Issue 1: repo scaffold and project structure

This commit is contained in:
Johnny 5
2026-06-07 19:20:02 +00:00
parent e0b31009b7
commit a70bdeaa5e
52 changed files with 6750 additions and 0 deletions
@@ -0,0 +1,437 @@
"""CSV import endpoints for entities and schedule of investments."""
import csv
import io
import re
from datetime import date, datetime
from typing import Any
from fastapi import APIRouter, Depends, File, HTTPException, Query, UploadFile
from sqlmodel import Session, select
from ten31portal.audit import record_audit
from ten31portal.auth import require_role
from ten31portal.database import get_session
from ten31portal.models import (
Entity, EntityStatus, EntityType, Holding, Position,
User, UserRole, Valuation, ValuationRound, RoundStatus,
)
router = APIRouter(prefix="/api/import", tags=["import"])
# --- Column maps ---
# These must be confirmed against real Carta CSV exports.
# Mark as UNCONFIRMED until a sample file is validated.
ENTITY_COLUMN_MAP: dict[str, str] = {
# Carta export header -> model field
# UNCONFIRMED: update these after inspecting a real Carta entities export
"Entity Name": "name",
"Entity Type": "type",
"Vintage Year": "vintage_year",
"Fund Size": "fund_size_cents",
}
ENTITY_TYPE_MAP: dict[str, EntityType] = {
"Fund": EntityType.fund,
"fund": EntityType.fund,
"SPV": EntityType.spv,
"spv": EntityType.spv,
"GP": EntityType.gp,
"gp": EntityType.gp,
"Mgmt Co": EntityType.mgmt_co,
"mgmt_co": EntityType.mgmt_co,
"Management Company": EntityType.mgmt_co,
}
SCHEDULE_COLUMN_MAP: dict[str, str] = {
# UNCONFIRMED: update after inspecting a real Carta schedule-of-investments export
"Company": "company_name",
"Security": "security_name",
"Investment Date": "investment_date",
"Shares": "shares",
"Cost": "cost_cents",
"Value": "value_cents",
}
def _parse_money(raw: str) -> int | None:
"""Parse dollar strings like '$3.3M', '$61.9M', '$1,234,567', '3300000' to cents."""
if not raw or not raw.strip():
return None
s = raw.strip().replace(",", "").replace("$", "")
multiplier = 1
if s.upper().endswith("M"):
multiplier = 1_000_000
s = s[:-1]
elif s.upper().endswith("K"):
multiplier = 1_000
s = s[:-1]
elif s.upper().endswith("B"):
multiplier = 1_000_000_000
s = s[:-1]
try:
return round(float(s) * multiplier * 100)
except ValueError:
return None
def _parse_date(raw: str) -> date | None:
"""Try common date formats."""
for fmt in ("%Y-%m-%d", "%m/%d/%Y", "%m/%d/%y", "%Y/%m/%d"):
try:
return datetime.strptime(raw.strip(), fmt).date()
except ValueError:
continue
return None
def _parse_shares(raw: str) -> str | None:
"""Parse share count, return as string (Decimal stored as string)."""
