| """ |
| RBAC-aware token resolver. |
| |
| For each [LABEL_hexid] token in a text chunk: |
| - If user has a matching RbacGrant → decrypt from Redis and substitute |
| - Otherwise → leave the token unchanged |
| |
| Decryption uses the same Presidio AnonymizerEngine decrypt path used by the |
| existing decrypt_file.py to stay consistent. |
| """ |
| import base64 |
| import re |
| import uuid |
| from typing import Set |
|
|
| from presidio_anonymizer import AnonymizerEngine |
| from presidio_anonymizer.entities import OperatorConfig, RecognizerResult |
|
|
| _TOKEN_RE = re.compile(r"\[([A-Z_]{1,30})_([0-9a-f]{8})\]") |
|
|
| _anonymizer = AnonymizerEngine() |
|
|
|
|
| def _presidio_decrypt(encrypted_value: str, aes_key: bytes) -> str: |
| key_b64 = base64.b64encode(aes_key).decode() |
| fake = [RecognizerResult(entity_type="PII", start=0, end=len(encrypted_value), score=1.0)] |
| result = _anonymizer.anonymize( |
| text=encrypted_value, |
| analyzer_results=fake, |
| operators={"DEFAULT": OperatorConfig("decrypt", {"key": key_b64})}, |
| ) |
| return result.text |
|
|
|
|
| def resolve_tokens( |
| text: str, |
| token_map: dict[str, str], |
| aes_key: bytes, |
| allowed_entity_types: Set[str], |
| dataset_token_map: dict[str, str] | None = None, |
| ) -> str: |
| """ |
| Replace tokens in *text* with their plaintext if the user has RBAC access. |
| Checks session token_map first, then dataset_token_map. |
| Tokens without RBAC access are left as-is. |
| """ |
| if not text: |
| return text |
|
|
| allow_all = "*" in allowed_entity_types |
|
|
| def _replace(match: re.Match) -> str: |
| label, _ = match.group(1), match.group(2) |
| token = match.group(0) |
|
|
| if not (allow_all or label in allowed_entity_types): |
| return token |
|
|
| encrypted = token_map.get(token) |
| if not encrypted and dataset_token_map: |
| encrypted = dataset_token_map.get(token) |
| if not encrypted: |
| return token |
|
|
| try: |
| return _presidio_decrypt(encrypted, aes_key) |
| except Exception: |
| return token |
|
|
| return _TOKEN_RE.sub(_replace, text) |
|
|
|
|
| async def get_allowed_entity_types( |
| user_id: uuid.UUID, |
| dataset_id: uuid.UUID | None, |
| role: str, |
| db, |
| ) -> Set[str]: |
| """ |
| Return the set of entity types this user may see in plaintext. |
| |
| Admin role always gets '*'. For other roles, query RbacGrant. |
| """ |
| if role == "admin": |
| return {"*"} |
|
|
| if not dataset_id: |
| return set() |
|
|
| from sqlalchemy import select |
| from api.models import RbacGrant |
|
|
| result = await db.execute( |
| select(RbacGrant.entity_type).where( |
| RbacGrant.user_id == user_id, |
| RbacGrant.dataset_id == dataset_id, |
| ) |
| ) |
| return {row[0] for row in result} |
|
|