quollio-core 0.4.7__tar.gz → 0.4.9__tar.gz

{quollio_core-0.4.7 → quollio_core-0.4.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: quollio-core
-Version: 0.4.7
+Version: 0.4.9
 Summary: Quollio Core
 Author-email: quollio-dev <qt.dev@quollio.com>
 Maintainer-email: RyoAriyama <ryo.arym@gmail.com>, tharuta <35373297+TakumiHaruta@users.noreply.github.com>
@@ -79,7 +79,7 @@ To see available commands and options, please run the following command. (ex: Sn
 コマンドやオプションの詳細については、下記のコマンドを実行してください。(例: Snowflake)
 
 ```
-$ python3 -m quollio_core.snowflake -h
+$ python -m quollio_core.snowflake -h
 ```
 
 Then run commands with the options provided.

{quollio_core-0.4.7 → quollio_core-0.4.9}/README.md

@@ -31,7 +31,7 @@ To see available commands and options, please run the following command. (ex: Sn
 コマンドやオプションの詳細については、下記のコマンドを実行してください。(例: Snowflake)
 
 ```
-$ python3 -m quollio_core.snowflake -h
+$ python -m quollio_core.snowflake -h
 ```
 
 Then run commands with the options provided.

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/__init__.py

@@ -1,4 +1,4 @@
 """Quollio Core"""
 
-__version__ = "0.4.7"
+__version__ = "0.4.9"
 __author__ = "Quollio Technologies, Inc"

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/bricks.py

@@ -2,7 +2,7 @@ import argparse
 import logging
 import os
 
-from quollio_core.helper.core import setup_dbt_profile
+from quollio_core.helper.core import setup_dbt_profile, trim_prefix
 from quollio_core.helper.env_default import env_default
 from quollio_core.profilers.databricks import (
     databricks_column_level_lineage,
@@ -59,20 +59,35 @@ def build_view(
 
 def load_lineage(
     conn: db.DatabricksConnectionConfig,
+    endpoint: str,
     qdc_client: qdc.QDCExternalAPIClient,
     tenant_id: str,
+    enable_column_lineage: bool = False,
 ) -> None:
     logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(name)s - %(message)s")
 
     logger.info("Generate Databricks table to table lineage.")
     databricks_table_level_lineage(
-        conn=conn, qdc_client=qdc_client, tenant_id=tenant_id, dbt_table_name="quollio_lineage_table_level"
+        conn=conn,
+        endpoint=endpoint,
+        qdc_client=qdc_client,
+        tenant_id=tenant_id,
+        dbt_table_name="quollio_lineage_table_level",
     )
 
-    logger.info("Generate Databricks column to column lineage.")
-    databricks_column_level_lineage(
-        conn=conn, qdc_client=qdc_client, tenant_id=tenant_id, dbt_table_name="quollio_lineage_column_level"
-    )
+    if enable_column_lineage:
+        logger.info(
+            f"enable_column_lineage is set to {enable_column_lineage}.Generate Databricks column to column lineage."
+        )
+        databricks_column_level_lineage(
+            conn=conn,
+            endpoint=endpoint,
+            qdc_client=qdc_client,
+            tenant_id=tenant_id,
+            dbt_table_name="quollio_lineage_column_level",
+        )
+    else:
+        logger.info("Skip column lineage ingestion. Set enable_column_lineage to True if you ingest column lineage.")
 
     logger.info("Lineage data is successfully loaded.")
     return
@@ -80,6 +95,7 @@ def load_lineage(
 
 def load_column_stats(
     conn: db.DatabricksConnectionConfig,
+    endpoint: str,
     qdc_client: qdc.QDCExternalAPIClient,
     tenant_id: str,
 ) -> None:
@@ -88,6 +104,7 @@ def load_column_stats(
     logger.info("Generate Databricks column stats.")
     databricks_column_stats(
         conn=conn,
+        endpoint=endpoint,
         qdc_client=qdc_client,
         tenant_id=tenant_id,
     )
@@ -207,7 +224,6 @@ if __name__ == "__main__":
               Please specify table name with blank delimiter like tableA tableB \
               if you want to create two or more tables",
     )
-
     parser.add_argument(
         "--monitoring_table_suffix",
         type=str,
@@ -217,11 +233,20 @@ if __name__ == "__main__":
               This is used to identify the monitoring tables created by the databricks monitoring tool. \
               Default value is _profile_metrics",
     )
+    parser.add_argument(
+        "--enable_column_lineage",
+        type=bool,
+        action=env_default("ENABLE_COLUMN_LINEAGE", store_true=True),
+        default=False,
+        required=False,
+        help="Whether to ingest column lineage into QDIC or not. Default value is False",
+    )
 
