paperhawk / schemas /financial_report.json
Nándorfi Vince
Initial paperhawk push to HF Space (LFS for binaries)
7ff7119
raw
history blame
1.59 kB
{
"type": "object",
"title": "FinancialReportSchema",
"description": "Structured-extraction schema for financial reports (P&L, balance sheet, cash flow).",
"properties": {
"report_type": {
"type": ["string", "null"],
"description": "e.g. income_statement (P&L), balance_sheet, cash_flow"
},
"period_start": { "type": ["string", "null"] },
"period_end": { "type": ["string", "null"] },
"company_name": { "type": ["string", "null"] },
"company_tax_id": { "type": ["string", "null"] },
"currency": { "type": "string", "default": "USD" },
"accounting_standard": {
"type": ["string", "null"],
"description": "IFRS, US-GAAP, HU-GAAP, DE-HGB, or other"
},
"line_items": {
"type": "array",
"items": {
"type": "object",
"properties": {
"description": { "type": "string" },
"value": { "type": ["number", "null"] },
"value_prior_period": { "type": ["number", "null"] }
}
}
},
"revenue": { "type": ["number", "null"] },
"operating_income": { "type": ["number", "null"] },
"pretax_income": { "type": ["number", "null"] },
"tax": { "type": ["number", "null"] },
"net_income": { "type": ["number", "null"] },
"_quotes": { "type": "array", "items": { "type": "string" } },
"_confidence": { "type": "object" },
"_source": {
"type": "object",
"properties": {
"file_name": { "type": "string" },
"page_number": { "type": ["integer", "null"] }
}
}
},
"required": ["_quotes", "_confidence"]
}