MyMidas/backend/app/api/v1/transactions.py
megaproxy 22fc1ce2f1 Add AI receipt parsing with provider API key settings
- Settings → AI: configure Anthropic or OpenAI provider with encrypted API key
- Sparkle button on each attachment in transaction drawer sends image/PDF to AI
- AI extracts merchant, amount, date, description, category hint
- "Apply to transaction" button patches the transaction with parsed fields
- Anthropic supports images and PDFs; OpenAI supports images only
- API key stored AES-256-GCM encrypted in users table (migration 0003)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-22 17:07:24 +00:00

468 lines
17 KiB
Python

import csv
import io
import mimetypes
import os
import uuid
from pathlib import Path
from typing import Annotated
from fastapi import APIRouter, Depends, File, Form, HTTPException, Query, UploadFile
from fastapi.responses import FileResponse
from sqlalchemy.ext.asyncio import AsyncSession
from app.config import get_settings
from app.core.audit import write_audit
from app.dependencies import get_current_user, get_db
from app.schemas.transaction import TransactionCreate, TransactionFilter, TransactionUpdate
from app.services.transaction_service import (
TransactionError,
create_transaction,
delete_transaction,
get_transaction,
import_csv,
list_transactions,
update_transaction,
_to_response,
)
MAX_IMPORT_FILE_BYTES = 10 * 1024 * 1024 # 10 MB
MAX_IMPORT_ROWS = 50_000
ALLOWED_MIME_TYPES = {
"image/jpeg",
"image/png",
"image/webp",
"application/pdf",
}
ALLOWED_EXTENSIONS = {".jpg", ".jpeg", ".png", ".webp", ".pdf"}
router = APIRouter()
@router.get("")
async def get_transactions(
account_id: uuid.UUID | None = None,
category_id: uuid.UUID | None = None,
type: str | None = None,
status: str | None = None,
date_from: str | None = None,
date_to: str | None = None,
search: str | None = None,
is_recurring: bool | None = None,
page: int = Query(default=1, ge=1),
page_size: int = Query(default=50, ge=1, le=200),
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
from datetime import date
filters = TransactionFilter(
account_id=account_id,
category_id=category_id,
type=type,
status=status,
date_from=date.fromisoformat(date_from) if date_from else None,
date_to=date.fromisoformat(date_to) if date_to else None,
search=search,
is_recurring=is_recurring,
page=page,
page_size=page_size,
)
return await list_transactions(db, user.id, filters)
@router.post("", status_code=201)
async def create(
body: TransactionCreate,
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
try:
result = await create_transaction(db, user.id, body, user.base_currency)
await write_audit(db, user_id=user.id, action="transaction_create")
await db.commit()
return result
except TransactionError as e:
raise HTTPException(status_code=e.status_code, detail=e.detail)
@router.get("/{txn_id}")
async def get_one(
txn_id: uuid.UUID,
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
try:
txn = await get_transaction(db, txn_id, user.id)
return _to_response(txn)
except TransactionError as e:
raise HTTPException(status_code=e.status_code, detail=e.detail)
@router.put("/{txn_id}")
async def update(
txn_id: uuid.UUID,
body: TransactionUpdate,
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
try:
result = await update_transaction(db, txn_id, user.id, body, user.base_currency)
await write_audit(db, user_id=user.id, action="transaction_update", resource_type="transaction", resource_id=txn_id)
await db.commit()
return result
except TransactionError as e:
raise HTTPException(status_code=e.status_code, detail=e.detail)
@router.delete("/{txn_id}", status_code=204)
async def delete(
txn_id: uuid.UUID,
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
try:
await delete_transaction(db, txn_id, user.id)
await write_audit(db, user_id=user.id, action="transaction_delete", resource_type="transaction", resource_id=txn_id)
await db.commit()
except TransactionError as e:
