/dbt-utils

Utility functions for dbt projects.

Primary LanguageTSQLApache License 2.0Apache-2.0

This dbt package contains macros that can be (re)used across dbt projects.

Installation Instructions

Check dbt Hub for the latest installation instructions, or read the docs for more information on installing packages.


Macros

Cross-database

While these macros are cross database, they do not support all databases. These macros are provided to make date calculations easier and are not a core part of dbt. Most date macros are not supported on postgres.

current_timestamp (source)

This macro returns the current timestamp.

Usage:

{{ dbt_utils.current_timestamp() }}

dateadd (source)

This macro adds a time/day interval to the supplied date/timestamp. Note: The datepart argument is database-specific.

Usage:

{{ dbt_utils.dateadd(datepart='day', interval=1, from_date_or_timestamp="'2017-01-01'") }}

datediff (source)

This macro calculates the difference between two dates.

Usage:

{{ dbt_utils.datediff("'2018-01-01'", "'2018-01-20'", 'day') }}

split_part (source)

This macro splits a string of text using the supplied delimiter and returns the supplied part number (1-indexed).

Usage:

{{ dbt_utils.split_part(string_text='1,2,3', delimiter_text=',', part_number=1) }}

date_trunc (source)

Truncates a date or timestamp to the specified datepart. Note: The datepart argument is database-specific.

Usage:

{{ dbt_utils.date_trunc(datepart, date) }}

last_day (source)

Gets the last day for a given date and datepart. Notes:

  • The datepart argument is database-specific.
  • This macro currently only supports dateparts of month and quarter.

Usage:

{{ dbt_utils.last_day(date, datepart) }}

width_bucket (source)

This macro is modeled after the width_bucket function natively available in Snowflake.

From the original Snowflake documentation:

Constructs equi-width histograms, in which the histogram range is divided into intervals of identical size, and returns the bucket number into which the value of an expression falls, after it has been evaluated. The function returns an integer value or null (if any input is null). Notes:

  • expr The expression for which the histogram is created. This expression must evaluate to a numeric value or to a value that can be implicitly converted to a numeric value.

  • min_value and max_value The low and high end points of the acceptable range for the expression. The end points must also evaluate to numeric values and not be equal.

  • num_buckets The desired number of buckets; must be a positive integer value. A value from the expression is assigned to each bucket, and the function then returns the corresponding bucket number.

    When an expression falls outside the range, the function returns:

    0 if the expression is less than min_value.

    num_buckets + 1 if the expression is greater than or equal to max_value.

Usage:

{{ dbt_utils.width_bucket(expr, min_value, max_value, num_buckets) }}

Date/Time

date_spine (source)

This macro returns the sql required to build a date spine.

Usage:

{{ dbt_utils.date_spine(
    datepart="minute",
    start_date="to_date('01/01/2016', 'mm/dd/yyyy')",
    end_date="dateadd(week, 1, current_date)"
   )
}}

Geo

haversine_distance (source)

This macro calculates the haversine distance between a pair of x/y coordinates.

Usage:

{{ dbt_utils.haversine_distance(lat1=<float>,lon1=<float>,lat2=<float>,lon2=<float>) }}

Schema Tests

equal_rowcount (source)

This schema test asserts the that two relations have the same number of rows.

Usage:

version: 2

models:
  - name: model_name
    tests:
      - dbt_utils.equal_rowcount:
          compare_model: ref('other_table_name')

equality (source)

This schema test asserts the equality of two relations. Optionally specify a subset of columns to compare.

Usage:

version: 2

models:
  - name: model_name
    tests:
      - dbt_utils.equality:
          compare_model: ref('other_table_name')
          compare_columns:
            - first_column
            - second_column

expression_is_true (source)

This schema test asserts that a valid sql expression is true for all records. This is useful when checking integrity across columns, for example, that a total is equal to the sum of its parts, or that at least one column is true.

Usage:

version: 2

models:
  - name: model_name
    tests:
      - dbt_utils.expression_is_true:
          expression: "col_a + col_b = total"

The macro accepts an optional parameter condition that allows for asserting the expression on a subset of all records.

Usage:

version: 2

models:
  - name: model_name
    tests:
      - dbt_utils.expression_is_true:
          expression: "col_a + col_b = total"
          condition: "created_at > '2018-12-31'"

recency (source)

This schema test asserts that there is data in the referenced model at least as recent as the defined interval prior to the current timestamp.

Usage:

version: 2

models:
  - name: model_name
    tests:
      - dbt_utils.recency:
          datepart: day
          field: created_at
          interval: 1

at_least_one (source)

This schema test asserts if column has at least one value.

