mirror of
https://github.com/K-Dense-AI/claude-scientific-skills.git
synced 2026-03-27 07:09:27 +08:00
9.2 KiB
9.2 KiB
edgartools — XBRL Reference
Table of Contents
- Core Classes
- XBRL Class
- Statements Access
- XBRLS — Multi-Period Analysis
- Facts Querying
- Statement to DataFrame
- Value Transformations
- Rendering
- Error Handling
- Import Reference
Core Classes
| Class | Purpose |
|---|---|
XBRL |
Parse single filing's XBRL |
XBRLS |
Multi-period analysis across filings |
Statements |
Access financial statements from single XBRL |
Statement |
Individual statement object |
StitchedStatements |
Multi-period statements interface |
StitchedStatement |
Multi-period individual statement |
FactsView |
Query interface for all XBRL facts |
FactQuery |
Fluent fact query builder |
XBRL Class
Creating an XBRL Object
from edgar.xbrl import XBRL
# From a Filing object (most common)
xbrl = XBRL.from_filing(filing)
# Via filing method
xbrl = filing.xbrl() # returns None if no XBRL
# From directory
xbrl = XBRL.from_directory("/path/to/xbrl/files")
# From file list
xbrl = XBRL.from_files(["/path/instance.xml", "/path/taxonomy.xsd"])
Core Properties
xbrl.statements # Statements object
xbrl.facts # FactsView object
# Convert all facts to DataFrame
df = xbrl.to_pandas()
# Columns: concept, value, period, label, ...
Statement Methods
stmt = xbrl.get_statement("BalanceSheet")
stmt = xbrl.get_statement("IncomeStatement")
stmt = xbrl.get_statement("CashFlowStatement")
stmt = xbrl.get_statement("StatementOfEquity")
# Render with rich formatting
rendered = xbrl.render_statement("BalanceSheet")
rendered = xbrl.render_statement("IncomeStatement", show_percentages=True, max_rows=50)
print(rendered)
Statements Access
statements = xbrl.statements
balance_sheet = statements.balance_sheet()
income_stmt = statements.income_statement()
cash_flow = statements.cash_flow_statement()
equity = statements.statement_of_equity()
comprehensive = statements.comprehensive_income()
All return Statement objects or None if not found.
XBRLS — Multi-Period Analysis
from edgar import Company
from edgar.xbrl import XBRLS
company = Company("AAPL")
# Get multiple filings (use amendments=False for clean stitching)
filings = company.get_filings(form="10-K", amendments=False).head(3)
# Stitch together
xbrls = XBRLS.from_filings(filings)
# Access stitched statements
stitched = xbrls.statements
income_stmt = stitched.income_statement()
balance_sheet = stitched.balance_sheet()
cashflow = stitched.cashflow_statement()
equity_stmt = stitched.statement_of_equity()
comprehensive = stitched.comprehensive_income()
StitchedStatements Parameters
All methods accept:
max_periods(int) — max periods to include (default: 8)standard(bool) — use standardized concept labels (default: True)use_optimal_periods(bool) — use entity info for period selection (default: True)show_date_range(bool) — show full date ranges (default: False)include_dimensions(bool) — include segment data (default: False)view(str) —"standard","detailed", or"summary"(overridesinclude_dimensions)
# Standard view (default)
income = stitched.income_statement()
# Detailed view with dimensional breakdowns
income_detailed = stitched.income_statement(view="detailed")
# Convert to DataFrame (periods as columns)
df = income.to_dataframe()
Facts Querying
FactsView — Starting a Query
facts = xbrl.facts
# Query by concept
revenue_q = facts.by_concept("Revenue")
revenue_q = facts.by_concept("us-gaap:Revenue", exact=True)
# Query by label
rd_q = facts.by_label("Research", exact=False)
# Query by value range
large_q = facts.by_value(min_value=1_000_000_000)
small_q = facts.by_value(max_value=100_000)
range_q = facts.by_value(min_value=100, max_value=1000)
# Query by period
period_q = facts.by_period(start_date="2023-01-01", end_date="2023-12-31")
FactQuery — Fluent Chaining
# Chain multiple filters
query = (xbrl.facts
.by_concept("Revenue")
.by_period(start_date="2023-01-01")
.by_value(min_value=1_000_000))
# Execute
facts_list = query.execute() # List[Dict]
facts_df = query.to_dataframe() # DataFrame
first_fact = query.first() # Dict or None
count = query.count() # int
# Filter by statement type
income_facts = xbrl.facts.by_statement("IncomeStatement")
Analysis Methods on FactsView
# Pivot: concepts as rows, periods as columns
pivot = facts.pivot_by_period(["Revenue", "NetIncomeLoss"])
# Time series for a concept
revenue_ts = facts.time_series("Revenue") # pandas Series
# Convert all to DataFrame
all_df = facts.to_dataframe()
Statement to DataFrame
Statement.to_dataframe()
statement = xbrl.statements.income_statement()
# Raw mode (default) — exact XML values
df_raw = statement.to_dataframe()
# Presentation mode — matches SEC HTML display
df_presentation = statement.to_dataframe(presentation=True)
# Additional options
df = statement.to_dataframe(
include_dimensions=True, # include segment breakdowns (default: True)
include_unit=True, # include unit column (USD, shares)
include_point_in_time=True # include point-in-time column
)
Columns in output
- Core:
concept,label, period date columns - Metadata (always):
balance,weight,preferred_sign - Optional:
dimension,unit,point_in_time
Get Concept Value
revenue = statement.get_concept_value("Revenue")
net_income = statement.get_concept_value("NetIncomeLoss")
Value Transformations
edgartools provides two layers of values:
Raw Values (default): Values exactly as in XML instance document. Consistent across companies, comparable to SEC CompanyFacts API.
