Extract (Function) |

Location	Street
221 B Baker St., London, England, United Kingdom	['221 B Baker St.']
London SW1A 1AA, London, England, United Kingdom

Parameter	Required	Data Type	Notes
input	✓	str, list
output	✓	str, list
dataType	✓	streets / cities / regions / countries	Specific part of the address to extract
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
output_format		string	Format of the extract output Allowed values: `list`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

Product Specs	Blade Diameter	Max. RPM
Saw with a 4.5 inch blade and maximum speed of 3600 rpm.	4.5	3600

Parameter	Required	Data Type	Notes
input		string, integer, array	Input column name, column index, or list of columns supplied together as DATA for each row. If omitted, all dataframe columns are supplied.
output		object, string, array	Desired extraction. Use an object keyed by output column name for structured fields, a string for one prompted value, or an array of field names/definitions. Each field may use the schema options below.
record_examples		array, object	Whole-record examples. Each example has a separate input value or record and the complete expected output record. Optional name and notes provide model-visible context. Use {name: ..., notes: ..., input: ..., output: ...}. Omitted nullable output fields are completed with null. Required non-null nested properties must be supplied. This differs from examples nested under one output field, which teach only that field.
api_key	✓	string	OpenAI API key used for this wrangle, normally supplied through a recipe variable.
model		string	OpenAI model ID for this call. If omitted, uses the configured extract.ai default; a saved model definition may supply its own model.
threads		integer	Maximum number of row-level requests sent in parallel. The configured default is 32.
timeout		number	Network timeout in seconds for each HTTP attempt. The configured default is 12. Each retry uses the same timeout.
retries		integer	Number of additional attempts per row after a retryable failure. The configured default is 1. Retry delays are separate from timeout.
url		string	Override the endpoint for the selected protocol. A chat/completions URL selects the legacy protocol only when protocol is omitted; new recipes should use the configured Responses endpoint.
provider		string	AI service provider. Currently only OpenAI is supported. Allowed values: `openai`.
protocol		string	OpenAI API protocol. Responses is the configured default and is required for web_search; chat_completions remains available for legacy definitions. Allowed values: `responses`, `chat_completions`.
store		boolean	Whether OpenAI may store Responses API results. Defaults to true.
metadata		object	Labels attached to OpenAI requests, such as recipe_name and wrangles_user. Available recipe name and Wrangles user are added automatically. Explicit labels override those defaults; an empty object disables automatic labels. These appear with stored logs and are separate from model instructions. Up to 16 string pairs; keys may contain up to 64 characters and values up to 512 characters. Does not enable workflow tracing.
cache		boolean	Reuse identical successful results from the bounded warm-instance cache. Defaults to true. Set false when fresh model or web results are required.
cache_ttl		number	Maximum age in seconds for a cached result used by this call. Applies to extracted values and web_search_sources together.
messages		str, list	Legacy alias for general instructions. Prefer `instructions` in new recipes.
instructions		string, array	General instructions applied to every input row, in addition to any GeneralInstructions in the saved model. Use this for decision rules, evidence priorities, normalization requirements, or other behavior that applies to the complete extraction.
model_id		string	ID of a saved extract.ai definition. Use it instead of defining an output schema. When output is also supplied with model_id in a recipe, output names the destination column or columns for the saved fields.
strict		boolean	Require OpenAI structured-output strict mode. Defaults to true. Definitions with dynamic dictionary keys automatically switch to non-strict provider mode and are still validated locally.
output_format		string	How extracted fields are written. columns writes one dataframe column per field (default); dictionary keeps one object; concatenate joins fields into one string using char. Allowed values: `dictionary`, `columns`, `concatenate`.
char		string	Separator used only when output_format is concatenate. Defaults to comma-space.
reasoning		string, object	Reasoning effort, such as none or low, or an object with effort and other reasoning options. Both forms are equivalent for setting effort. Defaults to none for every model. Explicit settings are always sent to the provider; unsupported settings produce a provider error.
verbosity		string	Responses API text verbosity for compatible models. Defaults to low when supported; ignored with a warning for incompatible models. Allowed values: `low`, `medium`, `high`.
web_search		boolean	Enable OpenAI Responses web search; the model decides when searching helps. When true, every row also receives web_search_sources: a deduplicated list of {title, url} objects in source order, or an empty list when no source was used. This reserved column is automatic. Requires protocol responses. Defaults to false.
where		str	Apply to rows matching a SQL WHERE expression.
where_params		list, object	Values for positional (?) or named (:name) WHERE parameters.
if		str	Condition controlling whether the whole wrangle runs.

