XBRL and Financials rewrite

[dgunning]

I pushed a rewrite of the XBRL Parser and Financials. The main features are

• Generate financials statements exactly as published by companies
• Extract all statements not just financials e.g. APPL 10-K has 78 statements
• Easier querying of XBRL data

Here is a link to a notebook with some of the features
Reading Data from XBRL

Feedback requested

[dgunning]

The new XBRL API allows you to get a statement by name or index

[amitgandhinz]

income_statement.get_concept("us-gaap:RevenueFromContractWithCustomerExcludingAssessedTax")

When I try this on TSLA's latest 10-Q (Q2) I get an error

/edgar/xbrl/xbrldata.py", line 535, in get_concept
    name=results.concept.iloc[0],
AttributeError: 'str' object has no attribute 'iloc'

[amitgandhinz]

created an issue for this here: #88

[tjf2007]

This is really excellent, Dwight, thank you.

Three questions for you:

1. Is there an easy way to get the full history of financial statements and for quarterly intervals?
2. How do you think about dealing with the re-statements point-in-time?
3. For Quarterly financial statement data with companies that publish Q1, Q2, Q3, and FY values, do you derive what the Q4 value is (e.g., Q4 = FY - [Q1 + Q2 + Q3])? An example is IBM

[tjf2007]

What do you think is the best way to get the explicit reference period for the Financials information?

For example:
Let's say I want to see the point-in-time snapshot of IBM's balance sheet, as of 2010-02-23 (the acceptance date for their 10-K submitted with ain: '0001047469-10-001151'):

ain = '0001047469-10-001151' # IBM 10-K, accepted on 2011-02-22T13:08:33.000Z
filing = get_by_accession_number(ain)
xbrl_data: XBRLData = XBRLData.extract(filing)
financials = Financials(xbrl_data)
bs = financials.get_balance_sheet()
print(bs.periods). # ['2009', '2008']```

What I'd really like to see is the "as of" reference period, so instead of ['2009', '2008'], it would be the actual fiscal period as of dates, 12/31/2009 and 12/31/2008. That will be more important when stitching together quarterly data, too.

[dgunning]

Gotcha. You would prefer to see the dates rather that 2023, 2024 as the column headers?
I could change it to that.