Toolbelt of classes and functions written in Python to use with the Federal Register API. The Federal Register (FR) is the daily journal of the U.S. government, containing rules, proposed rules, notices, and presidential documents from federal agencies.
Name inspired by the Requests Toolbelt package. "FR" is a common shorthand for the Federal Register, in part because document citations take the form of VOLUME FR STARTPAGE (e.g., 88 FR 21879).
Install using pip:
pip install fr-toolbeltThis package was developed with Python 3.12 and requires Python 3.10 or higher.
You can run the package as a module to get an illustration of how it processes documents:
python -m fr_toolbeltGenerally, you'll want to import specific classes or functions to meet your project's needs. More on what each module offers below.
The FR toolbelt package contains several modules:
api_requestsfor requesting documents from the Federal Register API.preprocessingfor processing Federal Register documents to further analyze.utilsfor providing utility classes and functions to the other modules
The api_requests module makes it easy to retrieve FR documents from the API by specifying a date range or providing a list of document numbers (document_number is the unique identifier for each document).
A simple request for documents published in January 2024 requires a start and end date and returns a tuple of the list of results and the count of retrieved documents.
from fr_toolbelt.api_requests import get_documents_by_date
start = "2024-01-01"
end = "2024-01-31"
results, count = get_documents_by_date(start, end)To only retrieve documents of a particular type (or types), pass the document_types parameter to the function call.
start = "2024-01-01"
end = "2024-01-31"
types = ["RULE", "PRORULE"]
results, count = get_documents_by_date(start, end, document_types=types)To deviate from the default set of fields, pass the fields parameter to the function call.
start = "2024-01-01"
end = "2024-01-31"
fields = ["document_number", "publication_date", "raw_text_url"]
results, count = get_documents_by_date(start, end, fields=fields)More customization is possible by examining the parameters and docstrings. Note that the get_documents_by_date function works around the FR API's maximum of 10,000 results per search by querying smaller subsets of documents and compiling them into a larger result set. So retrieving all 28,308 documents published in 2020 is now possible with a single function call.
To collect a particular set of documents, pass their document numbers as a parameter.
from fr_toolbelt.api_requests import get_documents_by_number
document_numbers = ["2024-02204", "2023-28203", "2023-25797"]
results, count = get_documents_by_number(document_numbers)The api_requests module may add support for endpoints other than the documents endpoint at a future point.
The preprocessing module handles common tasks to process the API data in a usable format. Below is an example of what the raw API data look like for a single illustrative document. Notice how fields like "agencies" and "regulation_id_number_info" are nested data structures that are difficult to use in their raw form.
{'agencies': [{'id': 12,
'json_url': 'https://www.federalregister.gov/api/v1/agencies/12',
'name': 'Agriculture Department',
'parent_id': None,
'raw_name': 'DEPARTMENT OF AGRICULTURE',
'slug': 'agriculture-department',
'url': 'https://www.federalregister.gov/agencies/agriculture-department'},
{'id': 456,
'json_url': 'https://www.federalregister.gov/api/v1/agencies/456',
'name': 'Rural Business-Cooperative Service',
'parent_id': 12,
'raw_name': 'Rural Business-Cooperative Service',
'slug': 'rural-business-cooperative-service',
'url': 'https://www.federalregister.gov/agencies/rural-business-cooperative-service'}],
'agency_names': ['Agriculture Department',
'Rural Business-Cooperative Service'],
'docket_ids': ['DOCKET #: RBS-23-BUSINESS-0024'],
'dockets': [{'agency_name': 'RBS',
'documents': [{'allow_late_comments': None,
'comment_count': 1,
'comment_end_date': '2024-04-02',
'comment_start_date': '2024-01-02',
'comment_url': 'https://www.regulations.gov/commenton/RBS-23-BUSINESS-0024-0001',
'id': 'RBS-23-BUSINESS-0024-0001',
'regulations_dot_gov_open_for_comment': True,
'updated_at': '2024-01-22T00:04:26.978-05:00'}],
'id': 'RBS-23-BUSINESS-0024',
'supporting_documents': [],
'supporting_documents_count': 0,
'title': 'Notice of Funding Opportunity for the Rural Innovation '
'Stronger Economy (RISE) Grant Program for Fiscal Year '
'2024'}],
'document_number': '2023-26792',
'president': {'identifier': 'joe-biden', 'name': 'Joseph R. Biden Jr.'},
'publication_date': '2024-01-02',
'regulation_id_number_info': {},
'title': 'Notice of Solicitation of Applications for the Rural Innovation '
'Stronger Economy (RISE) Grant Program for Fiscal Year 2024',
'type': 'Notice'}To preprocess the agency information in a set of documents, we would use the AgencyMetadata class to retrieve agency metadata from the API and then process the documents with the AgencyData class.
from fr_toolbelt.preprocessing import AgencyMetadata, AgencyData
# first we collect metadata for processing the agency information
agency_metadata = AgencyMetadata()
metadata, schema = agency_metadata.get_agency_metadata()
# then we process the documents using the AgencyData class
agency_data = AgencyData(results, metadata, schema)
processed_agencies = agency_data.process_data(return_format="name")Below, see how the illustrative document shown previously now contains new key: value pairs ("agency_slugs", "independent_reg_agency", "parent_name", "subagency_name") and removes the old ones ("agencies", "agency_names").
