Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -108,50 +108,103 @@ Just type your question or start a conversation, and I'll do my best to help you
|
|
| 108 |
return random.choice(responses)
|
| 109 |
|
| 110 |
def save_unanswered_question(self, question: str) -> bool:
|
| 111 |
-
"""Save unanswered question to the database"""
|
|
|
|
|
|
|
| 112 |
try:
|
| 113 |
# Try different possible endpoints for saving unanswered questions
|
| 114 |
endpoints = [
|
| 115 |
f"{self.database_url}/unanswered_questions",
|
| 116 |
f"{self.database_url}/api/unanswered_questions",
|
| 117 |
f"{self.database_url}/save_question",
|
| 118 |
-
f"{self.database_url}/api/save_question"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 119 |
]
|
| 120 |
|
|
|
|
|
|
|
|
|
|
| 121 |
for endpoint in endpoints:
|
|
|
|
|
|
|
|
|
|
| 122 |
try:
|
| 123 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
response = requests.post(
|
| 125 |
endpoint,
|
| 126 |
-
json=
|
| 127 |
-
"question": question,
|
| 128 |
-
"created_at": self._get_timestamp()
|
| 129 |
-
},
|
| 130 |
headers={"Content-Type": "application/json"},
|
| 131 |
timeout=10
|
| 132 |
)
|
|
|
|
|
|
|
|
|
|
| 133 |
if response.status_code in [200, 201]:
|
|
|
|
| 134 |
return True
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
|
| 148 |
-
|
| 149 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 150 |
|
|
|
|
| 151 |
return False
|
| 152 |
|
| 153 |
except Exception as e:
|
| 154 |
-
print(f"
|
| 155 |
return False
|
| 156 |
|
| 157 |
def _get_timestamp(self) -> str:
|
|
@@ -167,6 +220,30 @@ Just type your question or start a conversation, and I'll do my best to help you
|
|
| 167 |
text = text.translate(str.maketrans('', '', string.punctuation))
|
| 168 |
# Remove extra whitespace
|
| 169 |
text = ' '.join(text.split())
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 170 |
return text
|
| 171 |
|
| 172 |
def _extract_keywords(self, text: str) -> List[str]:
|
|
@@ -227,18 +304,35 @@ Just type your question or start a conversation, and I'll do my best to help you
|
|
| 227 |
if norm2 in norm1:
|
| 228 |
contains_similarity = max(contains_similarity, 0.9 * (len(norm2) / len(norm1)))
|
| 229 |
|
| 230 |
-
# Method 4:
|
| 231 |
words1 = norm1.split()
|
| 232 |
words2 = norm2.split()
|
| 233 |
word_order_similarity = 0.0
|
| 234 |
if words1 and words2:
|
| 235 |
-
# Check for common word sequences
|
| 236 |
common_sequences = 0
|
| 237 |
max_len = min(len(words1), len(words2))
|
| 238 |
for i in range(max_len):
|
| 239 |
if words1[i] == words2[i]:
|
| 240 |
common_sequences += 1
|
| 241 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 242 |
|
| 243 |
# Method 5: Semantic similarity using word relationships
|
| 244 |
semantic_similarity = self._calculate_semantic_similarity(keywords1, keywords2)
|
|
@@ -277,8 +371,9 @@ Just type your question or start a conversation, and I'll do my best to help you
|
|
| 277 |
'contact': {'contact', 'phone', 'email', 'address', 'office', 'reach', 'call'},
|
| 278 |
'requirements': {'requirement', 'need', 'required', 'must', 'prerequisite', 'condition'},
|
| 279 |
'application': {'apply', 'application', 'submit', 'process', 'procedure'},
|
| 280 |
-
'programs': {'program', 'course', 'major', 'degree', 'study', 'academic'},
|
| 281 |
-
'admission': {'admission', 'admit', 'accept', 'enroll', 'entry', 'enter'}
|
|
|