Parameter	Data Type	Notes
type	string	JSON data type required for this field. If omitted, common scalar types are accepted. Fields allow null by default. Allowed values: `string`, `number`, `integer`, `boolean`, `null`, `object`, `array`.
description	string	Plain-language definition of the value to extract, including any selection, normalization, unit, or evidence rules.
enum	array	Allowed output values. The model must choose one of these values; null is also allowed unless nullable is false.
default	string, number, integer, boolean, null, object, array	JSON Schema annotation for a preferred default. extract.ai does not substitute this value when evidence is missing; describe fallback behavior explicitly or allow null.
examples	array, object, string, number, integer, boolean, null	Field-specific examples. The backward-compatible form is a scalar or list of typical output values. A paired example may instead use input and output, with optional name and notes. Paired examples apply only to this output field; use record_examples for complete output records. Object outputs must include every required non-null nested property.
properties	object, array, string	Child fields when type is object. Use an object to define a schema for each child. A list or comma-separated string is a shortcut that creates fixed child names. Named child values are non-null by default.
required	array, string	Named object properties that must be returned. If omitted, every named property is required. Strings may use pipe or comma delimiters.
additionalProperties	boolean, object	Controls keys beyond properties when type is object. Set false for fixed keys, true for arbitrary values, or provide one schema applied to every dynamic value. Dynamic dictionaries use non-strict provider mode plus local validation. Defaults to false when named properties exist.
items	object	Schema applied to every element when this field's type is array.
nullable	boolean	Whether the field may return null. Defaults to true while a top-level field key remains required. Named nested properties default to false. Set this explicitly to override the applicable default.

Tools	Attributes (Span)
hammer 5kg, 0.5m	`{'length': ['0.5m'], 'mass': ['5kg']}`

Tools	Attributes (Object)
hammer 5kg, 0.5m	{'length': [{'unit': 'metre', 'value': 0.5}], 'mass': [{'unit': 'kilogram', 'value': 5.0}]}

Parameter	Required	Data Type	Notes
input	✓	str, list
output	✓	str, list
responseContent		span / object	If omited defaults to span (text found). Object returns an object with the value and unit.
attribute_type		See list of options below	Specify which attributes to return, if omitted defaults to all
desired_unit		str	Convert the extracted unit to the desired unit
bound		min/ mid/ max	When returning an object, if the input is a range (e.g. 10-20mm) set the value to return. min, mid or max. Default mid.
output_format		string	Format of the extract output Allowed values: `list`, `dictionary`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
first_element		bool	Get the first element from results, defaults to False.
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

Data	Output
{Hello}	Hello
[Wrangles]	Wrangles
(!)	!
<!>	!

Parameter	Required	Data Type	Notes
input	✓	str, list
output	✓	str, list
first_element		bool	Get the first element from results, defaults to False.
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
output_format		string	Format of the extract output Allowed values: `list`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
find		str, list	The type of brackets to find (round '()', square '[]', curly '{}', angled '<>'). Default is all brackets.
include_brackets		bool	Include the brackets in the output
if		str	A condition that will determine whether the action runs or not as a whole.

Secret	Code Extract
to gain access use Z1ON0101	['Z1ON0101']

code1	code2	Codes
code CH465517080-1	code CH465517080-2	['CH465517080-1', 'CH465517080-2']

Parameter	Required	Data Type	Notes
input	✓	str, list
output	✓	str, list
first_element		bool	Get the first element from results, defaults to False.
min_length		str, int	Minimum length of allowed results
max_length		str, int	Maximun length of allowed results
strategy		str	How aggressive to be at removing false positives such as measurements (lenient, balanced, or strict).
sort_order		str	`input` (default) retains input order; `longest` and `shortest` sort by code length. Equal-length matches retain input order.
disallowed_patterns		str	A pattern or JSON array of regex patterns to not include in the found codes
include_multi_part_tokens		bool	Whether to include multi-part tokens that have a space. Default True.
extract_raw		bool	Include adjacent non-whitespace characters around the detected token instead of returning only the cleaned token. Default `False`.
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
output_format		string	Format of the extract output Allowed values: `list`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

Product	Wood Types
Dining Oakwood Chair	Oakwood
Living Room Teakwood Frame Mirror	Teakwood

Part 1 of 2	Part 2 of 2	Wood Types
Dining Acacia Wood Table	Imitation Wood Table Chairs	['Acacia Wood', 'Imitation Wood']

Product	Wood Types	Item Type
Dining Oakwood Chair	Oakwood	Chair
Living Room Teakwood Frame Mirror	Teakwood	Mirror