     args = parser.parse_args()
 
     conn = db.DatabricksConnectionConfig(
-        host=args.host,
+        # MEMO: Metadata agent allows the string 'https://' as a host name but is not allowed by intelligence agent.
+        host=trim_prefix(args.host, "https://"),
         http_path=args.http_path,
         client_id=args.databricks_client_id,
         client_secret=args.databricks_client_secret,
@@ -243,7 +268,13 @@ if __name__ == "__main__":
         qdc_client = qdc.QDCExternalAPIClient(
             base_url=args.api_url, client_id=args.client_id, client_secret=args.client_secret
         )
-        load_lineage(conn=conn, qdc_client=qdc_client, tenant_id=args.tenant_id)
+        load_lineage(
+            conn=conn,
+            endpoint=args.host,
+            qdc_client=qdc_client,
+            tenant_id=args.tenant_id,
+            enable_column_lineage=args.enable_column_lineage,
+        )
 
     if "load_stats" in args.commands:
         qdc_client = qdc.QDCExternalAPIClient(
@@ -251,6 +282,7 @@ if __name__ == "__main__":
         )
         databricks_column_stats(
             conn=conn,
+            endpoint=args.host,
             qdc_client=qdc_client,
             tenant_id=args.tenant_id,
             monitoring_table_suffix=args.monitoring_table_suffix,

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/dbt_projects/databricks/models/quollio_lineage_table_level.yml

@@ -1,7 +1,7 @@
 version: 2
 
 model:
-  - name: quollio_lineage_column_level
+  - name: quollio_lineage_table_level
     columns: 
       - name: UPSTREAM_TABLES
         description: 'String column with all upstream tables in JSON format'

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/dbt_projects/redshift/dbt_project.yml

@@ -18,4 +18,4 @@ clean-targets:
 models:
   +dbt-osmosis: "{model}.yml"
   +grants:
-    select: ["{{ var('query_user') }}"]
+    select: ["\"{{ var('query_user') }}\""]

quollio_core-0.4.9/quollio_core/dbt_projects/redshift/macros/materialization/divided_view.sql

