Spaces:
Running
Running
File size: 9,737 Bytes
88bedd1 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 | # src/ticket_validator.py
# Ticket Input Validator โ Edge Case Handler
# SupportMind v1.0 โ Asmitha
import re
import logging
from typing import Dict, Tuple
logger = logging.getLogger(__name__)
# โโ Constants โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
MIN_WORDS = 4
MAX_WORDS = 500
MIN_CHARS = 10
MAX_CHARS = 3000
# Non-English character detection
# Covers: Arabic, Hindi/Devanagari, Tamil, Chinese,
# Japanese, Korean, Thai, Russian
NON_LATIN_PATTERN = re.compile(
r'[\u0600-\u06FF' # Arabic
r'\u0900-\u097F' # Devanagari (Hindi)
r'\u0B80-\u0BFF' # Tamil
r'\u4E00-\u9FFF' # Chinese
r'\u3040-\u30FF' # Japanese
r'\uAC00-\uD7AF' # Korean
r'\u0E00-\u0E7F' # Thai
r'\u0400-\u04FF]' # Russian/Cyrillic
)
# Gibberish detection
# No vowels in long sequences = likely gibberish
GIBBERISH_PATTERN = re.compile(r'\b[^aeiou\s]{6,}\b', re.IGNORECASE)
# Already resolved patterns
RESOLVED_PATTERNS = [
r'never ?mind',
r'problem (?:is )?(?:solved|fixed|resolved)',
r'(?:sorted|fixed) (?:it )?(?:out)?',
r'no longer (?:need|require)',
r'cancel (?:this )?(?:ticket|request)',
r'disregard',
r'ignore (?:this|my)',
r'thanks?,? (?:got it|all good|figured)',
]
# Greeting/test patterns
GREETING_PATTERNS = [
r'^hi+\s*[.!?]*$',
r'^hello+\s*[.!?]*$',
r'^hey+\s*[.!?]*$',
r'^test\s*[.!?]*$',
r'^testing\s*[.!?]*$',
r'^help\s*[.!?]*$',
r'^\?\s*$',
r'^\.+$',
]
# Abuse/spam patterns (basic)
SPAM_PATTERNS = [
r'(.)\1{9,}', # Same char repeated 10+ times
r'(\b\w+\b)(\s+\1){4,}', # Same word repeated 5+ times
]
class TicketValidator:
"""
Validates and pre-processes ticket text before ML inference.
Catches edge cases early so the ML pipeline never receives
bad input. Returns structured validation result with
specific response for each edge case.
"""
def validate(self, text: str) -> Dict:
"""
Validate ticket text and return result.
Returns:
{
'valid': bool,
'cleaned_text': str, # cleaned version if valid
'error_type': str | None, # type of error if invalid
'response': str, # what to show user
'should_route': bool, # proceed to ML?
}
"""
# โโ Check 1: Empty or None โโโโโโโโโโโโโโโโโโโโโโ
if not text or not text.strip():
return self._invalid(
error_type='empty',
response="It looks like your message is empty. "
"Please describe your issue and we'll help you right away."
)
# Clean whitespace
cleaned = ' '.join(text.strip().split())
# โโ Check 2: Too short โโโโโโโโโโโโโโโโโโโโโโโโโโ
words = cleaned.split()
if len(words) < MIN_WORDS or len(cleaned) < MIN_CHARS:
# Check if it's a greeting specifically
if any(re.match(p, cleaned.lower()) for p in GREETING_PATTERNS):
return self._invalid(
error_type='greeting',
response="Hi there! ๐ Could you describe the issue "
"you're experiencing? We're here to help."
)
return self._invalid(
error_type='too_short',
response="Could you share a bit more detail about your issue? "
"The more context you provide, the faster we can help."
)
# โโ Check 3: Too long โโโโโโโโโโโโโโโโโโโโโโโโโโโ
if len(words) > MAX_WORDS or len(cleaned) > MAX_CHARS:
# Truncate intelligently โ keep first 500 words
truncated_words = words[:MAX_WORDS]
cleaned = ' '.join(truncated_words)
logger.info(f"Ticket truncated from {len(words)} to {MAX_WORDS} words")
# Still valid โ just truncated
return self._valid(
cleaned_text=cleaned,
warning="Your message was very long โ "
"we've focused on the first part to route you correctly."
