Spaces:
Sleeping
Sleeping
Update engine.py
Browse files
engine.py
CHANGED
|
@@ -279,8 +279,10 @@ def process_question(question):
|
|
| 279 |
global LAST_PROMPT_TYPE, LAST_SUGGESTED_DATE
|
| 280 |
|
| 281 |
q = question.strip().lower()
|
| 282 |
-
|
|
|
|
| 283 |
# Patient summary intent
|
|
|
|
| 284 |
if is_patient_summary_question(q):
|
| 285 |
return {
|
| 286 |
"status": "ok",
|
|
@@ -290,16 +292,18 @@ def process_question(question):
|
|
| 290 |
"note": None
|
| 291 |
}
|
| 292 |
|
| 293 |
-
#
|
|
|
|
|
|
|
| 294 |
if is_why_question(q) and LAST_PROMPT_TYPE == "NO_DATA":
|
|
|
|
| 295 |
return {
|
| 296 |
"status": "ok",
|
| 297 |
"message": (
|
| 298 |
-
f"I suggested **{
|
| 299 |
f"for which data exists in the system.\n\n"
|
| 300 |
-
"Your database doesn’t contain newer records
|
| 301 |
-
"
|
| 302 |
-
"You can still explore:\n"
|
| 303 |
"• Data from 2021\n"
|
| 304 |
"• Trends over time\n"
|
| 305 |
"• Patient summaries"
|
|
@@ -307,7 +311,9 @@ def process_question(question):
|
|
| 307 |
"data": []
|
| 308 |
}
|
| 309 |
|
| 310 |
-
#
|
|
|
|
|
|
|
| 311 |
if is_confirmation(q) and LAST_PROMPT_TYPE == "NO_DATA":
|
| 312 |
return {
|
| 313 |
"status": "ok",
|
|
@@ -322,9 +328,19 @@ def process_question(question):
|
|
| 322 |
"data": []
|
| 323 |
}
|
| 324 |
|
|
|
|
|
|
|
|
|
|
| 325 |
question = correct_spelling(question)
|
| 326 |
question = normalize_time_question(question)
|
| 327 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 328 |
if any(x in question for x in ["what data", "what tables", "which data"]):
|
| 329 |
return {
|
| 330 |
"status": "ok",
|
|
@@ -332,6 +348,9 @@ def process_question(question):
|
|
| 332 |
"data": []
|
| 333 |
}
|
| 334 |
|
|
|
|
|
|
|
|
|
|
| 335 |
unsupported = get_unsupported_reason(question)
|
| 336 |
if unsupported:
|
| 337 |
return {
|
|
@@ -347,6 +366,9 @@ def process_question(question):
|
|
| 347 |
"data": []
|
| 348 |
}
|
| 349 |
|
|
|
|
|
|
|
|
|
|
| 350 |
sql = call_llm(build_prompt(question))
|
| 351 |
|
| 352 |
if sql == "NOT_ANSWERABLE":
|
|
@@ -359,9 +381,13 @@ def process_question(question):
|
|
| 359 |
sql = validate_sql(sanitize_sql(sql))
|
| 360 |
cols, rows = run_query(sql)
|
| 361 |
|
|
|
|
|
|
|
|
|
|
| 362 |
if is_aggregate_only_query(sql) and not has_underlying_data(sql):
|
| 363 |
LAST_PROMPT_TYPE = "NO_DATA"
|
| 364 |
LAST_SUGGESTED_DATE = get_latest_data_date()
|
|
|
|
| 365 |
return {
|
| 366 |
"status": "ok",
|
| 367 |
"message": friendly("No data is available for that time period."),
|
|
@@ -370,9 +396,13 @@ def process_question(question):
|
|
| 370 |
"sql": None
|
| 371 |
}
|
| 372 |
|
|
|
|
|
|
|
|
|
|
| 373 |
if not rows:
|
| 374 |
LAST_PROMPT_TYPE = "NO_DATA"
|
| 375 |
LAST_SUGGESTED_DATE = get_latest_data_date()
|
|
|
|
| 376 |
return {
|
| 377 |
"status": "ok",
|
| 378 |
"message": friendly("No records found."),
|
|
@@ -380,12 +410,13 @@ def process_question(question):
|
|
| 380 |
"data": []
|
| 381 |
}
|
| 382 |
|
| 383 |
-
|
| 384 |
-
|
| 385 |
-
|
| 386 |
return {
|
| 387 |
"status": "ok",
|
| 388 |
"sql": sql,
|
| 389 |
"columns": cols,
|
| 390 |
"data": rows
|
| 391 |
}
|
|
|
|
|
|
| 279 |
global LAST_PROMPT_TYPE, LAST_SUGGESTED_DATE
|
| 280 |
|
| 281 |
q = question.strip().lower()
|
| 282 |
+
|
| 283 |
+
# -------------------------------
|
| 284 |
# Patient summary intent
|
| 285 |
+
# -------------------------------
|
| 286 |
if is_patient_summary_question(q):
|
| 287 |
return {
|
| 288 |
"status": "ok",
|
|
|
|
| 292 |
"note": None
|
