ixbrl-parse

Introduction

• Python library code, parses iXBRL files.
• Script ixbrl-dump emits iXBRL tagged data in a semi-human-readable dump.
• Script ixbrl-report emits iXBRL tagged data in a human-readable report. This involves downloading referenced XBRL schema to get the human-readable fact labels.
• Script ixbrl-to-rdf emits iXBRL tagged data in RDF.
• Script ixbrl-to-csv outputs iXBRL tagged data in a CSV format.
• Script ixbrl-to-json emits iXBRL tagged data in JSON.
• Script ixbrl-to-xbrl emits iXBRL tagged data as an XBRL instance.
• Script ixbrl-to-kv emits iXBRL in a key-value form easily consumed by scripts.
• Script ixbrl-markdown emits iXBRL tagged data in a markdown report.

Sample data

There's a bunch of sample iXBRL files grabbed from various places in the ixbrl directory: US 10-K and 10-Q filings, a few random things from UK companies house, and a couple of sample ESEF filings. This is the data I've tested with.

Also, accts.html is a sample file created using gnucash-ixbrl.

Installation

From PyPI (when published)

pip install ixbrl-parse

From source

pip install git+https://github.com/cybermaggedon/ixbrl-parse

For development

git clone https://github.com/cybermaggedon/ixbrl-parse
cd ixbrl-parse
pip install -e ".[dev]"

The dev extra includes pytest and pytest-cov for running tests.

For markdown report support:

pip install ixbrl-parse[markdown]

Usage

Parse iXBRL and output in RDF (default n3 format):

ixbrl-to-rdf accts.html

Parse iXBRL and output in RDF/XML:

ixbrl-to-rdf accts.html --format xml

Parse iXBRL and output in CSV:

ixbrl-to-csv accts.html

Parse iXBRL and output in JSON:

ixbrl-to-json accts.html

Schema labels in JSON:

ixbrl-to-json ixbrl/10k/lyft-20201231.htm -f labeled \
    -b https://www.sec.gov/Archives/edgar/data/1759509/000175950921000011/lyft-20201231.htm

Dump iXBRL values:

ixbrl-dump accts.html

Human-readable report:

ixbrl-report accts.html

Human-readable report from SEC EDGAR. Note need to tell ixbrl-report the URL of the original report in order to know where to fetch the custom schema (relative URLs are used):

ixbrl-report ixbrl/10k/lyft-20201231.htm \
    -b https://www.sec.gov/Archives/edgar/data/1759509/000175950921000011/lyft-20201231.htm

Dump iXBRL as XBRL:

ixbrl-to-xbrl accts.html

API

You can use the library directly in your Python code:

from lxml import etree as ET
from ixbrl_parse.ixbrl import parse

# Parse an iXBRL file
tree = ET.parse('accts.html')
ixbrl = parse(tree)

# Get data in various formats
data_dict = ixbrl.to_dict()
flat_data = ixbrl.flatten()
rdf_triples = ixbrl.get_triples()

# Access contexts and values
for context in ixbrl.contexts.values():
    print(context.entity, context.period)

for value in ixbrl.values.values():
    print(value.name, value.to_value())

See the CLI implementations in ixbrl_parse/cli.py for more examples.

Development

Running Tests

The project includes a comprehensive pytest test suite:

# Run all tests
pytest

# Run with coverage
pytest --cov=ixbrl_parse --cov-report=term-missing

# Run specific test file
pytest tests/test_ixbrl.py -v

See tests/README.md for more details on the test suite.

Building

The project uses modern Python packaging with pyproject.toml:

# Build wheel and source distribution
python -m build

# Install in development mode
pip install -e .

What next?

This loads into a Redland RDF sqlite3 store:

ixbrl-to-rdf -i accts.html -f ntriples > accts.ntriples
rdfproc -n -s sqlite accts.db parse accts.ntriples ntriples
rdfproc -s sqlite accts.db print | head

I run a SPARQL store across the data, and view it with LodLive.