)
# โโ Check 4: Non-English โโโโโโโโโโโโโโโโโโโโโโโโ
non_latin_chars = len(NON_LATIN_PATTERN.findall(cleaned))
total_chars = len(re.sub(r'\s', '', cleaned))
non_latin_ratio = non_latin_chars / max(total_chars, 1)
if non_latin_ratio > 0.3:
language = self._detect_language_hint(cleaned)
return self._invalid(
error_type='non_english',
response=f"We noticed your message may be in another language. "
f"Our routing system currently works best in English. "
f"Could you resend your message in English? "
f"We want to make sure you reach the right team quickly."
)
# โโ Check 5: Already resolved โโโโโโโโโโโโโโโโโโโ
if any(re.search(p, cleaned.lower()) for p in RESOLVED_PATTERNS):
return self._invalid(
error_type='resolved',
response="Glad to hear it's sorted! ๐ "
"If you need anything else, don't hesitate to reach out.",
should_route=False
)
# โโ Check 6: Gibberish โโโโโโโโโโโโโโโโโโโโโโโโโโ
gibberish_matches = GIBBERISH_PATTERN.findall(cleaned)
total_words = len(words)
gibberish_ratio = len(gibberish_matches) / max(total_words, 1)
if gibberish_ratio > 0.4:
return self._invalid(
error_type='gibberish',
response="We couldn't quite understand your message. "
"Could you describe your issue in plain language?"
)
# โโ Check 7: Spam โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
for pattern in SPAM_PATTERNS:
if re.search(pattern, cleaned):
return self._invalid(
error_type='spam',
response="We weren't able to process your message. "
"Please describe your issue clearly."
)
# โโ Check 8: Only numbers/symbols โโโโโโโโโโโโโโโ
alpha_chars = len(re.findall(r'[a-zA-Z]', cleaned))
if alpha_chars < 5:
return self._invalid(
error_type='no_text',
response="Could you describe your issue in words? "
"We want to make sure you reach the right team."
)
# โโ All checks passed โโโโโโโโโโโโโโโโโโโโโโโโโโโโ
return self._valid(cleaned_text=cleaned)
def _valid(self, cleaned_text: str, warning: str = None) -> Dict:
return {
'valid': True,
'cleaned_text': cleaned_text,
'error_type': None,
'response': warning,
'should_route': True,
'warning': warning is not None,
}
def _invalid(self,
error_type: str,
response: str,
should_route: bool = False) -> Dict:
return {
'valid': False,
'cleaned_text': None,
'error_type': error_type,
'response': response,
'should_route': should_route,
'warning': False,
}
def _detect_language_hint(self, text: str) -> str:
"""Basic language hint for logging."""
if re.search(r'[\u0B80-\u0BFF]', text):
return 'Tamil'
if re.search(r'[\u0900-\u097F]', text):
return 'Hindi'
if re.search(r'[\u0600-\u06FF]', text):
return 'Arabic'
if re.search(r'[\u4E00-\u9FFF]', text):
return 'Chinese'
if re.search(r'[\uAC00-\uD7AF]', text):
return 'Korean'
return 'Unknown'
# โโ Quick test โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
if __name__ == '__main__':
validator = TicketValidator()
test_cases = [
("hi", "greeting"),
("", "empty"),
(" ", "empty"),
("asdfghjkl qwerty zxcvbnm poiuytrewq", "gibberish"),
("เฎเฎฉเฎคเฏ เฎเฎฃเฎเฏเฎเฎฟเฎฒเฏ เฎเฎฟเฎเฏเฎเฎฒเฏ เฎเฎณเฏเฎณเฎคเฏ", "tamil"),
("My invoice is wrong please help me fix this billing issue", "valid"),
("never mind got it sorted thanks", "resolved"),
("aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", "spam"),
("500 404 200 301 302", "no_text"),
("The API endpoint returns 500 error " * 200, "too_long"),
]
print("=" * 60)
print("TICKET VALIDATOR โ EDGE CASE TESTS")
print("=" * 60)
for text, expected in test_cases:
result = validator.validate(text)
status = "[OK]" if not result['valid'] or result['valid'] else "[ERROR]"
preview = text[:40] + "..." if len(text) > 40 else text
print(f"\nInput: '{preview}'")
print(f"Expected: {expected}")
print(f"Got: {result['error_type'] or 'valid'}")
print(f"Response: {result['response']}")
|