if not raw or not raw.strip():
return None
s = raw.strip().replace(",", "")
try:
float(s) # Validate it's numeric
return s
except ValueError:
return None
# --- Entity import ---
@router.post("/entities")
def import_entities(
file: UploadFile = File(...),
commit: bool = Query(default=True),
user: User = Depends(require_role(UserRole.approver, UserRole.cfo)),
session: Session = Depends(get_session),
) -> dict[str, Any]:
content = file.file.read().decode("utf-8-sig")
reader = csv.DictReader(io.StringIO(content))
results: list[dict] = []
errors: list[dict] = []
created = 0
updated = 0
for i, row in enumerate(reader, start=2): # row 1 is header
mapped: dict[str, Any] = {}
unmapped_cols: list[str] = []
for csv_col, model_field in ENTITY_COLUMN_MAP.items():
val = row.get(csv_col)
if val is None:
# Try case-insensitive match
for k, v in row.items():
if k.strip().lower() == csv_col.lower():
val = v
break
if val is not None:
mapped[model_field] = val.strip()
else:
unmapped_cols.append(csv_col)
# Parse fields
parsed: dict[str, Any] = {}
row_errors: list[str] = []
name = mapped.get("name")
if not name:
row_errors.append("Missing entity name")
else:
parsed["name"] = name
type_raw = mapped.get("type", "")
entity_type = ENTITY_TYPE_MAP.get(type_raw)
if entity_type is None and type_raw:
row_errors.append(f"Unknown entity type: {type_raw}")
elif entity_type:
parsed["type"] = entity_type
vy = mapped.get("vintage_year")
if vy:
try:
parsed["vintage_year"] = int(vy)
except ValueError:
row_errors.append(f"Invalid vintage year: {vy}")
fs = mapped.get("fund_size_cents")
if fs:
cents = _parse_money(fs)
if cents is None:
row_errors.append(f"Cannot parse fund size: {fs}")
else:
parsed["fund_size_cents"] = cents
if row_errors:
errors.append({"row": i, "errors": row_errors, "raw": dict(row)})
continue
if not parsed.get("name"):
continue
# Check existing
existing = session.exec(select(Entity).where(Entity.name == parsed["name"])).first()
action = "update" if existing else "create"
results.append({
"row": i,
"action": action,
"name": parsed["name"],
"type": parsed.get("type", EntityType.fund).value if parsed.get("type") else None,
"vintage_year": parsed.get("vintage_year"),
"fund_size_cents": parsed.get("fund_size_cents"),
})
if commit:
if existing:
if "type" in parsed:
existing.type = parsed["type"]
if "vintage_year" in parsed:
existing.vintage_year = parsed["vintage_year"]
if "fund_size_cents" in parsed:
existing.fund_size_cents = parsed["fund_size_cents"]
session.add(existing)
updated += 1
else:
entity = Entity(
name=parsed["name"],
type=parsed.get("type", EntityType.fund),
vintage_year=parsed.get("vintage_year"),
fund_size_cents=parsed.get("fund_size_cents"),
)
session.add(entity)
created += 1
if commit:
record_audit(session, user.id, "import_entities", "entity", None, {
"created": created, "updated": updated, "errors": len(errors),
})
session.commit()
return {
"committed": commit,
"preview": results,
"errors": errors,
"summary": {"created": created, "updated": updated, "error_rows": len(errors)},
}
# --- Schedule of investments import ---
@router.post("/schedule")
def import_schedule(
file: UploadFile = File(...),
entity_id: int = Query(...),
as_of: date = Query(...),
commit: bool = Query(default=True),
user: User = Depends(require_role(UserRole.approver, UserRole.cfo)),
session: Session = Depends(get_session),
) -> dict[str, Any]:
entity = session.get(Entity, entity_id)
if entity is None:
raise HTTPException(status_code=404, detail="Entity not found")
# Check for existing non-seed round at this quarter
existing_round = session.exec(
select(ValuationRound).where(
ValuationRound.entity_id == entity_id,
ValuationRound.quarter_end == as_of,
ValuationRound.is_seed == False,
)
).first()
if existing_round:
raise HTTPException(
status_code=409,
detail="A non-seed valuation round already exists for this entity and quarter.",
)
content = file.file.read().decode("utf-8-sig")
reader = csv.DictReader(io.StringIO(content))
holdings_preview: list[dict] = []
positions_preview: list[dict] = []
errors: list[dict] = []
current_company: str | None = None
for i, row in enumerate(reader, start=2):
mapped: dict[str, Any] = {}
for csv_col, model_field in SCHEDULE_COLUMN_MAP.items():
val = row.get(csv_col)
if val is None:
for k, v in row.items():
if k.strip().lower() == csv_col.lower():
val = v
break
if val is not None:
mapped[model_field] = val.strip()
company = mapped.get("company_name", "").strip()
security = mapped.get("security_name", "").strip()
# Grouped layout: company row has company name but may lack security detail
if company and not security:
current_company = company
holdings_preview.append({"row": i, "company_name": company})
continue
# Position row: may inherit company from grouping
if not company and current_company:
company = current_company
elif company:
# Both company and security on same row
if company not in [h["company_name"] for h in holdings_preview]:
holdings_preview.append({"row": i, "company_name": company})
current_company = company
if not security:
continue # Skip empty/subtotal rows
row_errors: list[str] = []
inv_date = None
if mapped.get("investment_date"):
inv_date = _parse_date(mapped["investment_date"])
if inv_date is None:
row_errors.append(f"Cannot parse date: {mapped['investment_date']}")
shares = _parse_shares(mapped.get("shares", ""))
cost_cents = None
if mapped.get("cost_cents"):
cost_cents = _parse_money(mapped["cost_cents"])
if cost_cents is None:
row_errors.append(f"Cannot parse cost: {mapped['cost_cents']}")
value_cents = None
if mapped.get("value_cents"):
value_cents = _parse_money(mapped["value_cents"])
if value_cents is None:
row_errors.append(f"Cannot parse value: {mapped['value_cents']}")
if row_errors:
errors.append({"row": i, "errors": row_errors, "raw": dict(row)})
continue
positions_preview.append({
"row": i,
"company_name": company,
"security_name": security,
"investment_date": str(inv_date) if inv_date else None,
"shares": shares,
"cost_cents": cost_cents,
"value_cents": value_cents,
})
if not commit:
return {
"committed": False,
"holdings": holdings_preview,
"positions": positions_preview,
"errors": errors,
"seed_round": {"entity_id": entity_id, "quarter_end": str(as_of)},
}
# Commit: create holdings, positions, seed round
holding_map: dict[str, Holding] = {}
for hp in holdings_preview:
name = hp["company_name"]
existing = session.exec(
select(Holding).where(
Holding.entity_id == entity_id,
Holding.company_name == name,
)
).first()
if existing:
holding_map[name] = existing
else:
h = Holding(entity_id=entity_id, company_name=name)
session.add(h)
session.flush()
holding_map[name] = h
# Create seed round
seed_round = ValuationRound(
entity_id=entity_id,
quarter_end=as_of,
status=RoundStatus.approved,
is_seed=True,
approved_by=user.id,
approved_at=datetime.utcnow(),
)
session.add(seed_round)
session.flush()
positions_created = 0
for pp in positions_preview:
company = pp["company_name"]
holding = holding_map.get(company)
if holding is None:
# Create holding on the fly for positions with inline company
existing = session.exec(
select(Holding).where(
Holding.entity_id == entity_id,
Holding.company_name == company,
)
).first()
if existing:
holding = existing
else:
holding = Holding(entity_id=entity_id, company_name=company)
session.add(holding)
session.flush()
holding_map[company] = holding
# Upsert position by (holding, security_name)
pos = session.exec(
select(Position).where(
Position.holding_id == holding.id,
Position.security_name == pp["security_name"],
)
).first()
if pos is None:
pos = Position(
holding_id=holding.id,
security_name=pp["security_name"],
investment_date=_parse_date(pp["investment_date"]) if pp["investment_date"] else as_of,
shares=pp["shares"],
cost_cents=pp["cost_cents"] or 0,
)
session.add(pos)
session.flush()
positions_created += 1
else:
if pp["investment_date"]:
pos.investment_date = _parse_date(pp["investment_date"])
if pp["shares"]:
pos.shares = pp["shares"]
if pp["cost_cents"] is not None:
pos.cost_cents = pp["cost_cents"]
session.add(pos)
session.flush()
# Attach valuation to seed round
val = Valuation(
round_id=seed_round.id,
position_id=pos.id,
value_cents=pp["value_cents"] or 0,
)
session.add(val)
record_audit(session, user.id, "import_schedule", "entity", entity_id, {
"holdings": len(holding_map),
"positions": positions_created,
"seed_quarter": str(as_of),
})
session.commit()
return {
"committed": True,
"holdings_count": len(holding_map),
"positions_count": len(positions_preview),
"seed_round_id": seed_round.id,
"errors": errors,
}