@@ -0,0 +1,136 @@
+{%- materialization divided_view, default %}
+{%- set identifier = model['alias'] %}
+{%- set target_relations = [] %}
+{%- set grant_config = config.get('grants') %}
+
+{{ run_hooks(pre_hooks, inside_transaction=False) }}
+-- `BEGIN` happens here:
+{{ run_hooks(pre_hooks, inside_transaction=True) }}
+
+-- fetch target_tables
+{%- set query_stats_target_tables -%}
+    SELECT
+      distinct
+      database_name
+      , schema_name
+      , table_name
+    FROM
+      {{ ref('quollio_stats_profiling_columns') }}
+    WHERE
+      table_name not like 'quollio_%%'
+{%- endset -%}
+{%- set results = run_query(query_stats_target_tables) -%}
+{%- if execute -%}
+{%- set stats_target_tables = results.rows -%}
+{%- else -%}
+{%- set stats_target_tables = [] -%}
+{%- endif -%}
+
+-- skip creating views if the target profiling columns don't exist.
+{%- if stats_target_tables | length == 0 -%}
+  {% call statement("main") %}
+    {{ log("No records found. Just execute select stmt for skipping call statement.", info=True) }}
+    select null
+  {% endcall %}
+  {%- set full_refresh_mode = (should_full_refresh()) -%}
+  {%- set should_revoke = should_revoke(target_relation, full_refresh_mode) %}
+{%- endif -%}
+
+-- build sql
+{%- for stats_target_table in stats_target_tables -%}
+  -- get columns for statistics. 
+  -- LISTAGG function can't be used for sys table, then it's necessary to get column for each table. 
+  -- See https://docs.aws.amazon.com/redshift/latest/dg/c_join_PG.html.
+  {%- set stats_target_columns %}
+      SELECT
+        database_name
+        , schema_name
+        , table_name
+        , column_name
+        , is_bool
+        , is_calculable
+      FROM
+        {{ ref('quollio_stats_profiling_columns') }}
+      WHERE
+        database_name = '{{stats_target_table[0]}}'
+        AND schema_name = '{{stats_target_table[1]}}'
+        AND table_name = '{{stats_target_table[2]}}'
+  {%- endset -%}
+
+  {%- set results = run_query(stats_target_columns) -%}
+  {%- set stats_target_columns = results.rows -%}
+
+  {%- set sql_for_column_stats %}
+  {%- for stats_target_column in stats_target_columns -%}
+    {%- if not loop.first -%}UNION{% endif %}
+    SELECT
+      main.db_name
+      , main.schema_name
+      , main.table_name
+      , main.column_name
+      , main.max_value
+      , main.min_value
+      , main.null_count
+      , main.cardinality
+      , main.avg_value
+      , main.median_value
+      , mode.mode_value
+      , main.stddev_value
+    FROM
+      (
+      SELECT
+        DISTINCT
+        '{{stats_target_column[0]}}'::varchar as db_name
+        , '{{stats_target_column[1]}}'::varchar as schema_name
+        , '{{stats_target_column[2]}}'::varchar as table_name
+        , '{{stats_target_column[3]}}'::varchar as column_name
+        , {% if var("aggregate_all") == True and stats_target_column[5] == True %}cast(max("{{stats_target_column[3]}}") as varchar){% else %}null::varchar{% endif %} AS max_value
+        , {% if var("aggregate_all") == True and stats_target_column[5] == True %}cast(min("{{stats_target_column[3]}}") as varchar){% else %}null::varchar{% endif %} AS min_value
+        -- requires full table scan
+        , {% if var("aggregate_all") == True %}cast(SUM(NVL2("{{stats_target_column[3]}}", 0, 1)) as integer){% else %}null::integer{% endif %} AS null_count
+        , APPROXIMATE COUNT(DISTINCT "{{stats_target_column[3]}}") AS cardinality
+        -- requires full table scan
+        , {% if var("aggregate_all") == True and stats_target_column[5] == True %}cast(avg("{{stats_target_column[3]}}")as varchar){% else %}null::varchar{% endif %} AS avg_value
+        , {% if var("aggregate_all") == True and stats_target_column[5] == True %}cast(median("{{stats_target_column[3]}}") as varchar){% else %}null::varchar{% endif %} AS median_value
+        -- requires full table scan
+        , {% if stats_target_column[5] == True %}cast(STDDEV_SAMP("{{stats_target_column[3]}}") as integer){% else %}null::integer{% endif %} AS stddev_value
+      FROM {{ stats_target_column[0] }}.{{ stats_target_column[1] }}.{{ stats_target_column[2] }}
+    ) main, (
+      {%- if var("aggregate_all") == True and stats_target_column[4] == false %}
+        SELECT
+          cast("{{stats_target_column[3]}}" as varchar) mode_value
+        FROM (
+           SELECT
+            DISTINCT
+            "{{stats_target_column[3]}}"
+            , ROW_NUMBER() OVER (ORDER BY COUNT(*) DESC) AS row_num
+          FROM {{ stats_target_column[0] }}.{{ stats_target_column[1] }}.{{ stats_target_column[2] }}
+          GROUP BY
+            "{{stats_target_column[3]}}"
+        )
+        WHERE
+          row_num = 1
+      {% else %}
+        SELECT null as mode_value {%- endif -%}
+    ) mode
+  {% endfor -%}
+  {%- endset %}
+  -- create a view with a index as suffix
+  {%- set target_identifier = "%s_%s_%s_%s"|format(model['name'], stats_target_table[0], stats_target_table[1], stats_target_table[2]) %}
+  {%- set target_relation = api.Relation.create(identifier=target_identifier, schema=schema, database=database, type='view') %}
+  -- {{ drop_relation_if_exists(target_relation) }}
+  {% call statement("main") %}
+    {{ get_replace_view_sql(target_relation, sql_for_column_stats) }}
+  {% endcall %}
+  {%- set full_refresh_mode = (should_full_refresh()) -%}
+  {%- set should_revoke = should_revoke(target_relation, full_refresh_mode) %}
+  {%- do apply_grants(target_relation, grant_config, should_revoke) %}
+  {%- set target_relations = target_relations.append(target_relation) %}
+{%- endfor -%}
+
+{{ run_hooks(post_hooks, inside_transaction=True) }}
+{{ adapter.commit() }}
+{{ run_hooks(post_hooks, inside_transaction=False) }}
+
+{{ return({'relations': target_relations}) }}
+{%- endmaterialization -%}

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/dbt_projects/redshift/models/quollio_stats_columns.sql

