
   {% if target.name == 'production' %}
-  {% if is_incremental() %}
-  AND _TABLE_SUFFIX > FORMAT_TIMESTAMP("%y%m%d", TIMESTAMP_SUB(_dbt_max_partition, INTERVAL 1 DAY))
-  {% else %}
-    {# 
-    If/when we do full refreshes we are currently only taking data from 2015
-    onward due to schema changes of the github events in the githubarchive 
-    #}
-    AND _TABLE_SUFFIX >= "150101" 
-  {% endif %}
-{% elif target.name in ['dev', 'playground'] %}
+    {% if is_incremental() %}
+      AND _TABLE_SUFFIX > FORMAT_TIMESTAMP("%y%m%d", TIMESTAMP_SUB(_dbt_max_partition, INTERVAL 1 DAY))
+    {% else %}
+      {# 
+      If/when we do full refreshes we are currently only taking data from 2015
+      onward due to schema changes of the github events in the githubarchive 
+      #}
+      AND _TABLE_SUFFIX >= "150101" 
+    {% endif %}
+  {% elif target.name in ['dev', 'playground'] %}
     {# 
       If this is not production then we only copy the most recent number of days
       (default to 14) 
     #}
     and created_at
     >= TIMESTAMP_SUB(
-      CURRENT_TIMESTAMP(), interval {{ env_var("PLAYGROUND_DAYS", '14') }} day
+      CURRENT_TIMESTAMP(),
+      interval {{ env_var("PLAYGROUND_DAYS", '14') }} day
     )
   {% endif %}