|
| 282 |
}
|
| 283 |
|
| 284 |
# Check if keywords belong to the same semantic group
|
|
@@ -313,6 +408,50 @@ Just type your question or start a conversation, and I'll do my best to help you
|
|
| 313 |
|
| 314 |
return min(phrase_score, 1.0)
|
| 315 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 316 |
def _find_best_match(self, user_question: str, database_questions: List[str], threshold: float = 0.25) -> Optional[str]:
|
| 317 |
"""Find the best matching question from database with improved logic"""
|
| 318 |
if not database_questions:
|
|
@@ -422,21 +561,33 @@ Just type your question or start a conversation, and I'll do my best to help you
|
|
| 422 |
continue
|
| 423 |
|
| 424 |
# If no answer found, save the question as unanswered
|
| 425 |
-
self.save_unanswered_question(question)
|
| 426 |
-
|
|
|
|
|
|
|
|
|
|
| 427 |
|
| 428 |
except requests.exceptions.Timeout:
|
| 429 |
# Save the question even if there's a timeout
|
| 430 |
-
self.save_unanswered_question(question)
|
| 431 |
-
|
|
|
|
|
|
|
|
|
|
| 432 |
except requests.exceptions.ConnectionError:
|
| 433 |
# Save the question even if there's a connection error
|
| 434 |
-
self.save_unanswered_question(question)
|
| 435 |
-
|
|
|
|
|
|
|
|
|
|
| 436 |
except Exception as e:
|
| 437 |
# Save the question even if there's an unexpected error
|
| 438 |
-
self.save_unanswered_question(question)
|
| 439 |
-
|
|
|
|
|
|
|
|
|
|
| 440 |
|
| 441 |
def _get_all_questions(self) -> List[str]:
|
| 442 |
"""Get all available questions from the database for smart matching"""
|
|
|
|
| 108 |
return random.choice(responses)
|
| 109 |
|
| 110 |
def save_unanswered_question(self, question: str) -> bool:
|
| 111 |
+
"""Save unanswered question to the database with detailed logging"""
|
| 112 |
+
print(f"Attempting to save unanswered question: '{question}'")
|
| 113 |
+
|
| 114 |
try:
|
| 115 |
# Try different possible endpoints for saving unanswered questions
|
| 116 |
endpoints = [
|
| 117 |
f"{self.database_url}/unanswered_questions",
|
| 118 |
f"{self.database_url}/api/unanswered_questions",
|
| 119 |
f"{self.database_url}/save_question",
|
| 120 |
+
f"{self.database_url}/api/save_question",
|
| 121 |
+
f"{self.database_url}/questions",
|
| 122 |
+
f"{self.database_url}/api/questions",
|
| 123 |
+
f"{self.database_url}/faq/unanswered",
|
| 124 |
+
f"{self.database_url}/api/faq/unanswered"
|
| 125 |
]
|
| 126 |
|
| 127 |
+
timestamp = self._get_timestamp()
|
| 128 |
+
print(f"Using timestamp: {timestamp}")
|
| 129 |
+
|
| 130 |
for endpoint in endpoints:
|
| 131 |
+
print(f"Trying endpoint: {endpoint}")
|
| 132 |
+
|
| 133 |
+
# Try POST request with JSON body - matching your table structure
|
| 134 |
try:
|
| 135 |
+
post_data = {
|
| 136 |
+
"question": question,
|
| 137 |
+
"created_at": timestamp
|
| 138 |
+
}
|
| 139 |
+
print(f"POST data: {post_data}")
|
| 140 |
+
|
| 141 |
response = requests.post(
|
| 142 |
endpoint,
|
| 143 |
+
json=post_data,
|
|
|
|
|
|
|
|
|
|
| 144 |
headers={"Content-Type": "application/json"},
|
| 145 |
timeout=10
|
| 146 |
)
|
| 147 |
+
print(f"POST response status: {response.status_code}")
|
| 148 |
+
print(f"POST response text: {response.text[:200]}")
|
| 149 |
+
|
| 150 |
if response.status_code in [200, 201]:
|
| 151 |
+
print(f"Successfully saved question via POST to {endpoint}")
|
| 152 |
return True
|
| 153 |
+
|