@@ -1,7 +1,6 @@
 {{
     config(
-        materialized='divided_view',
-        chunk=20
+        materialized='divided_view'
     )
 }}
 -- depends_on: {{ ref('quollio_stats_profiling_columns') }}

quollio_core-0.4.9/quollio_core/dbt_projects/snowflake/macros/materialization/divided_view.sql

@@ -0,0 +1,85 @@
+{%- materialization divided_view, default %}
+{%- set identifier = model['alias'] %}
+{%- set target_relations = [] %}
+{%- set grant_config = config.get('grants') %}
+
+{{ run_hooks(pre_hooks, inside_transaction=False) }}
+-- `BEGIN` happens here:
+{{ run_hooks(pre_hooks, inside_transaction=True) }}
+
+-- fetch target_tables
+{%- set query_stats_target_tables -%}
+    SELECT
+      TABLE_CATALOG
+      , TABLE_SCHEMA
+      , TABLE_NAME
+      , OBJECT_AGG(COLUMN_NAME, IS_CALCULABLE) AS COLUMNS_OBJ
+    FROM
+      {{ ref('quollio_stats_profiling_columns') }}
+    WHERE NOT startswith(table_name, 'QUOLLIO_')
+    GROUP BY
+      TABLE_CATALOG
+      , TABLE_SCHEMA
+      , TABLE_NAME
+{%- endset -%}
+{%- set results = run_query(query_stats_target_tables) -%}
+{%- if execute -%}
+{%- set stats_target_tables = results.rows -%}
+{%- else -%}
+{%- set stats_target_tables = [] -%}
+{%- endif -%}
+
+-- skip creating views if the target profiling columns don't exist.
+{%- if stats_target_tables | length == 0 -%}
+  {% call statement("main") %}
+    {{ log("No records found. Just execute select stmt for skipping call statement.", info=True) }}
+    select null
+  {% endcall %}
+  {%- set full_refresh_mode = (should_full_refresh()) -%}
+  {%- set should_revoke = should_revoke(target_relation, full_refresh_mode) %}
+{%- endif -%}
+
+-- create view for each table
+{%- for stats_target_table in stats_target_tables -%}
+  -- build sql for column value aggregation.
+  {%- set sql_for_column_stats %}
+  {% set columns_json = fromjson(stats_target_table[3]) %}
+  {%- for col_name, is_calclable in columns_json.items() -%}
+    {%- if not loop.first %}UNION{% endif %}
+    SELECT
+      DISTINCT
+      '{{stats_target_table[0]}}' as db_name
+      , '{{stats_target_table[1]}}' as schema_name
+      , '{{stats_target_table[2]}}' as table_name
+      , '{{col_name}}' as column_name
+      , {% if is_calclable == True %}CAST(MAX("{{col_name}}") AS STRING){% else %}NULL{% endif %} AS max_value
+      , {% if is_calclable == True %}CAST(MIN("{{col_name}}") AS STRING){% else %}NULL{% endif %} AS min_value
+      , COUNT_IF("{{col_name}}" IS NULL) AS null_count
+      , APPROX_COUNT_DISTINCT("{{col_name}}") AS cardinality
+      , {% if is_calclable == True %}AVG("{{col_name}}"){% else %}NULL{% endif %} AS avg_value
+      , {% if is_calclable == True %}MEDIAN("{{col_name}}"){% else %}NULL{% endif %} AS median_value
+      , {% if is_calclable == True %}APPROX_TOP_K("{{col_name}}")[0][0]{% else %}NULL{% endif %} AS mode_value
+      , {% if is_calclable == True %}STDDEV("{{col_name}}"){% else %}NULL{% endif %} AS stddev_value
+    FROM "{{stats_target_table[0]}}"."{{stats_target_table[1]}}"."{{stats_target_table[2]}}" {{ var("sample_method") }}
+  {% endfor -%}
+  {%- endset %}
+
+  -- create a view with a index as suffix
+  {%- set stats_view_identifier = "%s_%s_%s_%s"|format(model['name'], stats_target_table[0], stats_target_table[1], stats_target_table[2]) %}
+  {%- set target_relation = api.Relation.create(identifier=stats_view_identifier, schema=schema, database=database, type='view') %}
+  {% call statement("main") %}
+    {{ get_create_view_as_sql(target_relation, sql_for_column_stats) }}
+  {% endcall %}
+  {%- set full_refresh_mode = (should_full_refresh()) -%}
+  {%- set should_revoke = should_revoke(target_relation, full_refresh_mode) %}
+  {%- do apply_grants(target_relation, grant_config, should_revoke) %}
+  {%- set target_relations = target_relations.append(target_relation) %}
+{%- endfor -%}
+
+{{ run_hooks(post_hooks, inside_transaction=True) }}
+-- `COMMIT` happens here:
+{{ adapter.commit() }}
+{{ run_hooks(post_hooks, inside_transaction=False) }}
+
+{{ return({'relations': target_relations}) }}
+{%- endmaterialization -%}