raise HTTPException(status_code=e.status_code, detail=e.detail)
@router.post("/{txn_id}/attachments")
async def upload_attachment(
txn_id: uuid.UUID,
file: UploadFile = File(...),
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
settings = get_settings()
# Validate extension
filename = file.filename or "upload"
ext = Path(filename).suffix.lower()
if ext not in ALLOWED_EXTENSIONS:
raise HTTPException(status_code=400, detail="Unsupported file type. Allowed: JPG, PNG, WebP, PDF")
# Verify transaction ownership
try:
txn = await get_transaction(db, txn_id, user.id)
except TransactionError as e:
raise HTTPException(status_code=e.status_code, detail=e.detail)
current_refs: list = txn.get("attachment_refs", []) if isinstance(txn, dict) else []
# Fetch raw model for JSONB mutation
from sqlalchemy import select
from app.db.models.transaction import Transaction as TxnModel
result = await db.execute(
select(TxnModel).where(TxnModel.id == txn_id, TxnModel.user_id == user.id)
)
txn_row = result.scalar_one_or_none()
if not txn_row:
raise HTTPException(status_code=404, detail="Transaction not found")
current_refs = list(txn_row.attachment_refs or [])
if len(current_refs) >= settings.max_attachments_per_txn:
raise HTTPException(status_code=400, detail=f"Maximum {settings.max_attachments_per_txn} attachments per transaction")
# Read and size-check
content = await file.read(settings.max_attachment_bytes + 1)
if len(content) > settings.max_attachment_bytes:
raise HTTPException(status_code=413, detail="File too large (max 10 MB)")
# Sniff MIME from content
import magic # python-magic
detected_mime = magic.from_buffer(content[:2048], mime=True)
if detected_mime not in ALLOWED_MIME_TYPES:
raise HTTPException(status_code=400, detail="File content does not match an allowed type (JPEG, PNG, WebP, PDF)")
# Store file
attachment_id = str(uuid.uuid4())
user_upload_dir = Path(settings.upload_dir) / str(user.id)
user_upload_dir.mkdir(parents=True, exist_ok=True)
stored_name = f"{attachment_id}{ext}"
stored_path = user_upload_dir / stored_name
stored_path.write_bytes(content)
# Update attachment_refs
ref = {
"id": attachment_id,
"filename": filename,
"mime_type": detected_mime,
"size": len(content),
"stored_name": stored_name,
}
from sqlalchemy import update as sql_update
import copy
new_refs = copy.copy(current_refs)
new_refs.append(ref)
await db.execute(
sql_update(TxnModel)
.where(TxnModel.id == txn_id)
.values(attachment_refs=new_refs)
)
await write_audit(db, user_id=user.id, action="transaction_update", resource_type="transaction", resource_id=txn_id, metadata={"attachment_added": attachment_id})
await db.commit()
return ref
@router.get("/{txn_id}/attachments/{attachment_id}")
async def download_attachment(
txn_id: uuid.UUID,
attachment_id: str,
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
settings = get_settings()
from sqlalchemy import select
from app.db.models.transaction import Transaction as TxnModel
result = await db.execute(
select(TxnModel).where(TxnModel.id == txn_id, TxnModel.user_id == user.id)
)
txn_row = result.scalar_one_or_none()
if not txn_row:
raise HTTPException(status_code=404, detail="Transaction not found")
ref = next((r for r in (txn_row.attachment_refs or []) if r["id"] == attachment_id), None)
if not ref:
raise HTTPException(status_code=404, detail="Attachment not found")
path = Path(settings.upload_dir) / str(user.id) / ref["stored_name"]
if not path.exists():
raise HTTPException(status_code=404, detail="Attachment file missing")
return FileResponse(
path=str(path),
media_type=ref["mime_type"],
filename=ref["filename"],
)
@router.delete("/{txn_id}/attachments/{attachment_id}", status_code=204)
async def delete_attachment(
txn_id: uuid.UUID,
attachment_id: str,
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
settings = get_settings()
from sqlalchemy import select, update as sql_update
from app.db.models.transaction import Transaction as TxnModel