Usage:

version: 2

models:
  - name: model_name
    columns:
      - name: col_name
        tests:
          - dbt_utils.at_least_one

not_constant (source)

This schema test asserts if column does not have same value in all rows.

Usage:

version: 2

models:
  - name: model_name
    columns:
      - name: column_name
        tests:
          - dbt_utils.not_constant

cardinality_equality (source)

This schema test asserts if values in a given column have exactly the same cardinality as values from a different column in a different model.

Usage:

version: 2

models:
  - name: model_name
    columns:
      - name: from_column
        tests:
          - dbt_utils.cardinality_equality:
              field: other_column_name
              to: ref('other_model_name')

relationships_where (source)

This test validates the referential integrity between two relations (same as the core relationships schema test) with an added predicate to filter out some rows from the test. This is useful to exclude records such as test entities, rows created in the last X minutes/hours to account for temporary gaps due to ETL limitations, etc.

Usage:

version: 2

models:
  - name: model_name
    columns:
      - name: id
        tests:
          - dbt_utils.relationships_where:
              to: ref('other_model_name')
              field: client_id
              from_condition: id <> '4ca448b8-24bf-4b88-96c6-b1609499c38b'

mutually_exclusive_ranges (source)

This test confirms that for a given lower_bound_column and upper_bound_column, the ranges of between the lower and upper bounds do not overlap with the ranges of another row.

Usage:

version: 2

models:
  # test that age ranges do not overlap
  - name: age_brackets
    tests:
      - dbt_utils.mutually_exclusive_ranges:
          lower_bound_column: min_age
          upper_bound_column: max_age
          gaps: not_allowed

  # test that each customer can only have one subscription at a time
  - name: subscriptions
    tests:
      - dbt_utils.mutually_exclusive_ranges:
          lower_bound_column: started_at
          upper_bound_column: ended_at
          partition_by: customer_id
          gaps: required

Args:

  • lower_bound_column (required): The name of the column that represents the lower value of the range. Must be not null.
  • upper_bound_column (required): The name of the column that represents the upper value of the range. Must be not null.
  • partition_by (optional): If a subset of records should be mutually exclusive (e.g. all periods for a single subscription_id are mutually exclusive), use this argument to indicate which column to partition by. default=none
  • gaps (optional): Whether there can be gaps are allowed between ranges. default='allowed', one_of=['not_allowed', 'allowed', 'required']

Note: Both lower_bound_column and upper_bound_column should be not null. If this is not the case in your data source, consider passing a coalesce function to the lower_ and upper_bound_column arguments, like so:

version: 2

models:
- name: subscriptions
  tests:
    - dbt_utils.mutually_exclusive_ranges:
        lower_bound_column: coalesce(started_at, '1900-01-01')
        upper_bound_column: coalesce(ended_at, '2099-12-31')
        partition_by: customer_id
        gaps: allowed

Understanding the gaps parameter: Here are a number of examples for each allowed gaps parameter.

  • gaps:not_allowed: The upper bound of one record must be the lower bound of the next record.
lower_bound upper_bound
0 1
1 2
2 3
  • gaps:allowed (default): There may be a gap between the upper bound of one record and the lower bound of the next record.
lower_bound upper_bound
0 1
2 3
3 4
  • gaps:required: There must be a gap between the upper bound of one record and the lower bound of the next record (common for date ranges).
lower_bound upper_bound
0 1
2 3
4 5

unique_combination_of_columns (source)

This test confirms that the combination of columns is unique. For example, the combination of month and product is unique, however neither column is unique in isolation.

We generally recommend testing this uniqueness condition by either:

  • generating a surrogate_key for your model and testing the uniqueness of said key, OR
  • passing the unique test a coalesce of the columns (as discussed here).

However, these approaches can become non-perfomant on large data sets, in which case we recommend using this test instead.

Usage:

- name: revenue_by_product_by_month
  tests:
    - dbt_utils.unique_combination_of_columns:
        combination_of_columns:
          - month
          - product

SQL helpers

get_query_results_as_dict (source)

This macro returns a dictionary from a sql query, so that you don't need to interact with the Agate library to operate on the result

Usage:

-- Returns a dictionary of the users table where the state is California
{% set california_cities = dbt_utils.get_query_results_as_dict("select * from" ~ ref('cities') ~ "where state = 'CA' and city is not null ") %}
select
  city,
{% for city in california_cities %}
  sum(case when city = {{ city }} then 1 else 0 end) as users_in_{{ city }},
{% endfor %}
  count(*) as total
from {{ ref('users') }}

group by 1

get_column_values (source)

This macro returns the unique values for a column in a given relation. It takes an options default argument for compiling when the relation does not already exist.