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/dbt_projects/snowflake/models/quollio_stats_columns.sql

@@ -1,7 +1,6 @@
 {{
     config(
-        materialized='divided_view',
-        chunk=20
+        materialized='divided_view'
     )
 }}
 -- depends_on: {{ ref('quollio_stats_profiling_columns') }}

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/helper/core.py

@@ -31,3 +31,7 @@ def setup_dbt_profile(connections_json: Dict[str, str], template_path: str, temp
     with open(profile_path, "w") as profiles:
         yaml.dump(yaml.safe_load(profiles_body), profiles, default_flow_style=False, allow_unicode=True)
     return
+
+
+def trim_prefix(s: str, prefix: str) -> str:
+    return s.lstrip(prefix)

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/helper/env_default.py

@@ -6,6 +6,8 @@ Currently requires explicit naming of env vars to check for
 
 import argparse
 import os
+from distutils.util import strtobool
+from typing import Union
 
 
 # Courtesy of http://stackoverflow.com/a/10551190 with env-var retrieval fixed
@@ -28,9 +30,30 @@ class EnvDefault(argparse.Action):
         setattr(namespace, self.dest, values)
 
 
+class EnvStoreTrue(argparse._StoreTrueAction):
+    """An argparse action class that auto-sets missing default values from env vars for store_true."""
+
+    def __init__(self, envvar, required=True, default=None, **kwargs):
+        # Only pass the arguments that argparse._StoreTrueAction expects
+        action_kwargs = {key: value for key, value in kwargs.items() if key in ("option_strings", "dest")}
+        if envvar in os.environ:
+            default = _convert_value_to_bool(os.environ[envvar])
+        if required and default:
+            required = False
+        super(EnvStoreTrue, self).__init__(default=default, required=required, **action_kwargs)
+
+
 # functional sugar for the above
-def env_default(envvar):
+def env_default(envvar, store_true=False):
     def wrapper(**kwargs):
+        if store_true:
+            return EnvStoreTrue(envvar, **kwargs)
         return EnvDefault(envvar, **kwargs)
 
     return wrapper
+
+
+def _convert_value_to_bool(v: Union[str, bool]) -> bool:
+    if isinstance(v, str):
+        return bool(strtobool(v))
+    return v

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/profilers/databricks.py

@@ -14,6 +14,7 @@ logger = logging.getLogger(__name__)
 
 def databricks_table_level_lineage(
     conn: databricks.DatabricksConnectionConfig,
+    endpoint: str,
     qdc_client: qdc.QDCExternalAPIClient,
     tenant_id: str,
     dbt_table_name: str = "quollio_lineage_table_level",
@@ -31,7 +32,7 @@ def databricks_table_level_lineage(
         tables = parse_databricks_table_lineage(results)
         update_table_lineage_inputs = gen_table_lineage_payload(
             tenant_id=tenant_id,
-            endpoint=conn.host,
+            endpoint=endpoint,
             tables=tables,
         )
 
@@ -55,6 +56,7 @@ def databricks_table_level_lineage(
 
 def databricks_column_level_lineage(
     conn: databricks.DatabricksConnectionConfig,
+    endpoint: str,
     qdc_client: qdc.QDCExternalAPIClient,
     tenant_id: str,
     dbt_table_name: str = "quollio_lineage_column_level",
@@ -72,7 +74,7 @@ def databricks_column_level_lineage(
 
     update_column_lineage_inputs = gen_column_lineage_payload(
         tenant_id=tenant_id,
-        endpoint=conn.host,
+        endpoint=endpoint,
         columns=results,
     )
 