result = await db.execute(
select(TxnModel).where(TxnModel.id == txn_id, TxnModel.user_id == user.id)
)
txn_row = result.scalar_one_or_none()
if not txn_row:
raise HTTPException(status_code=404, detail="Transaction not found")
refs = list(txn_row.attachment_refs or [])
ref = next((r for r in refs if r["id"] == attachment_id), None)
if not ref:
raise HTTPException(status_code=404, detail="Attachment not found")
# Delete file
path = Path(settings.upload_dir) / str(user.id) / ref["stored_name"]
try:
path.unlink(missing_ok=True)
except OSError:
pass
new_refs = [r for r in refs if r["id"] != attachment_id]
await db.execute(
sql_update(TxnModel)
.where(TxnModel.id == txn_id)
.values(attachment_refs=new_refs)
)
await write_audit(db, user_id=user.id, action="transaction_update", resource_type="transaction", resource_id=txn_id, metadata={"attachment_deleted": attachment_id})
await db.commit()
@router.post("/{txn_id}/attachments/{attachment_id}/parse")
async def parse_attachment(
txn_id: uuid.UUID,
attachment_id: str,
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
"""Send an attachment to the configured AI provider and extract receipt fields."""
import base64
import json
import httpx
from sqlalchemy import select
from app.db.models.transaction import Transaction as TxnModel
from app.db.models.user import User as UserModel
from app.core.security import decrypt_field
settings = get_settings()
# Reload user to get ai fields (get_current_user may be cached without them)
user_row = await db.get(UserModel, user.id)
if not user_row or not user_row.ai_provider or not user_row.ai_api_key_enc:
raise HTTPException(status_code=400, detail="No AI provider configured. Add your API key in Settings → AI.")
api_key = decrypt_field(user_row.ai_api_key_enc)
result = await db.execute(
select(TxnModel).where(TxnModel.id == txn_id, TxnModel.user_id == user.id)
)
txn_row = result.scalar_one_or_none()
if not txn_row:
raise HTTPException(status_code=404, detail="Transaction not found")
ref = next((r for r in (txn_row.attachment_refs or []) if r["id"] == attachment_id), None)
if not ref:
raise HTTPException(status_code=404, detail="Attachment not found")
mime_type = ref["mime_type"]
path = Path(settings.upload_dir) / str(user.id) / ref["stored_name"]
if not path.exists():
raise HTTPException(status_code=404, detail="Attachment file missing")
file_bytes = path.read_bytes()
b64 = base64.standard_b64encode(file_bytes).decode()
prompt = (
"You are a receipt parser. Extract information from this receipt and return ONLY a JSON object "
"with exactly these keys (use null for any field you cannot determine):\n"
'{"merchant": "store name", "amount": 0.00, "currency": "GBP", '
'"date": "YYYY-MM-DD", "description": "brief description", '
'"category": "one of: Food & Drink, Transport, Shopping, Entertainment, Health, Travel, Bills & Utilities, Other"}\n'
"Return ONLY the JSON object. No markdown, no explanation, no code fences."
)
try:
if user_row.ai_provider == "anthropic":
if mime_type == "application/pdf":
content_block = {
"type": "document",
"source": {"type": "base64", "media_type": "application/pdf", "data": b64},
}
else:
content_block = {
"type": "image",
"source": {"type": "base64", "media_type": mime_type, "data": b64},
}
async with httpx.AsyncClient(timeout=30) as client:
resp = await client.post(
"https://api.anthropic.com/v1/messages",
headers={
"x-api-key": api_key,
"anthropic-version": "2023-06-01",
"content-type": "application/json",
},
json={
"model": "claude-haiku-4-5-20251001",
"max_tokens": 512,
"messages": [{"role": "user", "content": [content_block, {"type": "text", "text": prompt}]}],
},
)
resp.raise_for_status()
text = resp.json()["content"][0]["text"].strip()
elif user_row.ai_provider == "openai":
if mime_type == "application/pdf":
raise HTTPException(status_code=400, detail="PDF parsing is not supported with the OpenAI provider. Use an image format or switch to Anthropic.")