Parameter	Required	Data Type	Notes
input	✓	str, list
output	✓	str, list
model_id	✓	str, list	Model id of the custom Extract Wrangle being used.
use_labels		bool	Adds labels (if given) to the extracted data, defaults to false. See here for more information.
first_element		bool	If true, returns only the first element of matched values.
case_sensitive		bool	Allows the wrangle to be case sensitive if set to True, default is False.
use_spellcheck		bool	Use spellcheck to correct spelling mistakes in the input.
sort		str	Sort the results based on length, training data, input data alphabetical etc. Allowed values: training_order, input_order, longest, shortest, alphabetical, reverse_alphabetical, ascending, descending.
extract_raw		bool	Extract the raw data from the wrangle.
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
output_format		string	Format of the extract output Allowed values: `list`, `dictionary`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
include_empty_labels		boolean	Include labels with no found values in the output when using use_labels=True
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

Date	Output
1992-08-13 00:00:00	August

Parameter	Required	Data Type	Notes
input	✓	str
output		str	If not given, overwrites input
property	✓	day / day_of_year / month / month_name / weekday / week_day_name / week_year / quarter	Property to extract from date
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

Start	End	Output
1992-08-13 00:00:00	2023-08-13 00:00:00	371

Parameter	Required	Data Type	Notes
start_time	✓	str	Name of the start date column
end_time	✓	str	Name of the end date column
output	✓	str	Name of the output column
range		See below for options	Type of frequency to count, defaults to day
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

HTML	Text
`<a href="https://www.wrangleworks.com/">Wrangle Works!</a>`	Wrangle Works!

HTML	Links
`<a href="https://www.wrangleworks.com/">Wrangle Works!</a>`	['https://www.wrangleworks.com/']

Parameter	Required	Data Type	Notes
input	✓	str, list
output		str, list
data_type	✓	text / links	The type of data to extract
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
output_format		string	Format of the extract output Allowed values: `list`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

Products	Product Properties
Stainless Steel Blue Bottle	`{'Colours': ['Blue'], 'Materials': ['Stainless Steel']}`
Plastic Yellow Bottle	`{'Colours': ['Yellow'], 'Materials': ['Plastic']}`

Products	Product Colours
Stainless Steel Blue Bottle	['Blue']
Plastic Yellow Bottle	['Yellow']

Parameter	Required	Data Type	Notes
input	✓	str, list
output	✓	str, list
property_type		str	The specific type of properties to extract (Colours, Materials, Shapes, Standards). Defaults to None, which will return all properties.
return_data_type		str	The format to return the data (list or string), defaults to list
first_element		bool	Get the first element from results, defaults to False.
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
output_format		string	Format of the extract output Allowed values: `list`, `dictionary`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

Product	GPM
3.4 gpm water pump	3.4 gpm
2gpm water pump	2gpm

Product	GPM
3.4 gpm water pump for 5.5 gallon tank	3.4 Gallons Per Minute
2gpm water pump for 2 gal tank	2 Gallons Per Minute

Parameter	Required	Data Type	Notes
input	✓	str
output	✓	str
find	✓	str	Pattern to find using regex
output_pattern		str	Specifies the format to output matches and specific capture groups using backreferences (e.g., `\1`, `\2`). Default is to return entire matches.
first_element		bool	Get the first element from results, defaults to False.
where		str	Filter the data to only apply the wrangle to certain rows using an equivalent to a SQL where criteria, such as column1 = 123 OR column2 = 'abc'
output_format		string	Format of the extract output Allowed values: `list`, `columns`, `concatenate`.
char		string	Character to use when output_format is concatenate
where_params		str	Variables to use in conjunctions with where. This allows the query to be parameterized. This uses sqlite syntax (? or :name)
if		str	A condition that will determine whether the action runs or not as a whole.

¶ Address

¶ Extracting Street Name

¶ AI

¶ Use a Saved Definition

¶ Extraction Output Formats

¶ Attributes

¶ Extracting All Attributes

¶ Extracting Specific Attributes

¶ Brackets

¶ Extracting Data Within Brackets

¶ Codes

¶ Extracting Codes From Single Column

¶ Extracting Codes From Multiple Columns

¶ Custom

¶ Extracting Wood Types From Single Column

¶ Extracting Wood Types From Multiple Columns

¶ Using Multiple Extract Models

¶ Date Properties

¶ Extracting Month From Date

¶ Date Range

¶ Extracting Number of Months From Range

¶ HTML

¶ Extracting Text From HTML

¶ Extracting Links From HTML

¶ Properties

¶ Extracting All Properties

¶ Extracting Colour

¶ Regex

¶ Extracting Number of Months From Range

¶ Implementing output_pattern