@@ -110,7 +112,9 @@ def _get_monitoring_tables(
             CONCAT(table_catalog, '.', table_schema, '.', table_name) AS table_fqdn
         FROM
             system.information_schema.tables
-        WHERE table_name LIKE "%{monitoring_table_suffix}"
+        WHERE
+            table_name LIKE "%{monitoring_table_suffix}"
+            AND table_name NOT LIKE ('quollio_%')
         """
     with databricks.DatabricksQueryExecutor(config=conn) as databricks_executor:
         tables = databricks_executor.get_query_results(query)
@@ -153,6 +157,8 @@ def _get_column_stats(
                     MAX(t.window) AS LATEST
                 FROM
                     {monitoring_table} t
+                WHERE
+                    t.column_name not in (':table')
                 GROUP BY
                     t.COLUMN_NAME,
                     t.DATA_TYPE,
@@ -176,13 +182,14 @@ def _get_column_stats(
 
 def databricks_column_stats(
     conn: databricks.DatabricksConnectionConfig,
+    endpoint: str,
     qdc_client: qdc.QDCExternalAPIClient,
     tenant_id: str,
     monitoring_table_suffix: str = "_profile_metrics",
 ) -> None:
     table_stats = _get_column_stats(conn, monitoring_table_suffix)
     for table in table_stats:
-        stats = gen_table_stats_payload(tenant_id, conn.host, table)
+        stats = gen_table_stats_payload(tenant_id=tenant_id, endpoint=endpoint, stats=table)
         for stat in stats:
             status_code = qdc_client.update_stats_by_id(
                 global_id=stat.global_id,

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/redshift.py

@@ -16,7 +16,7 @@ logger = logging.getLogger(__name__)
 
 def build_view(
     conn: redshift.RedshiftConnectionConfig,
-    skip_heavy: bool,
+    aggregate_all: bool = False,
     target_tables: str = "",
     log_level: str = "info",
 ) -> None:
@@ -29,9 +29,9 @@ def build_view(
     project_path = f"{current_dir}/dbt_projects/redshift"
     template_path = f"{current_dir}/dbt_projects/redshift/profiles"
     template_name = "profiles_template.yml"
-    options = '{{"query_user": {query_user}, "skip_heavy": {skip_heavy}, "target_database": {database}}}'.format(
+    options = '{{"query_user": {query_user}, "aggregate_all": {aggregate_all}, "target_database": {database}}}'.format(
         query_user=conn.query_user,
-        skip_heavy=skip_heavy,
+        aggregate_all=aggregate_all,
         database=conn.database,
     )
 
@@ -210,12 +210,12 @@ if __name__ == "__main__":
         help="Target schema name where the views are built by dbt",
     )
     parser.add_argument(
-        "--skip_heavy",
+        "--aggregate_all",
         type=bool,
-        action=env_default("REDSHIFT_SKIP_HEAVY"),
-        default=True,
+        action=env_default("REDSHIFT_AGGREGATE_ALL", store_true=True),
+        default=False,
         required=False,
-        help="Skip heavy queries when building views by dbt",
+        help="Aggregate all stats values. False by default.",
     )
     parser.add_argument(
         "--target_tables",
@@ -283,7 +283,7 @@ if __name__ == "__main__":
     if "build_view" in args.commands:
         build_view(
             conn=conn,
-            skip_heavy=args.skip_heavy,
+            aggregate_all=args.aggregate_all,
             target_tables=args.target_tables,
             log_level=args.log_level,
         )

{quollio_core-0.4.7 → quollio_core-0.4.9}/quollio_core/snowflake.py

@@ -72,6 +72,7 @@ def load_lineage(
     conn: snowflake.SnowflakeConnectionConfig,
     qdc_client: qdc.QDCExternalAPIClient,
     tenant_id: str,
+    enable_column_lineage: bool = False,
 ) -> None:
     logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(name)s - %(message)s")
 
@@ -82,12 +83,17 @@ def load_lineage(
         tenant_id=tenant_id,
     )
 
-    logger.info("Generate Snowflake column to column lineage.")
-    snowflake_column_to_column_lineage(
-        conn=conn,
-        qdc_client=qdc_client,
-        tenant_id=tenant_id,
-    )
+    if enable_column_lineage:
+        logger.info(
+            f"enable_column_lineage is set to {enable_column_lineage}.Generate Snowflake column to column lineage."
+        )
+        snowflake_column_to_column_lineage(
+            conn=conn,
+            qdc_client=qdc_client,
+            tenant_id=tenant_id,
+        )
+    else:
+        logger.info("Skip column lineage ingestion. Set enable_column_lineage to True if you ingest column lineage.")
 
     logger.info("Lineage data is successfully loaded.")
 