{'agency_slugs': ['rural-business-cooperative-service',
'agriculture-department'],
'docket_ids': ['DOCKET #: RBS-23-BUSINESS-0024'],
'dockets': [{'agency_name': 'RBS',
'documents': [{'allow_late_comments': None,
'comment_count': 1,
'comment_end_date': '2024-04-02',
'comment_start_date': '2024-01-02',
'comment_url': 'https://www.regulations.gov/commenton/RBS-23-BUSINESS-0024-0001',
'id': 'RBS-23-BUSINESS-0024-0001',
'regulations_dot_gov_open_for_comment': True,
'updated_at': '2024-01-22T00:04:26.978-05:00'}],
'id': 'RBS-23-BUSINESS-0024',
'supporting_documents': [],
'supporting_documents_count': 0,
'title': 'Notice of Funding Opportunity for the Rural Innovation '
'Stronger Economy (RISE) Grant Program for Fiscal Year '
'2024'}],
'document_number': '2023-26792',
'independent_reg_agency': False,
'parent_name': 'Agriculture Department',
'president': {'identifier': 'joe-biden', 'name': 'Joseph R. Biden Jr.'},
'publication_date': '2024-01-02',
'regulation_id_number_info': {},
'subagency_name': 'Rural Business-Cooperative Service',
'title': 'Notice of Solicitation of Applications for the Rural Innovation '
'Stronger Economy (RISE) Grant Program for Fiscal Year 2024',
'type': 'Notice'}A similar series of commands accomplishes data processing for other fields. Classes are available for preprocessing "president" (Presidents), "regulation_id_number_info" (RegInfoData), and fields related to public commenting dockets (Dockets and RegDotGovData).
Alternatively, the process_documents function provides a simpler interface for combining these functionalities together.
from fr_toolbelt.preprocessing import process_documents
# passing the del_keys parameter deletes those keys from the resulting dict
processed_docs = process_documents(results, del_keys=("type", "docket_ids")){'agency_slugs': ['rural-business-cooperative-service',
'agriculture-department'],
'docket_id': 'RBS-23-BUSINESS-0024',
'document_number': '2023-26792',
'independent_reg_agency': False,
'parent_slug': 'agriculture-department',
'president_id': 'joe-biden',
'publication_date': '2024-01-02',
'rin': None,
'rin_priority': None,
'subagency_slug': 'rural-business-cooperative-service',
'title': 'Notice of Solicitation of Applications for the Rural Innovation '
'Stronger Economy (RISE) Grant Program for Fiscal Year 2024'}You can select which fields you want to process with this function too.
from fr_toolbelt.preprocessing import process_documents
# passing the which parameter to select which fields to process
processed_docs = process_documents(results, which=("agencies", "presidents"), del_keys="docket_ids"){'agency_slugs': ['rural-business-cooperative-service',
'agriculture-department'],
'dockets': [{'agency_name': 'RBS',
'documents': [{'allow_late_comments': None,
'comment_count': 1,
'comment_end_date': '2024-04-02',
'comment_start_date': '2024-01-02',
'comment_url': 'https://www.regulations.gov/commenton/RBS-23-BUSINESS-0024-0001',
'id': 'RBS-23-BUSINESS-0024-0001',
'regulations_dot_gov_open_for_comment': True,
'updated_at': '2024-01-22T00:04:26.978-05:00'}],
'id': 'RBS-23-BUSINESS-0024',
'supporting_documents': [],
'supporting_documents_count': 0,
'title': 'Notice of Funding Opportunity for the Rural Innovation '
'Stronger Economy (RISE) Grant Program for Fiscal Year '
'2024'}],
'document_number': '2023-26792',
'independent_reg_agency': False,
'parent_name': 'Agriculture Department',
'president_id': 'joe-biden',
'publication_date': '2024-01-02',
'regulation_id_number_info': {},
'subagency_name': 'Rural Business-Cooperative Service',
'title': 'Notice of Solicitation of Applications for the Rural Innovation '
'Stronger Economy (RISE) Grant Program for Fiscal Year 2024',
'type': 'Notice'}These functions handle date formatting under the hood and provide functionality for identifying and removing duplicate entries (not a current bug in the FR API if passing the order=oldest or order=newest parameter in a request).