Spaces:
Runtime error
Runtime error
Update validators.py
Browse files- validators.py +91 -0
validators.py
CHANGED
|
@@ -41,3 +41,94 @@ def with_fallback_author_role(df):
|
|
| 41 |
fixed.loc[(fixed["author_role"] == "") & teacher_name_hint, "author_role"] = "교사"
|
| 42 |
fixed.loc[fixed["author_role"] == "", "author_role"] = "학생"
|
| 43 |
return fixed
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
fixed.loc[(fixed["author_role"] == "") & teacher_name_hint, "author_role"] = "교사"
|
| 42 |
fixed.loc[fixed["author_role"] == "", "author_role"] = "학생"
|
| 43 |
return fixed
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
ALLOWED_TEXT_PATTERN = re.compile(r"^[0-9A-Za-z가-힣ㄱ-ㅎㅏ-ㅣ _.,!?():;@#&/\-]*$")
|
| 48 |
+
|
| 49 |
+
|
| 50 |
+
def _contains_forbidden_word(text, forbidden_words=None):
|
| 51 |
+
words = forbidden_words or []
|
| 52 |
+
lowered = str(text or "").lower()
|
| 53 |
+
for word in words:
|
| 54 |
+
if str(word).lower() in lowered:
|
| 55 |
+
return True
|
| 56 |
+
return False
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
def _validate_text_field(
|
| 60 |
+
raw_text,
|
| 61 |
+
*,
|
| 62 |
+
field_name,
|
| 63 |
+
max_len,
|
| 64 |
+
allow_empty=False,
|
| 65 |
+
forbidden_words=None,
|
| 66 |
+
allowed_pattern=ALLOWED_TEXT_PATTERN,
|
| 67 |
+
):
|
| 68 |
+
normalized = normalize_user_text(raw_text, max_len=max_len)
|
| 69 |
+
if not normalized:
|
| 70 |
+
if allow_empty:
|
| 71 |
+
return True, "", None, None
|
| 72 |
+
return False, "", "VALIDATION_EMPTY", f"{field_name}을(를) 입력해 주세요."
|
| 73 |
+
|
| 74 |
+
if len(normalized) > max_len:
|
| 75 |
+
return False, "", "VALIDATION_TOO_LONG", f"{field_name}은(는) {max_len}자 이하여야 합니다."
|
| 76 |
+
|
| 77 |
+
if allowed_pattern and not allowed_pattern.match(normalized):
|
| 78 |
+
return False, "", "VALIDATION_INVALID_CHARS", f"{field_name}에 허용되지 않은 문자가 포함되어 있습니다."
|
| 79 |
+
|
| 80 |
+
if _contains_forbidden_word(normalized, forbidden_words=forbidden_words):
|
| 81 |
+
return False, "", "VALIDATION_FORBIDDEN", f"{field_name}에 사용할 수 없는 단어가 포함되어 있습니다."
|
| 82 |
+
|
| 83 |
+
return True, normalized, None, None
|
| 84 |
+
|
| 85 |
+
|
| 86 |
+
def validate_room_name(raw_text, max_len=60):
|
| 87 |
+
collapsed = normalize_room_name(raw_text, max_len=max_len)
|
| 88 |
+
return _validate_text_field(
|
| 89 |
+
collapsed,
|
| 90 |
+
field_name="방 이름",
|
| 91 |
+
max_len=max_len,
|
| 92 |
+
allow_empty=False,
|
| 93 |
+
forbidden_words=["관리자", "운영자"],
|
| 94 |
+
)
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
def validate_student_name(raw_text, max_len=30):
|
| 98 |
+
return _validate_text_field(
|
| 99 |
+
raw_text,
|
| 100 |
+
field_name="학생 이름/학번",
|
| 101 |
+
max_len=max_len,
|
| 102 |
+
allow_empty=False,
|
| 103 |
+
forbidden_words=["관리자", "운영자", "admin"],
|
| 104 |
+
)
|
| 105 |
+
|
| 106 |
+
|
| 107 |
+
def validate_opinion_content(raw_text, max_len=700):
|
| 108 |
+
return _validate_text_field(
|
| 109 |
+
raw_text,
|
| 110 |
+
field_name="의견 내용",
|
| 111 |
+
max_len=max_len,
|
| 112 |
+
allow_empty=False,
|
| 113 |
+
)
|
| 114 |
+
|
| 115 |
+
|
| 116 |
+
def validate_entry_code(raw_text, max_len=60):
|
| 117 |
+
return _validate_text_field(
|
| 118 |
+
raw_text,
|
| 119 |
+
field_name="입장 암호",
|
| 120 |
+
max_len=max_len,
|
| 121 |
+
allow_empty=True,
|
| 122 |
+
allowed_pattern=re.compile(r"^[0-9A-Za-z가-힣ㄱ-ㅎㅏ-ㅣ _!@#$%^&*()+=.?/:;\-]*$"),
|
| 123 |
+
)
|
| 124 |
+
|
| 125 |
+
|
| 126 |
+
def validate_teacher_credential(raw_text, *, field_name, max_len=60):
|
| 127 |
+
return _validate_text_field(
|
| 128 |
+
raw_text,
|
| 129 |
+
field_name=field_name,
|
| 130 |
+
max_len=max_len,
|
| 131 |
+
allow_empty=False,
|
| 132 |
+
forbidden_words=[" "],
|
| 133 |
+
allowed_pattern=re.compile(r"^[0-9A-Za-z_.@\-]+$"),
|
| 134 |
+
)
|