| 154 |
+
except requests.exceptions.RequestException as e:
|
| 155 |
+
print(f"POST request failed for {endpoint}: {e}")
|
| 156 |
+
|
| 157 |
+
# Try GET request with query parameters
|
| 158 |
+
try:
|
| 159 |
+
get_params = {
|
| 160 |
+
"question": question,
|
| 161 |
+
"created_at": timestamp
|
| 162 |
+
}
|
| 163 |
+
print(f"GET params: {get_params}")
|
| 164 |
+
|
| 165 |
+
response = requests.get(
|
| 166 |
+
endpoint,
|
| 167 |
+
params=get_params,
|
| 168 |
+
timeout=10
|
| 169 |
+
)
|
| 170 |
+
print(f"GET response status: {response.status_code}")
|
| 171 |
+
print(f"GET response text: {response.text[:200]}")
|
| 172 |
+
|
| 173 |
+
if response.status_code in [200, 201]:
|
| 174 |
+
print(f"Successfully saved question via GET to {endpoint}")
|
| 175 |
+
return True
|
| 176 |
+
|
| 177 |
+
except requests.exceptions.RequestException as e:
|
| 178 |
+
print(f"GET request failed for {endpoint}: {e}")
|
| 179 |
+
|
| 180 |
+
# Try POST with form data
|
| 181 |
+
try:
|
| 182 |
+
form_data = {
|
| 183 |
+
"question": question,
|
| 184 |
+
"created_at": timestamp
|
| 185 |
+
}
|
| 186 |
+
print(f"Form data: {form_data}")
|
| 187 |
+
|
| 188 |
+
response = requests.post(
|
| 189 |
+
endpoint,
|
| 190 |
+
data=form_data,
|
| 191 |
+
timeout=10
|
| 192 |
+
)
|
| 193 |
+
print(f"Form POST response status: {response.status_code}")
|
| 194 |
+
print(f"Form POST response text: {response.text[:200]}")
|
| 195 |
+
|
| 196 |
+
if response.status_code in [200, 201]:
|
| 197 |
+
print(f"Successfully saved question via form POST to {endpoint}")
|
| 198 |
+
return True
|
| 199 |
+
|
| 200 |
+
except requests.exceptions.RequestException as e:
|
| 201 |
+
print(f"Form POST request failed for {endpoint}: {e}")
|
| 202 |
|
| 203 |
+
print("All endpoints failed to save the question")
|
| 204 |
return False
|
| 205 |
|
| 206 |
except Exception as e:
|
| 207 |
+
print(f"Unexpected error saving unanswered question: {e}")
|
| 208 |
return False
|
| 209 |
|
| 210 |
def _get_timestamp(self) -> str:
|
|
|
|
| 220 |
text = text.translate(str.maketrans('', '', string.punctuation))
|
| 221 |
# Remove extra whitespace
|
| 222 |
text = ' '.join(text.split())
|
| 223 |
+
|
| 224 |
+
# Additional normalization for better matching
|
| 225 |
+
# Replace common variations
|
| 226 |
+
replacements = {
|
| 227 |
+
'what are the': 'what',
|
| 228 |
+
'what is the': 'what',
|
| 229 |
+
'what are': 'what',
|
| 230 |
+
'what is': 'what',
|
| 231 |
+
'how do i': 'how',
|
| 232 |
+
'how can i': 'how',
|
| 233 |
+
'how to': 'how',
|
| 234 |
+
'when is the': 'when',
|
| 235 |
+
'when are the': 'when',
|
| 236 |
+
'where is the': 'where',
|
| 237 |
+
'where are the': 'where',
|
| 238 |
+
'who is the': 'who',
|
| 239 |
+
'who are the': 'who'
|
| 240 |
+
}
|
| 241 |
+
|
| 242 |
+
for old, new in replacements.items():
|
| 243 |
+
if text.startswith(old):
|
| 244 |
+
text = text.replace(old, new, 1)
|
| 245 |
+
break
|
| 246 |
+
|
| 247 |
return text
|
| 248 |
|
| 249 |
def _extract_keywords(self, text: str) -> List[str]:
|
|
|
|
| 304 |
if norm2 in norm1:
|
| 305 |
contains_similarity = max(contains_similarity, 0.9 * (len(norm2) / len(norm1)))
|
| 306 |
|
| 307 |
+
# Method 4: Enhanced word order similarity
|
| 308 |
words1 = norm1.split()
|
| 309 |
words2 = norm2.split()
|
| 310 |
word_order_similarity = 0.0
|
| 311 |
if words1 and words2:
|
| 312 |
