A "filing" on EDGAR isn't a single document. It's a container. Understanding the container saves you a lot of confusion when the press release you're looking for isn't where you expected.
The accession number
Every filing gets a unique ID that looks like 0000320193-25-000079. It breaks into three parts:
0000320193: the filer's CIK (Apple, in this case)25: the year, roughly000079: a sequence number assigned by the filer's EDGAR agent
You'll see it written with or without dashes. SEC URLs strip the dashes; API calls take the dashed form.
The container: one big .txt file
EDGAR submissions are delivered as a single SGML text file, the "complete submission." It has two parts:
- A structured header with filer info: form type, company name, CIK, period of report, and for 8-Ks the item numbers being reported.
- Every member document concatenated together, each wrapped in
<DOCUMENT>tags with a filename and sequence number.
The header is boring but useful. The ITEM INFORMATION block on an 8-K is where the item numbers (2.02, 5.02, and so on) are declared.
Member documents
Inside the container you get a handful of files, and what they are depends on the form:
- The primary document: the filing itself, usually an .htm named something like
aapl-20250927.htm - Exhibits:
EX-99.1,EX-10.1, and friends. Press releases, contracts, material agreements. For an earnings 8-K, the actual earnings release is an EX-99 exhibit, which trips people up when they search only the main document. - XBRL files: instance documents, the Financial Report
R*.htmrenderings, andFilingSummary.xmlwhich lists every rendered report. - The submission .txt itself: the container as filed.
The filing's directory also exposes index.json, which lists the member files with sizes and timestamps. SEC's URL scheme is predictable:
https://www.sec.gov/Archives/edgar/data/{cik}/{accession-no-dashes}/{filename}
Sections vs. documents
Two ways to slice the same filing, and the difference matters:
- Documents are the files. The main doc plus exhibits.
- Sections are the carved-up text inside the main doc: Item 1A (Risk Factors), Item 7 (MD&A), and so on.
A company's press release is a document (an exhibit). Its risk factors are a section (inside the primary document). When you're searching filing text, "scope" means choosing which of those to look at.
How this maps to the API
Each layer has an endpoint:
GET /v1/filings/0000320193-25-000079 # the filing + everything embedded
GET /v1/filings/0000320193-25-000079/sgml # the raw container, header + all docs
GET /v1/filings/0000320193-25-000079/sections # item-level text
GET /v1/filings/.../documents/aapl-20250927.htm # a member document
The filing response itself embeds the document list, section list, parsed XBRL facts, and form-specific tables like 13F holdings and insider transactions. The docs have the full shape.