A Go library to parse xbrl documents into their facts, contexts, and units.

This library is based around the XBRL 2.1 spec. It implements support for parsing basic facts (not tuples of facts), contexts and units through the xml.Unmarshaler interface.

See the package example in the godocs for how to unmarshal into the XBRL struct.

This library supports basic validation that checks for malformed facts and broken references between facts and contexts/units (see XBRL.Validate()), but it does not implement full semantic validation of XBRL documents.

There are no abstractions added on-top of the XBRL data structure, which makes this library flexible and simple, but it also means you might have to read up a bit on how XBRL works to take full advantage of it.

To give you a head start, here's some basics about XBRL:

At a high level, XBRL is an XML spec intended to be the standard for digital business reporting.

Digital business reporting is a very broad topic and encompasses many things from financial statements, to compliance reporting and more.

XBRL is able to support so many use-cases by defining a framework and relying on supplemental taxonomies to describe schemas for specific use cases. For example the US GAAP Financial Reporting Taxonomy defines schemas for facts that relate to US GAAP financial reporting, which is used heavily in quarterly reports submitted to the SEC, among many other things.

Let's jump right into the main components of an XBRL document...

The core of XBRL documents are facts (also referred to as items). Facts represent a single business measurement. Here's an example of a fact whose schema is defined in the US GAAP Financial Reporting Taxonomy:

<us-gaap:EarningsPerShareBasic contextRef="c1" decimals="2" unitRef="u1">1.41</us-gaap:EarningsPerShareBasic>

A fact by itself is only a fragment of a useful piece of information. In the above example we see that earnings per share is 1.41, but we need more context around when this fact was true, and how it was measured.

In many cases (for example in an SEC quarterly report), there might be more than one instance of a us-gaap:EarningsPerShareBasic fact in the document (or any fact for that matter). This happens because a quarterly report compares EPS for the current quarter with EPS from previous quarter.

The above fact doesn't directly tell us in which quarter EPS was 1.41. That's what contexts are for...

A Context describes a business entity, period of time, and an optional scenario (this library doesn't currently support scenarios, so we're going to gloss over them).

When a fact references a context, it gives the fact more detail to help us understand what it means.

Note that many facts can reference the same context where it makes sense to do so.

The fact in the above example references a context called "c1", let's see what that context might look like:

<context id="c1">
    <entity>
        <identifier scheme="http://www.sec.gov/CIK">0000320193</identifier>
    </entity>
    <period>
        <startDate>2020-12-27</startDate>
        <endDate>2021-03-27</endDate>
    </period>
</context>

With the information in this context, we now know that in Q1 of 2021 (between 2020-12-27 and 2021-03-27), Apple Inc.'s (CIK 0000320193) EPS was 1.41.

We're closer to having a useful piece of information now, but there's one thing we're still missing. EPS is 1.41...what? What unit are we measuring it in?

A Unit describes a unit of measure for a numeric fact.

A unit can represent something simple like number of shares, something slightly more complex like dollars per share, or any other kind of unit you can think of.

Just like contexts, more than one fact can reference the same unit when it makes sense to do so.

Note that only numeric facts have units. Sometimes a fact is a block of text, which doesn't make sense to have a unit.

Let's look at what a simple unit like number of shares might look like:

<unit id="shares">
    <measure>shares</measure>
</unit>

That's great and all, but the fact in the above example references a unit called "u1". Let's see what that more complex unit might look like:

<unit id="u1">
    <divide>
        <unitNumerator>
            <measure>iso4217:USD</measure>
        </unitNumerator>
        <unitDenominator>
            <measure>shares</measure>
        </unitDenominator>
    </divide>
</unit>

This unit is a ratio of two simple units: USD / shares.

And with that we now fully understand the fact from the example above:

In Q1 of 2021 (between 2020-12-27 and 2021-03-27), Apple Inc.'s (CIK 0000320193) EPS was 1.41 dollars per share.

That was a very brief overview of the XBRL format, hopefully it empowers you enough to understand the basics of how to get information out of an XBRL document.

If you need to dig a little deeper, the models in this library are well documented and contain links to their definitions in the XBRL spec for your reference. Beyond that, the official spec is a bit large, but it's pretty clear and will almost definitely have the information you need (and probably a lot you don't need too!).