Skip to main content

a python parser for the .fec file format

Project description

This is a library for converting campaign finance filings stored in the .fec format into native python objects. It maps the comma/ASCII 28 delimited fields to canonical names based on the version the filing uses and then converts the values that are dates and numbers into the appropriate int, float, or datetime objects.

This library is in relatively early testing. I've used it on a couple of projects, but I wouldn't trust it to work on all filings. That said, if you do try using it, I'd love to hear about it!


The FEC makes a ton of data available via the "export" links on the main site and the developer API. For cases where those data sources are sufficient, they are almost certainly the easiest/best way to go. A few cases where one might need to be digging into raw filings are:

  • Getting information from individual itemizations including addresses. (The FEC doesn't include street addresses in bulk downloads.)

  • Getting data as soon as it has been filed, instead of waiting for it to be coded. (The FEC generally codes all filings received by 7pm eastern by 7am the next day. However, that means that a filing received at 11:59pm on Monday wouldn't be available until 7am on Wednesday, for example.)

  • Getting more data than the rate-limit on the developer API would allow.

  • Maintaining one's own database with all relevant campaign finance data, perhaps synced with another data source.

Raw filings can be found by either downloading the bulk data zip files or from http requests like this. This library includes helper methods for both.


To get started, install from pypi by running the following command in your preferred terminal:

pip install fecfile


For the vast majority of filings, the easiest way to use this library will be to load filings all at once by using the from_http(file_number), from_file(file_path), or loads(input) methods.

These methods will return a Python dictionary, with keys for header, filing, itemizations, and text. The itemizations dictionary contains lists of itemizations grouped by type (Schedule A, Schedule B, etc.).


import fecfile

filing1 = fecfile.from_file('1229017.fec')


filing2 = fecfile.from_http(1146148)


filing3 = fecfile.from_http(1146148)

all_contributions = filing3['itemizations']['Schedule B']

mid_size_contributions = [item for item in all_contributions if 500 <= item[contribution_amount] < 1000]


with open('1229017.fec') as file:

    parsed = fecfile.loads(

    num_disbursements = len(parsed['itemizations']['Schedule B'])


url = ''

r = requests.get(url, headers={'User-Agent': 'Mozilla/5.0'})

parsed = fecfile.loads(r.text)


Note: the urls cause problems with the requests library when a user-agent is not supplied. There may be a cleaner fix to that though.

Advanced Usage

FEC filings can be arbitrarily large. Loading enormous filings into memory all at once can cause problems (including running out of memory).

The fecfile library exposes the iter_file and iter_http methods to read large filings one line at a time. Both are generator functions that yield FecItem objects, which consist of data and data_type attributes. The data_type attribute can be one of "header", "summary", "itemization", "text", or "F99_text". The data attribute is a dictionary for all data types except for "F99_text", for which it is a string.

import fecfile

import imaginary_database

# Sometimes we only care about summary data, but want to be able to handle all filings, without

# knowing anything about them before we attempt to parse.

no_itemizations = {'filter_itemizations': []}

for i in range(1300000, 1320000):

    for item in fecfile.iter_http(i, options=no_itemizations):

        if item.data_type == 'summary':


# Sometimes we only care about one type of itemization, but from a very large filing.

# In this example, we add up all the contributions from Delaware in ActBlue's 2018

# post-general filing

only_contributions = {'filter_itemizations': ['SA']}

de_total = 0

for item in fecfile.iter_http(1300352, options=only_contributions):

    if item.data_type == 'itemization':

        if['contributor_state'] == 'DE':

            de_total +=['contribution_amount']


# Sometimes we want to maintain a database where different types of itemizations live in their own

# tables and have foreign key relationships to a summary record.

file_path = '/path/to/99840.fec'

filing = None

for item in fecfile.iter_file(file_path):

    if item.data_type == 'summary':

        filing = imaginary_database.add_filing(file_number=99840, **

    if item.data_type == 'itemization':


            imaginary_database.add_contribution(filing=filing, **


            imaginary_database.add_disbursement(filing=filing, **


            imaginary_database.add_loan(filing=filing, **

You can also choose to use the parse_header and parse_line methods if you are implementing a different method of

iterating over a filing's content. Before version 0.6, the below example was the only way to use fecfile to parse

filings without loading the entire filing into memory. This approach should no longer be necessary, but is kept to

show how example usage for those methods.

import fecfile

version = None

with open('1263179.fec') as file:

    for line in file:

        if version is None:

            header, version = fecfile.parse_header(line)


            parsed = fecfile.parse_line(line, version)


API Reference


loads(input, options={})

Deserialize input (a str instance

containing an FEC document) to a Python object.

