Ui
n U d Z ddlmZ ddlZ ej d ej d ej d ej d ej d ej d ej d
ej d gZd ed
< ddZddZddZ ddZ
dddZ ddZ y)u? engine_v2/content_sanitizer.py — 3레이어 인젝션 방어. ) annotationsNz-(?i)ignore\s+(all\s+)?previous\s+instructionsz!(?i)you\s+are\s+now\s+(?:a|an)\s+z(?i)system\s*:\s*z(?i)<\s*/?\s*system\s*>z (?i)act\s+as\s+(?:a|an|if)\s+z"(?i)forget\s+(everything|all|your)z(?i)new\s+instructions?\s*:z (?i)override\s+(your|the)\s+zlist[re.Pattern[str]]_INJECTION_PATTERNSc d}| }t D ]3 }|j | }|t | z
}|j d| }5 ||fS )ua L1: 패턴 기반 정규식 필터링.
Returns:
(sanitized_text, flagged_count)
r z
[REDACTED])r findalllensub)textflaggedresultpatternmatchess J/home/jay/workspace/services/multimodel-bot/engine_v2/content_sanitizer.py sanitize_l1r sU G
F& 3//&)3w< \623 7? c $ t | }d| dS )u= L2: 구조적 격리 — 태그로 감싸기.z
z
)_escape_envelope_tags)r escapeds r
wrap_upstreamr # s #D)G
wi'9 ::r c & | j dd S )u" L2 보조: 태그 이스케이프.zz</UPSTREAM_DATA>)replace)r s r r r ) s <<*,D EEr c < t | \ }}t | }||fS )uw 전체 Sanitize 파이프라인 (L1 + L2).
Returns:
(sanitized_and_wrapped_text, total_flagged_count)
)r r )r cleanedr
wrappeds r sanitizer . s( #4(GWG$G G r c | |k\ S )uZ L3: 오케스트레이터 게이트. flagged_count >= threshold이면 True(중단 필요). )
flagged_count thresholds r
check_gater 9 s
I %%r c | xr | S )uV L4: 에러 자동 제외. error=True AND fallback 미사용이면 True(중단 필요).r )error
fallback_useds r check_error_gater# > s & &&r )r strreturnztuple[str, int])r r$ r% r$ ) )r intr r' r% bool)r! r( r" r( r% r( )
__doc__
__future__r recompiler __annotations__r r r r r r# r r r r. s E " BJJ?@BJJ34BJJ#$BJJ)*BJJ/0BJJ45BJJ-.BJJ./ . * ; F
&
'r