+
# Check for common word sequences (exact order)
|
| 313 |
common_sequences = 0
|
| 314 |
max_len = min(len(words1), len(words2))
|
| 315 |
for i in range(max_len):
|
| 316 |
if words1[i] == words2[i]:
|
| 317 |
common_sequences += 1
|
| 318 |
+
exact_order_similarity = common_sequences / max_len if max_len > 0 else 0.0
|
| 319 |
+
|
| 320 |
+
# Check for word order flexibility (any order)
|
| 321 |
+
set1 = set(words1)
|
| 322 |
+
set2 = set(words2)
|
| 323 |
+
common_words = set1.intersection(set2)
|
| 324 |
+
total_words = set1.union(set2)
|
| 325 |
+
flexible_order_similarity = len(common_words) / len(total_words) if total_words else 0.0
|
| 326 |
+
|
| 327 |
+
# Check for phrase patterns (like "available courses" vs "courses available")
|
| 328 |
+
phrase_similarity = self._calculate_phrase_order_similarity(words1, words2)
|
| 329 |
+
|
| 330 |
+
# Combine different word order methods
|
| 331 |
+
word_order_similarity = (
|
| 332 |
+
exact_order_similarity * 0.3 +
|
| 333 |
+
flexible_order_similarity * 0.5 +
|
| 334 |
+
phrase_similarity * 0.2
|
| 335 |
+
)
|
| 336 |
|
| 337 |
# Method 5: Semantic similarity using word relationships
|
| 338 |
semantic_similarity = self._calculate_semantic_similarity(keywords1, keywords2)
|
|
|
|
| 371 |
'contact': {'contact', 'phone', 'email', 'address', 'office', 'reach', 'call'},
|
| 372 |
'requirements': {'requirement', 'need', 'required', 'must', 'prerequisite', 'condition'},
|
| 373 |
'application': {'apply', 'application', 'submit', 'process', 'procedure'},
|
| 374 |
+
'programs': {'program', 'course', 'major', 'degree', 'study', 'academic', 'available', 'offered', 'listings'},
|
| 375 |
+
'admission': {'admission', 'admit', 'accept', 'enroll', 'entry', 'enter'},
|
| 376 |
+
'courses': {'course', 'courses', 'program', 'programs', 'major', 'majors', 'degree', 'degrees', 'available', 'offered', 'listings', 'what', 'which'}
|
| 377 |
}
|
| 378 |
|
| 379 |
# Check if keywords belong to the same semantic group
|
|
|
|
| 408 |
|
| 409 |
return min(phrase_score, 1.0)
|
| 410 |
|
| 411 |
+
def _calculate_phrase_order_similarity(self, words1: List[str], words2: List[str]) -> float:
|
| 412 |
+
"""Calculate similarity based on phrase order flexibility"""
|
| 413 |
+
if not words1 or not words2:
|
| 414 |
+
return 0.0
|
| 415 |
+
|
| 416 |
+
# Common phrase patterns that should match regardless of order
|
| 417 |
+
phrase_patterns = [
|
| 418 |
+
(['available', 'courses'], ['courses', 'available']),
|
| 419 |
+
(['admission', 'requirements'], ['requirements', 'admission']),
|
| 420 |
+
(['financial', 'aid'], ['aid', 'financial']),
|
| 421 |
+
(['tuition', 'cost'], ['cost', 'tuition']),
|
| 422 |
+
(['application', 'deadline'], ['deadline', 'application']),
|
| 423 |
+
(['contact', 'admissions'], ['admissions', 'contact']),
|
| 424 |
+
(['gpa', 'requirement'], ['requirement', 'gpa']),
|
| 425 |
+
(['academic', 'requirements'], ['requirements', 'academic']),
|
| 426 |
+
(['programs', 'available'], ['available', 'programs']),
|
| 427 |
+
(['what', 'programs'], ['programs', 'what']),
|
| 428 |
+
(['what', 'courses'], ['courses', 'what']),
|
| 429 |
+
(['what', 'available'], ['available', 'what'])
|
| 430 |
+
]
|
| 431 |
+
|
| 432 |
+
# Check for phrase pattern matches
|
| 433 |
+
for pattern1, pattern2 in phrase_patterns:
|
| 434 |
+
# Check if words1 contains pattern1 and words2 contains pattern2
|