| 293 |
}
|
| 294 |
|
| 295 |
+
# -------------------------------
|
| 296 |
+
# WHY follow-up handling
|
| 297 |
+
# -------------------------------
|
| 298 |
if is_why_question(q) and LAST_PROMPT_TYPE == "NO_DATA":
|
| 299 |
+
year = LAST_SUGGESTED_DATE[:4] if LAST_SUGGESTED_DATE else "the latest available year"
|
| 300 |
return {
|
| 301 |
"status": "ok",
|
| 302 |
"message": (
|
| 303 |
+
f"I suggested **{year}** because that’s the most recent year "
|
| 304 |
f"for which data exists in the system.\n\n"
|
| 305 |
+
"Your database doesn’t contain newer records yet.\n\n"
|
| 306 |
+
"You can explore:\n"
|
|
|
|
| 307 |
"• Data from 2021\n"
|
| 308 |
"• Trends over time\n"
|
| 309 |
"• Patient summaries"
|
|
|
|
| 311 |
"data": []
|
| 312 |
}
|
| 313 |
|
| 314 |
+
# -------------------------------
|
| 315 |
+
# YES / confirmation handling
|
| 316 |
+
# -------------------------------
|
| 317 |
if is_confirmation(q) and LAST_PROMPT_TYPE == "NO_DATA":
|
| 318 |
return {
|
| 319 |
"status": "ok",
|
|
|
|
| 328 |
"data": []
|
| 329 |
}
|
| 330 |
|
| 331 |
+
# -------------------------------
|
| 332 |
+
# Normalize question
|
| 333 |
+
# -------------------------------
|
| 334 |
question = correct_spelling(question)
|
| 335 |
question = normalize_time_question(question)
|
| 336 |
|
| 337 |
+
# Reset state once user asks a fresh question
|
| 338 |
+
LAST_PROMPT_TYPE = None
|
| 339 |
+
LAST_SUGGESTED_DATE = None
|
| 340 |
+
|
| 341 |
+
# -------------------------------
|
| 342 |
+
# Metadata queries
|
| 343 |
+
# -------------------------------
|
| 344 |
if any(x in question for x in ["what data", "what tables", "which data"]):
|
| 345 |
return {
|
| 346 |
"status": "ok",
|
|
|
|
| 348 |
"data": []
|
| 349 |
}
|
| 350 |
|
| 351 |
+
# -------------------------------
|
| 352 |
+
# Unsupported questions
|
| 353 |
+
# -------------------------------
|
| 354 |
unsupported = get_unsupported_reason(question)
|
| 355 |
if unsupported:
|
| 356 |
return {
|
|
|
|
| 366 |
"data": []
|
| 367 |
}
|
| 368 |
|
| 369 |
+
# -------------------------------
|
| 370 |
+
# LLM → SQL
|
| 371 |
+
# -------------------------------
|
| 372 |
sql = call_llm(build_prompt(question))
|
| 373 |
|
| 374 |
if sql == "NOT_ANSWERABLE":
|
|
|
|
| 381 |
sql = validate_sql(sanitize_sql(sql))
|
| 382 |
cols, rows = run_query(sql)
|
| 383 |
|
| 384 |
+
# -------------------------------
|
| 385 |
+
# No data (aggregate case)
|
| 386 |
+
# -------------------------------
|
| 387 |
if is_aggregate_only_query(sql) and not has_underlying_data(sql):
|
| 388 |
LAST_PROMPT_TYPE = "NO_DATA"
|
| 389 |
LAST_SUGGESTED_DATE = get_latest_data_date()
|
| 390 |
+
|
| 391 |
return {
|
| 392 |
"status": "ok",
|
| 393 |
"message": friendly("No data is available for that time period."),
|
|
|
|
| 396 |
"sql": None
|
| 397 |
}
|
| 398 |
|
| 399 |
+
# -------------------------------
|
| 400 |
+
# Empty result set
|
| 401 |
+
# -------------------------------
|
| 402 |
if not rows:
|
| 403 |
LAST_PROMPT_TYPE = "NO_DATA"
|
| 404 |
LAST_SUGGESTED_DATE = get_latest_data_date()
|
| 405 |
+
|
| 406 |
return {
|
| 407 |
"status": "ok",
|
| 408 |
"message": friendly("No records found."),
|
|
|
|
| 410 |
"data": []
|
| 411 |
}
|
| 412 |
|
| 413 |
+
# -------------------------------
|
| 414 |
+
# Successful response
|
| 415 |
+
# -------------------------------
|
| 416 |
return {
|
| 417 |
"status": "ok",
|
| 418 |
"sql": sql,
|
| 419 |
"columns": cols,
|
| 420 |
"data": rows
|
| 421 |
}
|
| 422 |
+
|