Upload api.py
Browse files
api.py
CHANGED
|
@@ -364,11 +364,14 @@ async def correct_text(request: CorrectionRequest):
|
|
| 364 |
if len(request.text) > 100000:
|
| 365 |
raise HTTPException(status_code=400, detail="Text too long (max 100000 characters)")
|
| 366 |
|
|
|
|
|
|
|
| 367 |
# Process text
|
| 368 |
corrected = process_text(request.text)
|
| 369 |
|
| 370 |
# Calculate processing time
|
| 371 |
processing_time = (time.time() - start_time) * 1000
|
|
|
|
| 372 |
|
| 373 |
# Include timing if requested
|
| 374 |
response = CorrectionResponse(
|
|
@@ -401,6 +404,8 @@ async def correct_batch(request: BatchCorrectionRequest):
|
|
| 401 |
if not request.texts:
|
| 402 |
raise HTTPException(status_code=400, detail="No texts provided")
|
| 403 |
|
|
|
|
|
|
|
| 404 |
# Validate text lengths
|
| 405 |
validated_texts = []
|
| 406 |
for text in request.texts:
|
|
@@ -411,10 +416,18 @@ async def correct_batch(request: BatchCorrectionRequest):
|
|
| 411 |
|
| 412 |
# Process all texts in batch (GPU efficient!)
|
| 413 |
# Step 1: Grammar correction (batched)
|
|
|
|
|
|
|
| 414 |
gec_corrected_texts = apply_gec_correction_batch(validated_texts)
|
|
|
|
|
|
|
| 415 |
|
| 416 |
# Step 2: Punctuation and capitalization (batched)
|
|
|
|
|
|
|
| 417 |
final_texts = apply_punctuation_batch(gec_corrected_texts)
|
|
|
|
|
|
|
| 418 |
|
| 419 |
# Mark texts that were too long
|
| 420 |
corrected_texts = []
|
|
@@ -427,6 +440,8 @@ async def correct_batch(request: BatchCorrectionRequest):
|
|
| 427 |
# Calculate processing time
|
| 428 |
processing_time = (time.time() - start_time) * 1000
|
| 429 |
|
|
|
|
|
|
|
| 430 |
response = BatchCorrectionResponse(
|
| 431 |
success=True,
|
| 432 |
corrected_texts=corrected_texts
|
|
|
|
| 364 |
if len(request.text) > 100000:
|
| 365 |
raise HTTPException(status_code=400, detail="Text too long (max 100000 characters)")
|
| 366 |
|
| 367 |
+
logger.info(f"π Single text request received ({len(request.text)} chars)")
|
| 368 |
+
|
| 369 |
# Process text
|
| 370 |
corrected = process_text(request.text)
|
| 371 |
|
| 372 |
# Calculate processing time
|
| 373 |
processing_time = (time.time() - start_time) * 1000
|
| 374 |
+
logger.info(f"β
Completed in {processing_time:.1f}ms")
|
| 375 |
|
| 376 |
# Include timing if requested
|
| 377 |
response = CorrectionResponse(
|
|
|
|
| 404 |
if not request.texts:
|
| 405 |
raise HTTPException(status_code=400, detail="No texts provided")
|
| 406 |
|
| 407 |
+
logger.info(f"π¦ Batch request received: {len(request.texts)} texts")
|
| 408 |
+
|
| 409 |
# Validate text lengths
|
| 410 |
validated_texts = []
|
| 411 |
for text in request.texts:
|
|
|
|
| 416 |
|
| 417 |
# Process all texts in batch (GPU efficient!)
|
| 418 |
# Step 1: Grammar correction (batched)
|
| 419 |
+
logger.info(f"π§ Starting GEC batch processing ({len(validated_texts)} texts)...")
|
| 420 |
+
gec_start = time.time()
|
| 421 |
gec_corrected_texts = apply_gec_correction_batch(validated_texts)
|
| 422 |
+
gec_time = (time.time() - gec_start) * 1000
|
| 423 |
+
logger.info(f"β GEC completed in {gec_time:.1f}ms")
|
| 424 |
|
| 425 |
# Step 2: Punctuation and capitalization (batched)
|
| 426 |
+
logger.info(f"π Starting punctuation batch processing...")
|
| 427 |
+
punct_start = time.time()
|
| 428 |
final_texts = apply_punctuation_batch(gec_corrected_texts)
|
| 429 |
+
punct_time = (time.time() - punct_start) * 1000
|
| 430 |
+
logger.info(f"β Punctuation completed in {punct_time:.1f}ms")
|
| 431 |
|
| 432 |
# Mark texts that were too long
|
| 433 |
corrected_texts = []
|
|
|
|
| 440 |
# Calculate processing time
|
| 441 |
processing_time = (time.time() - start_time) * 1000
|
| 442 |
|
| 443 |
+
logger.info(f"β
Batch completed: {len(corrected_texts)} texts in {processing_time:.1f}ms (avg {processing_time/len(corrected_texts):.1f}ms/text)")
|
| 444 |
+
|
| 445 |
response = BatchCorrectionResponse(
|
| 446 |
success=True,
|
| 447 |
corrected_texts=corrected_texts
|