| 435 |
+
if (all(word in words1 for word in pattern1) and
|
| 436 |
+
all(word in words2 for word in pattern2)):
|
| 437 |
+
return 0.8
|
| 438 |
+
|
| 439 |
+
# Check if words1 contains pattern2 and words2 contains pattern1
|
| 440 |
+
if (all(word in words1 for word in pattern2) and
|
| 441 |
+
all(word in words2 for word in pattern1)):
|
| 442 |
+
return 0.8
|
| 443 |
+
|
| 444 |
+
# Check for partial phrase matches
|
| 445 |
+
for pattern1, pattern2 in phrase_patterns:
|
| 446 |
+
# Check if at least 2 words from each pattern are present
|
| 447 |
+
words1_matches = sum(1 for word in pattern1 if word in words1)
|
| 448 |
+
words2_matches = sum(1 for word in pattern2 if word in words2)
|
| 449 |
+
|
| 450 |
+
if words1_matches >= 2 and words2_matches >= 2:
|
| 451 |
+
return 0.6
|
| 452 |
+
|
| 453 |
+
return 0.0
|
| 454 |
+
|
| 455 |
def _find_best_match(self, user_question: str, database_questions: List[str], threshold: float = 0.25) -> Optional[str]:
|
| 456 |
"""Find the best matching question from database with improved logic"""
|
| 457 |
if not database_questions:
|
|
|
|
| 561 |
continue
|
| 562 |
|
| 563 |
# If no answer found, save the question as unanswered
|
| 564 |
+
saved = self.save_unanswered_question(question)
|
| 565 |
+
if saved:
|
| 566 |
+
return "I'm sorry, I couldn't find a specific answer to your question in our database. I've saved your question for review, and we'll work on providing a better answer in the future. Could you try rephrasing your question or ask me something else?"
|
| 567 |
+
else:
|
| 568 |
+
return "I'm sorry, I couldn't find a specific answer to your question in our database. I tried to save your question for review, but there was an issue with our database connection. Could you try rephrasing your question or ask me something else?"
|
| 569 |
|
| 570 |
except requests.exceptions.Timeout:
|
| 571 |
# Save the question even if there's a timeout
|
| 572 |
+
saved = self.save_unanswered_question(question)
|
| 573 |
+
if saved:
|
| 574 |
+
return "I'm sorry, the database is taking too long to respond. I've saved your question for review. Please try again in a moment."
|
| 575 |
+
else:
|
| 576 |
+
return "I'm sorry, the database is taking too long to respond. Please try again in a moment."
|
| 577 |
except requests.exceptions.ConnectionError:
|
| 578 |
# Save the question even if there's a connection error
|
| 579 |
+
saved = self.save_unanswered_question(question)
|
| 580 |
+
if saved:
|
| 581 |
+
return "I'm sorry, I'm having trouble connecting to our database right now. I've saved your question for review. Please try again later."
|
| 582 |
+
else:
|
| 583 |
+
return "I'm sorry, I'm having trouble connecting to our database right now. Please try again later."
|
| 584 |
except Exception as e:
|
| 585 |
# Save the question even if there's an unexpected error
|
| 586 |
+
saved = self.save_unanswered_question(question)
|
| 587 |
+
if saved:
|
| 588 |
+
return f"I encountered an error while searching our database: {str(e)}. I've saved your question for review. Please try again."
|
| 589 |
+
else:
|
| 590 |
+
return f"I encountered an error while searching our database: {str(e)}. Please try again."
|
| 591 |
|
| 592 |
def _get_all_questions(self) -> List[str]:
|
| 593 |
"""Get all available questions from the database for smart matching"""
|