Files
eveAI/common/services/entitlements/license_usage_services.py

144 lines
6.2 KiB
Python

from dateutil.relativedelta import relativedelta
from flask import session, current_app, flash
from sqlalchemy.exc import SQLAlchemyError
from sqlalchemy.sql.expression import text
from common.extensions import db, cache_manager
from common.models.entitlements import PartnerServiceLicenseTier, License, LicenseUsage, LicensePeriod, PeriodStatus
from common.models.user import Partner, PartnerTenant
from common.services.entitlements import LicensePeriodServices
from common.utils.database import Database
from common.utils.eveai_exceptions import EveAINoManagementPartnerService, EveAINoActiveLicense, \
EveAIStorageQuotaExceeded, EveAIEmbeddingQuotaExceeded, EveAIInteractionQuotaExceeded, EveAILicensePeriodsExceeded, \
EveAIException
from common.utils.model_logging_utils import set_logging_information, update_logging_information
from datetime import datetime as dt, timezone as tz
from common.utils.security_utils import current_user_has_role
class LicenseUsageServices:
@staticmethod
def check_storage_and_embedding_quota(tenant_id: int, file_size_mb: float) -> None:
"""
Check if a tenant can add a new document without exceeding storage and embedding quotas
Args:
tenant_id: ID of the tenant
file_size_mb: Size of the file in MB
Raises:
EveAIStorageQuotaExceeded: If storage quota would be exceeded
EveAIEmbeddingQuotaExceeded: If embedding quota would be exceeded
EveAINoActiveLicense: If no active license is found
EveAIException: For other errors
"""
# Get active license period
license_period = LicensePeriodServices.find_current_license_period_for_usage(tenant_id)
# Early return if both overruns are allowed - no need to check usage at all
if license_period.additional_storage_allowed and license_period.additional_embedding_allowed:
return
# Check storage quota only if overruns are not allowed
if not license_period.additional_storage_allowed:
LicenseUsageServices._validate_storage_quota(license_period, file_size_mb)
# Check embedding quota only if overruns are not allowed
if not license_period.additional_embedding_allowed:
LicenseUsageServices._validate_embedding_quota(license_period, file_size_mb)
@staticmethod
def check_embedding_quota(tenant_id: int, file_size_mb: float) -> None:
"""
Check if a tenant can re-embed a document without exceeding embedding quota
Args:
tenant_id: ID of the tenant
file_size_mb: Size of the file in MB
Raises:
EveAIEmbeddingQuotaExceeded: If embedding quota would be exceeded
EveAINoActiveLicense: If no active license is found
EveAIException: For other errors
"""
# Get active license period
license_period = LicensePeriodServices.find_current_license_period_for_usage(tenant_id)
# Early return if both overruns are allowed - no need to check usage at all
if license_period.additional_embedding_allowed:
return
# Check embedding quota
LicenseUsageServices._validate_embedding_quota(license_period, file_size_mb)
@staticmethod
def check_interaction_quota(tenant_id: int) -> None:
"""
Check if a tenant can execute a specialist without exceeding interaction quota. As it is impossible to estimate
the number of interaction tokens, we only check if the interaction quota are exceeded. So we might have a
limited overrun.
Args:
tenant_id: ID of the tenant
Raises:
EveAIInteractionQuotaExceeded: If interaction quota would be exceeded
EveAINoActiveLicense: If no active license is found
EveAIException: For other errors
"""
# Get active license period
license_period = LicensePeriodServices.find_current_license_period_for_usage(tenant_id)
# Early return if both overruns are allowed - no need to check usage at all
if license_period.additional_interaction_allowed:
return
# Convert tokens to M tokens and check interaction quota
LicenseUsageServices._validate_interaction_quota(license_period)
@staticmethod
def _validate_storage_quota(license_period: LicensePeriod, additional_mb: float) -> None:
"""Check storage quota and raise exception if exceeded"""
current_storage = license_period.license_usage.storage_mb_used or 0
projected_storage = current_storage + additional_mb
max_storage = license_period.max_storage_mb
# Hard limit check (we only get here if overruns are NOT allowed)
if projected_storage > max_storage:
raise EveAIStorageQuotaExceeded(
current_usage=current_storage,
limit=max_storage,
additional=additional_mb
)
@staticmethod
def _validate_embedding_quota(license_period: LicensePeriod, additional_mb: float) -> None:
"""Check embedding quota and raise exception if exceeded"""
current_embedding = license_period.license_usage.embedding_mb_used or 0
projected_embedding = current_embedding + additional_mb
max_embedding = license_period.included_embedding_mb
# Hard limit check (we only get here if overruns are NOT allowed)
if projected_embedding > max_embedding:
raise EveAIEmbeddingQuotaExceeded(
current_usage=current_embedding,
limit=max_embedding,
additional=additional_mb
)
@staticmethod
def _validate_interaction_quota(license_period) -> None:
"""Check interaction quota and raise exception if exceeded (tokens in millions). We might have an overrun!"""
current_tokens = license_period.license_usage.interaction_total_tokens_used / 1_000_000 or 0
max_tokens = license_period.included_interaction_tokens
# Hard limit check (we only get here if overruns are NOT allowed)
if current_tokens > max_tokens:
raise EveAIInteractionQuotaExceeded(
current_usage=current_tokens,
limit=max_tokens
)