Presentation Values (presentation=True): Transformed to match SEC HTML display. Cash flow outflows shown as negative. Good for investor-facing reports.
statement = xbrl.statements.cash_flow_statement()
# Raw: dividends paid appears as positive
df_raw = statement.to_dataframe()
# Presentation: dividends paid appears as negative (matches HTML)
df_pres = statement.to_dataframe(presentation=True)
Metadata columns explain semantics:
balance: debit/credit from schemaweight: calculation weight (+1.0 or -1.0)preferred_sign: presentation hint (+1 or -1)
When to use each:
| Use Raw | Use Presentation |
|---|---|
| Cross-company analysis | Matching SEC HTML display |
| Data science / ML | Investor-facing reports |
| Comparison with CompanyFacts API | Traditional financial statement signs |
Rendering
# Render single statement
rendered = xbrl.render_statement("BalanceSheet")
print(rendered) # Rich formatted output
# Render Statement object
stmt = xbrl.statements.income_statement()
rendered = stmt.render()
rendered = stmt.render(show_percentages=True, max_rows=50)
print(rendered)
# Multi-period render
stitched_stmt = xbrls.statements.income_statement()
rendered = stitched_stmt.render(show_date_range=True)
print(rendered)
Advanced Examples
Complex Fact Query
from edgar import Company
from edgar.xbrl import XBRL
company = Company("MSFT")
filing = company.latest("10-K")
xbrl = XBRL.from_filing(filing)
# Query with multiple filters
results = (xbrl.facts
.by_concept("Revenue")
.by_value(min_value=50_000_000_000)
.by_period(start_date="2023-01-01")
.to_dataframe())
# Pivot analysis
pivot = xbrl.facts.pivot_by_period([
"Revenue",
"NetIncomeLoss",
"OperatingIncomeLoss"
])
Cross-Company Comparison
from edgar import Company
from edgar.xbrl import XBRL
companies = ["AAPL", "MSFT", "GOOGL"]
for ticker in companies:
company = Company(ticker)
filing = company.latest("10-K")
xbrl = XBRL.from_filing(filing)
if xbrl and xbrl.statements.income_statement():
stmt = xbrl.statements.income_statement()
revenue = stmt.get_concept_value("Revenue")
print(f"{ticker}: ${revenue/1e9:.1f}B")
Error Handling
from edgar.xbrl import XBRL, XBRLFilingWithNoXbrlData
try:
xbrl = XBRL.from_filing(filing)
except XBRLFilingWithNoXbrlData:
print("No XBRL data in this filing")
# Check availability
xbrl = filing.xbrl()
if xbrl is None:
print("No XBRL available")
text = filing.text() # fallback
# Check statement availability
if xbrl and xbrl.statements.income_statement():
income = xbrl.statements.income_statement()
df = income.to_dataframe()
Import Reference
# Core
from edgar.xbrl import XBRL, XBRLS
# Statements
from edgar.xbrl import Statements, Statement
from edgar.xbrl import StitchedStatements, StitchedStatement
# Facts
from edgar.xbrl import FactsView, FactQuery
from edgar.xbrl import StitchedFactsView, StitchedFactQuery
# Rendering & standardization
from edgar.xbrl import StandardConcept, RenderedStatement
# Utilities
from edgar.xbrl import stitch_statements, render_stitched_statement, to_pandas