Optionally, pass an array of strings to options['filter_itemizations'].

If included, loads will only parse lines that start with any of the

strings in that array. For example, passing

{'filter_itemizations': ['SC', 'SD']} to options, will only include

Schedule C and Schedule D itemizations. Also, passing

{'filter_itemizations': []} to options will result in only the header

and the filing being parsed and returned.

Including {'as_strings': True} in the options dictionary will not attempt to convert values that are normally numeric or datetimes to their native python types and will return dictionaries with all values as strings.



Deserialize a str or a list of str instances containing

header information for an FEC document. Returns an Python object, the

version str used in the document, and the number of lines used

by the header.

The third return value from parse_header--the number of lines used by the header--is only

useful for early versions of the FEC file format, typically predating 2001. Versions 1 and 2 of the FEC file format allowed headers to be a multiline string beginning and ending with /*.

Returning the number of lines in the header allows us to know where the non-header lines begin.


parse_line(line, version, line_num=None)

Deserialize a line (a str instance

containing a line from an FEC document) to a Python object.

version is a str instance for the version of the FEC file format

to be used, and is required.

line_num is optional and is used for debugging. If an error or

warning is encountered, whatever is passed in to line_num will be

included in the error/warning message. Normally the line number of the input file will be passed in, so that the user is shown the error and the line number in the original file that triggered the error.


from_http(file_number, options={})

Utility method for retrieving a parsed Python representation of an FEC

filing when it is not available as a local file. This method takes

either a str or int as a file_number and requests the corresponding filing from

the server. It returns the parsed response.

See above for how documentation on how to use the optional

options argument.


from_file(file_path, options={})

Utility method for getting a parsed Python representation of an FEC

filing that exists as a .fec file on a local machine. This method takes

a str of the path to the file, and returns the parsed Python object.

See above for how documentation on how to use the optional

options argument.


iter_http(file_number, options={})

Makes an http request for the given file_number and iterates over the response, yielding FecItem instances, which consist of data and data_type attributes. The data_type attribute can be one of "header", "summary", "itemization", "text", or "F99_text". The data attribute is a dictionary for all data types except for "F99_text", for which it is a string. This method avoids loading the entire filing into memory, as the from_http method does.

See above for how documentation on how to use the optional

options argument.


iter_file(file_path, options={})

Opens a file at the given file_path and iterates over its contents, yielding FecItem instances, which consist of data and data_type attributes. The data_type attribute can be one of "header", "summary", "itemization", "text", or "F99_text". The data attribute is a dictionary for all data types except for "F99_text", for which it is a string. This method avoids loading the entire filing into memory, as the from_file method does.

See above for how documentation on how to use the optional

options argument.



Utility method for debugging - prints out a representative subset of

the Python object returned by one of the deserialization methods. For

filings with itemizations, it only prints the first of each type of

itemization included in the object.

Developing locally

Assuming you already have Python3 and the ability to create virtual environments installed, first clone this repository from github and cd into it:

git clone

cd fecfile

Then create a virtual environment for this project (I use the following commands, but there are several ways to get the desired result):

python3 -m venv env

source env/bin/activate

Next, install the dependencies:

python install

Finally, make some changes, and run:



This project would be impossible without the work done by the kind folks at The New York Times Newsdev team. In particular, this project relies heavily on fech.

Many thanks to Derek Willis for creating and maintaining fech-sources which serves as the upstream source of truth for parsing libaries in various languages.

A huge thank you to Chris Zubak-Skees, whose work on the fec-parse javascript library inspired this project.

Many thanks to Jacob Fenton for writing the caching logic and for providing valuable feedback about the overall design of this library.


I would love some help with this, particularly with the mapping from strings to int, float, and datetime types. Please create an issue or make a pull request. Or reach out privately via email - that works too.

To do:

Almost too much to list:

  • Handle files from before v6 when they were comma-delimited

  • create a dumps method for writing .fec files for round-trip tests

  • add more types to the types.json file

  • elegantly handle errors


See the changelog for a list of notable changes introduced in each version of fecfile.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

fecfile-0.7.2.tar.gz (34.4 kB view hashes)

Uploaded source

Built Distribution

fecfile-0.7.2-py3-none-any.whl (30.5 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page