async with httpx.AsyncClient(timeout=30) as client:
resp = await client.post(
"https://api.openai.com/v1/chat/completions",
headers={"Authorization": f"Bearer {api_key}", "content-type": "application/json"},
json={
"model": "gpt-4o-mini",
"max_tokens": 512,
"messages": [{
"role": "user",
"content": [
{"type": "image_url", "image_url": {"url": f"data:{mime_type};base64,{b64}"}},
{"type": "text", "text": prompt},
],
}],
},
)
resp.raise_for_status()
text = resp.json()["choices"][0]["message"]["content"].strip()
else:
raise HTTPException(status_code=400, detail="Unknown provider")
except httpx.HTTPStatusError as e:
raise HTTPException(status_code=502, detail=f"AI provider error: {e.response.status_code}")
except httpx.RequestError:
raise HTTPException(status_code=502, detail="Could not reach AI provider")
# Strip markdown fences if model wrapped the JSON
if text.startswith("```"):
text = text.split("```")[1]
if text.startswith("json"):
text = text[4:]
text = text.strip()
try:
parsed = json.loads(text)
except json.JSONDecodeError:
raise HTTPException(status_code=502, detail="AI returned an unexpected response. Try again.")
return {
"merchant": parsed.get("merchant"),
"amount": parsed.get("amount"),
"currency": parsed.get("currency"),
"date": parsed.get("date"),
"description": parsed.get("description"),
"category": parsed.get("category"),
}
@router.post("/import")
async def import_transactions(
file: UploadFile = File(...),
account_id: uuid.UUID = Form(...),
date_col: str = Form(default="date"),
description_col: str = Form(default="description"),
amount_col: str = Form(default="amount"),
db: AsyncSession = Depends(get_db),
user=Depends(get_current_user),
):
if not file.filename or not file.filename.lower().endswith(".csv"):
raise HTTPException(status_code=400, detail="Only CSV files are supported")
content = await file.read(MAX_IMPORT_FILE_BYTES + 1)
if len(content) > MAX_IMPORT_FILE_BYTES:
raise HTTPException(status_code=413, detail="File too large (max 10 MB)")
try:
text = content.decode("utf-8-sig") # handle BOM
except UnicodeDecodeError:
text = content.decode("latin-1")
reader = csv.DictReader(io.StringIO(text))
rows = []
for row in reader:
if len(rows) >= MAX_IMPORT_ROWS:
raise HTTPException(status_code=400, detail=f"File contains too many rows (max {MAX_IMPORT_ROWS:,})")
mapped = {}
# Flexible column mapping
for key, col in [("date", date_col), ("description", description_col), ("amount", amount_col)]:
val = row.get(col) or row.get(col.lower()) or row.get(col.upper())
if val is not None:
mapped[key] = val.strip()
if "date" in mapped and "amount" in mapped:
mapped.setdefault("description", "Imported transaction")
rows.append(mapped)
if not rows:
raise HTTPException(status_code=400, detail="No valid rows found. Check column names.")
result = await import_csv(db, user.id, account_id, rows, user.base_currency)
await write_audit(db, user_id=user.id, action="import_data", metadata=result)
await db.commit()
return result
@router.get("/import/template")
async def import_template():
from fastapi.responses import Response
csv_content = "date,description,amount,merchant,notes\n2026-01-15,Tesco Groceries,-45.67,Tesco,\n2026-01-14,Salary,2500.00,Employer,January salary\n"
return Response(
content=csv_content,
media_type="text/csv",
headers={"Content-Disposition": "attachment; filename=import_template.csv"},
)