@@ -264,6 +270,14 @@ if __name__ == "__main__":
         required=False,
         help="The client secrete that is created on Quollio console to let clients access Quollio External API",
     )
+    parser.add_argument(
+        "--enable_column_lineage",
+        type=bool,
+        action=env_default("ENABLE_COLUMN_LINEAGE", store_true=True),
+        default=False,
+        required=False,
+        help="Whether to ingest column lineage into QDIC or not. Default value is False",
+    )
     args = parser.parse_args()
     conn = snowflake.SnowflakeConnectionConfig(
         account_id=args.account_id,
@@ -296,6 +310,7 @@ if __name__ == "__main__":
             conn=conn,
             qdc_client=qdc_client,
             tenant_id=args.tenant_id,
+            enable_column_lineage=args.enable_column_lineage,
         )
     if "load_stats" in args.commands:
         qdc_client = qdc.QDCExternalAPIClient(

quollio_core-0.4.7/quollio_core/dbt_projects/redshift/macros/materialization/divided_view.sql

@@ -1,97 +0,0 @@
-{%- materialization divided_view, default %}
-{%- set identifier = model['alias'] %}
-{%- set target_relations = [] %}
-{%- set chunk = config.get('chunk') %}
-{%- set grant_config = config.get('grants') %}
-
-{{ run_hooks(pre_hooks, inside_transaction=False) }}
--- `BEGIN` happens here:
-{{ run_hooks(pre_hooks, inside_transaction=True) }}
-
--- fetch records
-{%- set query_quollio_stats_profiling_columns -%}
-SELECT * FROM {{  ref('quollio_stats_profiling_columns')  }} WHERE table_name not like 'quollio_%'
-{%- endset -%}
-{%- set results = run_query(query_quollio_stats_profiling_columns) -%}
-{%- if execute -%}
-{%- set records = results.rows -%}
-{%- else -%}
-{%- set records = [] -%}
-{%- endif -%}
-
--- build sql
-{%- for i in range(0, records|length, chunk) -%}
-  {%- set build_sql %}
-  {%- for record in records[i: i+chunk] -%}
-    {%- if not loop.first -%}UNION{% endif %}
-    SELECT
-      main.db_name
-      , main.schema_name
-      , main.table_name
-      , main.column_name
-      , main.max_value
-      , main.min_value
-      , main.null_count
-      , main.cardinality
-      , main.avg_value
-      , main.median_value
-      , mode.mode_value
-      , main.stddev_value
-    FROM
-      (
-      SELECT
-        DISTINCT
-        '{{record[0]}}'::varchar as db_name
-        , '{{record[1]}}'::varchar as schema_name
-        , '{{record[2]}}'::varchar as table_name
-        , '{{record[3]}}'::varchar as column_name
-        , {% if var("skip_heavy") == false and record[5] == true %}cast(max("{{record[3]}}") as varchar){% else %}null::varchar{% endif %} AS max_value
-        , {% if var("skip_heavy") == false and record[5] == true %}cast(min("{{record[3]}}") as varchar){% else %}null::varchar{% endif %} AS min_value
-        -- requires full table scan
-        , {% if var("skip_heavy") == false %}cast(SUM(NVL2("{{record[3]}}", 0, 1)) as integer){% else %}null::integer{% endif %} AS null_count
-        , APPROXIMATE COUNT(DISTINCT "{{record[3]}}") AS cardinality
-        -- requires full table scan
-        , {% if var("skip_heavy") == false and record[5] == true %}cast(avg("{{record[3]}}")as varchar){% else %}null::varchar{% endif %} AS avg_value
-        , {% if var("skip_heavy") == false and record[5] == true %}cast(median("{{record[3]}}") as varchar){% else %}null::varchar{% endif %} AS median_value
-        -- requires full table scan
-        , {% if record[5] == true %}cast(STDDEV_SAMP("{{record[3]}}") as integer){% else %}null::integer{% endif %} AS stddev_value
-      FROM {{ record[0] }}.{{ record[1] }}.{{ record[2] }}
-    ) main, (
-      {%- if var("skip_heavy") == false and record[4] == false %}
-        SELECT
-          cast("{{record[3]}}" as varchar) mode_value
-        FROM (
-           SELECT
-            DISTINCT
-            "{{record[3]}}"
-            , ROW_NUMBER() OVER (ORDER BY COUNT(*) DESC) AS row_num
-          FROM {{ record[0] }}.{{ record[1] }}.{{ record[2] }}
-          GROUP BY
-            "{{record[3]}}"
-        )
-        WHERE
-          row_num = 1
-      {% else %}
-        SELECT null as mode_value {%- endif -%}
-    ) mode
-  {% endfor -%}
-  {%- endset %}
-  -- create a view with a index as suffix
-  {%- set target_identifier = "%s_%d"|format(model['name'], loop.index) %}
-  {%- set target_relation = api.Relation.create(identifier=target_identifier, schema=schema, database=database, type='view') %}
-  -- {{ drop_relation_if_exists(target_relation) }}
-  {% call statement("main") %}
-    {{ get_replace_view_sql(target_relation, build_sql) }}
-  {% endcall %}
-  {%- set full_refresh_mode = (should_full_refresh()) -%}
-  {%- set should_revoke = should_revoke(target_relation, full_refresh_mode) %}
-  {%- do apply_grants(target_relation, grant_config, should_revoke) %}
-  {%- set target_relations = target_relations.append(target_relation) %}
-{%- endfor -%}
-
-{{ run_hooks(post_hooks, inside_transaction=True) }}
-{{ adapter.commit() }}
-{{ run_hooks(post_hooks, inside_transaction=False) }}
-
-{{ return({'relations': target_relations}) }}
-{%- endmaterialization -%}