Usage:

-- Returns a list of the top 50 states in the `users` table
{% set states = dbt_utils.get_column_values(table=ref('users'), column='state', max_records=50, default=[]) %}

{% for state in states %}
    ...
{% endfor %}

...

get_relations_by_prefix

This replaces the get_tables_by_prefix macro. Note that the get_tables_by_prefix macro will be deprecated in a future release of this package.

Returns a list of Relations that match a given prefix, with an optional exclusion pattern. It's particularly handy paired with union_relations. Usage:

-- Returns a list of relations that match schema.prefix%
{% set relations = dbt_utils.get_relations_by_prefix('my_schema', 'my_prefix') %}

-- Returns a list of relations as above, excluding any that end in `deprecated`
{% set relations = dbt_utils.get_relations_by_prefix('my_schema', 'my_prefix', '%deprecated') %}

-- Example using the union_relations macro
{% set event_relations = dbt_utils.get_relations_by_prefix('events', 'event_') %}
{{ dbt_utils.union_relations(relations = event_relations) }}

Args:

  • schema (required): The schema to inspect for relations.
  • prefix (required): The prefix of the table/view (case insensitive)
  • exclude (optional): Exclude any relations that match this pattern.
  • database (optional, default = target.database): The database to inspect for relations.

group_by (source)

This macro build a group by statement for fields 1...N

Usage:

{{ dbt_utils.group_by(n=3) }} --> group by 1,2,3

star (source)

This macro generates a list of all fields that exist in the from relation, excluding any fields listed in the except argument. The construction is identical to select * from {{ref('my_model')}}, replacing star (*) with the star macro. This macro also has an optional relation_alias argument that will prefix all generated fields with an alias.

Usage:

select
{{ dbt_utils.star(from=ref('my_model'), except=["exclude_field_1", "exclude_field_2"]) }}
from {{ref('my_model')}}

union_relations (source)

This replaces the union_tables macro. Note that the union_tables macro will be deprecated in a future release of this package.

This macro unions together an array of Relations, even when columns have differing orders in each Relation, and/or some columns are missing from some relations. Any columns exclusive to a subset of these relations will be filled with null where not present. An new column (_dbt_source_relation) is also added to indicate the source for each record.

Usage:

{{ dbt_utils.union_relations(
    relations=[ref('my_model'), source('my_source', 'my_table')],
    exclude=["_loaded_at"]
) }}

Args:

  • relations (required): An array of Relations.
  • exclude (optional): A list of column names that should be excluded from the final query.
  • include (optional): A list of column names that should be included in the final query. Note the include and exclude parameters are mutually exclusive.
  • column_override (optional): A dictionary of explicit column type overrides, e.g. {"some_field": "varchar(100)"}.``
  • source_column_name (optional, default="_dbt_source_relation"): The name of the column that records the source of this row.

generate_series (source)

This macro implements a cross-database mechanism to generate an arbitrarily long list of numbers. Specify the maximum number you'd like in your list and it will create a 1-indexed SQL result set.

Usage:

{{ dbt_utils.generate_series(upper_bound=1000) }}

surrogate_key (source)

Implements a cross-database way to generate a hashed surrogate key using the fields specified.

Usage:

{{ dbt_utils.surrogate_key(['field_a', 'field_b'[,...]]) }}

safe_add (source)

Implements a cross-database way to sum nullable fiellds using the fields specified.

Usage:

{{ dbt_utils.safe_add('field_a', 'field_b'[,...]) }}

pivot (source)

This macro pivots values from rows to columns.

Usage:

{{ dbt_utils.pivot(<column>, <list of values>) }}

Example:

Input: orders

| size | color |
|------|-------|
| S    | red   |
| S    | blue  |
| S    | red   |
| M    | red   |

select
  size,
  {{ dbt_utils.pivot(
      'color',
      dbt_utils.get_column_values(ref('orders'), 'color')
  ) }}
from {{ ref('orders') }}
group by size

Output:

| size | red | blue |
|------|-----|------|
| S    | 2   | 1    |
| M    | 1   | 0    |

Arguments:

- column: Column name, required
- values: List of row values to turn into columns, required
- alias: Whether to create column aliases, default is True
- agg: SQL aggregation function, default is sum
- cmp: SQL value comparison, default is =
- prefix: Column alias prefix, default is blank
- suffix: Column alias postfix, default is blank
- then_value: Value to use if comparison succeeds, default is 1
- else_value: Value to use if comparison fails, default is 0
- quote_identifiers: Whether to surround column aliases with double quotes, default is true

unpivot (source)

This macro "un-pivots" a table from wide format to long format. Functionality is similar to pandas melt function.

