reference/ (stable, app-backed)JSON loaded at runtime (sec_edgar.reference_data, data/manifest.json).
| File | Role |
|---|---|
edgar_api_schema.json |
SEC company facts / submissions shapes and mapping notes. |
taxonomies.json |
XBRL taxonomy namespaces (us-gaap, dei, …). |
financial_model.json |
US-GAAP concept groups (tag preference) and derived KPI definitions. |
accounting_model.json |
How accounting sources merge into the canonical map. |
sic_codes.json |
SEC SIC master list. |
generated/us_gaap_account_map.json |
Generated — one entry per us-gaap concept: label, description, optional acct_category. |
reference/enrichment/)Loaded by enrichment.reference at runtime and in tests (see docs/enrichment.md).
| File | Role |
|---|---|
iso3166.json |
ISO 3166-1 countries (iso3, UN region/subregion, centroid, currency, ccTLD, EU/OECD) and ISO 3166-2 subdivisions. |
business_registries.json |
Company register per jurisdiction with its GLEIF Registration Authority code and whether it has a free API, plus the legal-form → jurisdiction table. |
A legal form is a registration fact: “GmbH” exists because a German, Austrian or Swiss registrar created it. That makes form-vs-country checkable offline, and tells you which register to query next.
reference/sources/accounting/)| File | Role |
|---|---|
acct_facts.csv |
Compact seed list (us_gaap_list, acct_label, acct_description). |
acct_facts_overlay.json |
JSON array with the full concept set; optional acct_category. Loaded after the CSV so it wins per concept. |
This replaces the older trio acct_facts.csv + acct_facts.json + acct_facts_updated.json (the two JSON files had identical concept keys; only the overlay file is kept).
Merge into the canonical reference file:
PYTHONPATH=src DJANGO_SETTINGS_MODULE=config.settings python src/manage.py sync_accounting_reference
If generated/us_gaap_account_map.json is missing, the loader falls back to acct_facts_overlay.json, then acct_facts.csv.
From the repo root (USER_AGENT_EMAIL set per SEC policy):
python src/manage.py generate_edgar_reference
Add --with-accounting to run sync_accounting_reference afterward.
samples/ (tracked, small)Example rows for documentation and local experiments: CRM-style CSV, CRM sync snippet, optional us_gaap_facts.csv.
local/ (gitignored)Place large or proprietary exports here (not committed). Examples:
companies-clean.json — default path for load_crm_companies_jsoncompanies.csv / companies.json — full CRM/ERP exportserp-clients.csv / erp-clients.json — CRM/ERP export stagingThe repo root .gitignore keeps data/local/* ignored except data/local/.gitkeep.
enrichment/Enrichment input lists and exported customer/vendor masters (see docs/enrichment.md).
| File | Role |
|---|---|
master_list_v2.csv |
Input: 10,982 manufacturing suppliers (name, footprint, verticals). |
vendor_master_v3.csv |
Export: the golden record, with provenance and trust columns. |
Rebuild the export from the database at any time:
cd src && python manage.py enrich_export --dataset manufacturing --out ../data/enrichment/vendor_master_v3.csv
hard_completeness is the column to trust — it excludes modelled values. revenue_basis keeps reported and estimated apart; unverified_website flags the most dangerous cell in the file, because it looks exactly like a good one.