quollio_core-0.4.7/quollio_core/dbt_projects/snowflake/macros/materialization/divided_view.sql

@@ -1,63 +0,0 @@
-{%- materialization divided_view, default %}
-{%- set identifier = model['alias'] %}
-{%- set target_relations = [] %}
-{%- set chunk = config.get('chunk') %}
-{%- set grant_config = config.get('grants') %}
-
-{{ run_hooks(pre_hooks, inside_transaction=False) }}
--- `BEGIN` happens here:
-{{ run_hooks(pre_hooks, inside_transaction=True) }}
-
--- fetch records
-{%- set query_quollio_stats_profiling_columns -%}
-SELECT * FROM {{  ref('quollio_stats_profiling_columns')  }} WHERE NOT startswith(table_name, 'QUOLLIO_')
-{%- endset -%}
-{%- set results = run_query(query_quollio_stats_profiling_columns) -%}
-{%- if execute -%}
-{%- set records = results.rows -%}
-{%- else -%}
-{%- set records = [] -%}
-{%- endif -%}
-
--- build sql
-{%- for i in range(0, records|length, chunk) -%}
-  {%- set build_sql %}
-  {%- for record in records[i: i+chunk] -%}
-    {%- if not loop.first %}UNION{% endif %}
-
-    SELECT
-      DISTINCT
-      '{{record[0]}}' as db_name
-      , '{{record[1]}}' as schema_name
-      , '{{record[2]}}' as table_name
-      , '{{record[3]}}' as column_name
-      , {% if record[5] == true %}CAST(max("{{record[3]}}") AS STRING){% else %}null{% endif %} AS max_value
-      , {% if record[5] == true %}CAST(min("{{record[3]}}") AS STRING){% else %}null{% endif %} AS min_value
-      , COUNT_IF("{{record[3]}}" IS NULL) AS null_count
-      , APPROX_COUNT_DISTINCT("{{record[3]}}") AS cardinality
-      , {% if record[5] == true %}avg("{{record[3]}}"){% else %}null{% endif %} AS avg_value
-      , {% if record[5] == true %}median("{{record[3]}}"){% else %}null{% endif %} AS median_value
-      , {% if record[5] == true %}approx_top_k("{{record[3]}}")[0][0]{% else %}null{% endif %} AS mode_value
-      , {% if record[5] == true %}stddev("{{record[3]}}"){% else %}null{% endif %} AS stddev_value
-    FROM "{{record[0]}}"."{{record[1]}}"."{{record[2]}}" {{ var("sample_method") }}
-  {% endfor -%}
-  {%- endset %}
-  -- create a view with a index as suffix
-  {%- set target_identifier = "%s_%d"|format(model['name'], loop.index) %}
-  {%- set target_relation = api.Relation.create(identifier=target_identifier, schema=schema, database=database, type='view') %}
-  {% call statement("main") %}
-    {{ get_create_view_as_sql(target_relation, build_sql) }}
-  {% endcall %}
-  {%- set full_refresh_mode = (should_full_refresh()) -%}
-  {%- set should_revoke = should_revoke(target_relation, full_refresh_mode) %}
-  {%- do apply_grants(target_relation, grant_config, should_revoke) %}
-  {%- set target_relations = target_relations.append(target_relation) %}
-{%- endfor -%}
-
-{{ run_hooks(post_hooks, inside_transaction=True) }}
--- `COMMIT` happens here:
-{{ adapter.commit() }}
-{{ run_hooks(post_hooks, inside_transaction=False) }}
-
-{{ return({'relations': target_relations}) }}
-{%- endmaterialization -%}