Usage:

{{ dbt_utils.unpivot(
  relation=ref('table_name'),
  cast_to='datatype',
  exclude=[<list of columns to exclude from unpivot>],
  remove=[<list of columns to remove>],
  field_name=<column name for field>,
  value_name=<column name for value>
) }}

Usage:

Input: orders

| date       | size | color | status     |
|------------|------|-------|------------|
| 2017-01-01 | S    | red   | complete   |
| 2017-03-01 | S    | red   | processing |

{{ dbt_utils.unpivot(ref('orders'), cast_to='varchar', exclude=['date','status']) }}

Output:

| date       | status     | field_name | value |
|------------|------------|------------|-------|
| 2017-01-01 | complete   | size       | S     |
| 2017-01-01 | complete   | color      | red   |
| 2017-03-01 | processing | size       | S     |
| 2017-03-01 | processing | color      | red   |

Args:

  • relation: The Relation to unpivot.
  • cast_to: The data type to cast the unpivoted values to, default is varchar
  • exclude: A list of columns to exclude from the unpivot operation but keep in the resulting table.
  • remove: A list of columns to remove from the resulting table.
  • field_name: column name in the resulting table for field
  • value_name: column name in the resulting table for value

Web

get_url_parameter (source)

This macro extracts a url parameter from a column containing a url.

Usage:

{{ dbt_utils.get_url_parameter(field='page_url', url_parameter='utm_source') }}

get_url_host (source)

This macro extracts a hostname from a column containing a url.

Usage:

{{ dbt_utils.get_url_host(field='page_url') }}

get_url_path (source)

This macro extracts a page path from a column containing a url.

Usage:

{{ dbt_utils.get_url_path(field='page_url') }}

Logger

pretty_time (source)

This macro returns a string of the current timestamp, optionally taking a datestring format.

{#- This will return a string like '14:50:34' -#}
{{ dbt_utils.pretty_time() }}

{#- This will return a string like '2019-05-02 14:50:34' -#}
{{ dbt_utils.pretty_time(format='%Y-%m-%d %H:%M:%S') }}

pretty_log_format (source)

This macro formats the input in a way that will print nicely to the command line when you log it.

{#- This will return a string like:
"11:07:31 + my pretty message"
-#}

{{ dbt_utils.pretty_log_format("my pretty message") }}

log_info (source)

This macro logs a formatted message (with a timestamp) to the command line.

{{ dbt_utils.log_info("my pretty message") }}
11:07:28 | 1 of 1 START table model analytics.fct_orders........................ [RUN]
11:07:31 + my pretty message

Materializations

insert_by_period (source)

insert_by_period allows dbt to insert records into a table one period (i.e. day, week) at a time.

This materialization is appropriate for event data that can be processed in discrete periods. It is similar in concept to the built-in incremental materialization, but has the added benefit of building the model in chunks even during a full-refresh so is particularly useful for models where the initial run can be problematic.

Should a run of a model using this materialization be interrupted, a subsequent run will continue building the target table from where it was interrupted (granted the --full-refresh flag is omitted).

Progress is logged in the command line for easy monitoring.

Usage:

{{
  config(
    materialized = "insert_by_period",
    period = "day",
    timestamp_field = "created_at",
    start_date = "2018-01-01",
    stop_date = "2018-06-01")
}}

with events as (

  select *
  from {{ ref('events') }}
  where __PERIOD_FILTER__ -- This will be replaced with a filter in the materialization code

)

....complex aggregates here....

Configuration values:

  • period: period to break the model into, must be a valid datepart (default='Week')
  • timestamp_field: the column name of the timestamp field that will be used to break the model into smaller queries
  • start_date: literal date or timestamp - generally choose a date that is earlier than the start of your data
  • stop_date: literal date or timestamp (default=current_timestamp)

Caveats:

  • This materialization is compatible with dbt 0.10.1.
  • This materialization has been written for Redshift.
  • This materialization can only be used for a model where records are not expected to change after they are created.
  • Any model post-hooks that use {{ this }} will fail using this materialization. For example:
models:
    project-name:
        post-hook: "grant select on {{ this }} to db_reader"

A useful workaround is to change the above post-hook to:

        post-hook: "grant select on {{ this.schema }}.{{ this.name }} to db_reader"

Contributing

We welcome contributions to this repo! To contribute a new feature or a fix, please open a Pull Request with 1) your changes, 2) updated documentation for the README.md file, and 3) a working integration test. See this page for more information.


Getting started with dbt

Code of Conduct

Everyone interacting in the dbt project's codebases, issue trackers, chat rooms, and mailing lists is expected to follow the PyPA Code of Conduct.