castor-extractor 0.21.5__tar.gz → 0.21.7__tar.gz

{castor_extractor-0.21.5 → castor_extractor-0.21.7}/CHANGELOG.md

@@ -1,6 +1,18 @@
 
 # Changelog
 
+## 0.21.8 - 2024-11-26
+
+* Redshift: improve deduplication of columns
+
+## 0.21.7 - 2024-11-26
+
+* Metabase: stop using deprecated table `view_log`
+
+## 0.21.6 - 2024-11-22
+
+* bump dependencies: ruff, setuptools
+
 ## 0.21.5 - 2024-11-20
 
 * PostgreSQL: Fix schema extraction when owner is a role without login privilege

{castor_extractor-0.21.5 → castor_extractor-0.21.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: castor-extractor
-Version: 0.21.5
+Version: 0.21.7
 Summary: Extract your metadata assets.
 Home-page: https://www.castordoc.com/
 License: EULA
@@ -52,7 +52,7 @@ Requires-Dist: pymssql (>=2.2.11,<3.0.0) ; extra == "sqlserver" or extra == "all
 Requires-Dist: pymysql[rsa] (>=1.1.0,<2.0.0) ; extra == "mysql" or extra == "all"
 Requires-Dist: python-dateutil (>=2.0.0,<=3.0.0)
 Requires-Dist: requests (>=2.0.0,<3.0.0)
-Requires-Dist: setuptools (>=74,<75)
+Requires-Dist: setuptools (>=75.3.0,<75.4.0)
 Requires-Dist: snowflake-connector-python (>=3.4.0,<4.0.0) ; extra == "snowflake" or extra == "all"
 Requires-Dist: snowflake-sqlalchemy (!=1.2.5,<2.0.0) ; extra == "snowflake" or extra == "all"
 Requires-Dist: sqlalchemy (>=1.4,<1.5)
@@ -208,6 +208,18 @@ For any questions or bug report, contact us at [support@castordoc.com](mailto:su
 
 # Changelog
 
+## 0.21.8 - 2024-11-26
+
+* Redshift: improve deduplication of columns
+
+## 0.21.7 - 2024-11-26
+
+* Metabase: stop using deprecated table `view_log`
+
+## 0.21.6 - 2024-11-22
+
+* bump dependencies: ruff, setuptools
+
 ## 0.21.5 - 2024-11-20
 
 * PostgreSQL: Fix schema extraction when owner is a role without login privilege

castor_extractor-0.21.7/castor_extractor/visualization/metabase/client/db/queries/card.sql

@@ -0,0 +1,4 @@
+SELECT
+    *
+FROM
+    {schema}.report_card

castor_extractor-0.21.7/castor_extractor/visualization/metabase/client/db/queries/dashboard.sql

@@ -0,0 +1,4 @@
+SELECT
+    *
+FROM
+    {schema}.report_dashboard

{castor_extractor-0.21.5 → castor_extractor-0.21.7}/castor_extractor/warehouse/redshift/queries/column.sql

@@ -155,9 +155,9 @@ external_columns AS (
         db.datname || '.' || s.schemaname || '.' || c.tablename AS table_id,
         c.columnname AS column_name,
         db.datname || '.' || s.schemaname || '.' || c.tablename || '.' || c.columnname AS column_id,
-        c.external_type AS data_type,
+        MIN(c.external_type) AS data_type,
         MIN(c.columnnum) AS ordinal_position,
-        CASE c.is_nullable WHEN 'false' THEN 'NO' ELSE 'YES' END AS is_nullable,
+        MIN(CASE c.is_nullable WHEN 'false' THEN 'NO' ELSE 'YES' END) AS is_nullable,
         NULL AS character_maximum_length,
         NULL AS character_octet_length,
         NULL AS numeric_precision,
@@ -172,7 +172,7 @@ external_columns AS (
         JOIN pg_catalog.pg_database AS db ON CURRENT_DATABASE() = db.datname
 
     -- To remove duplicate column names that can occur in external tables (no check on CSVs)
-    GROUP BY database_name, database_id, schema_name, schema_id, table_name, table_id, column_name, column_id, data_type, is_nullable
+    GROUP BY database_name, database_id, schema_name, schema_id, table_name, table_id, column_name, column_id
 )
 
 SELECT * FROM tables

{castor_extractor-0.21.5 → castor_extractor-0.21.7}/pyproject.toml

@@ -3,7 +3,7 @@ requires = ["setuptools>=61.2"]
 
 [tool.poetry]
 name = "castor-extractor"
-version = "0.21.5"
+version = "0.21.7"
 description = "Extract your metadata assets."
 authors = ["Castor <support@castordoc.com>"]
 license = "EULA"
@@ -51,7 +51,7 @@ python-dateutil = ">=2.0.0, <=3.0.0"
 requests = ">=2.0.0, <3.0.0"
 snowflake-connector-python = { version = "^3.4.0", optional = true }
 snowflake-sqlalchemy = { version = "!=1.2.5, <2.0.0", optional = true }
-setuptools = "^74"
+setuptools = "~75.3.0"  # locked, requires python >= 3.9
 sqlalchemy = "~1.4"
 sqlalchemy-bigquery = { version = ">=1.0.0, <= 2.0.0", optional = true, extras = [
   "bqstorage",
@@ -108,10 +108,10 @@ all = [
 pytest = "^8"
 pytest-freezegun = "^0.4"
 requests-mock = "^1.11"
-ruff = "^0.6.4"
+ruff = "^0.7.4"
 
 [tool.poetry.group.dev.dependencies]
-deptry = "^0.20"
+deptry = "^0.20"  # locked, requires python >= 3.9
 
 [tool.poetry.scripts]
 castor-extract-bigquery = "castor_extractor.commands.extract_bigquery:main"

castor_extractor-0.21.5/castor_extractor/visualization/metabase/client/db/queries/card.sql

@@ -1,21 +0,0 @@
-WITH card_view_log AS
-(
-    SELECT
-        model_id,
-        COUNT(id) AS view_count,
-        MAX("timestamp") AS last_viewed_at
-    FROM
-        {schema}.view_log
-    WHERE
-        model = 'card'
-    GROUP BY
-        1
-)
-
-SELECT
-    rc.*,
-    vl.last_viewed_at AS last_viewed_at,
-    COALESCE(vl.view_count, 0) AS view_count
-FROM
-    {schema}.report_card AS rc
-    LEFT JOIN card_view_log AS vl ON rc.id = vl.model_id

castor_extractor-0.21.5/castor_extractor/visualization/metabase/client/db/queries/dashboard.sql

@@ -1,21 +0,0 @@
-WITH dashboard_view_log AS
-(
-    SELECT
-        model_id,
-        COUNT(id) AS view_count,
-        MAX("timestamp") AS last_viewed_at
-    FROM
-        {schema}.view_log
-    WHERE
-        model = 'dashboard'
-    GROUP BY
-        1
-)
-
-SELECT
-    rd.*,
-    vl.last_viewed_at AS last_viewed_at,
-    COALESCE(vl.view_count, 0) AS view_count
-FROM
-    {schema}.report_dashboard AS rd
-    LEFT JOIN dashboard_view_log AS vl ON rd.id = vl.model_id