Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -36,6 +36,46 @@ def custom_encoder(obj: Any) -> Any:
|
|
| 36 |
# For any other type, fall back to the default encoder
|
| 37 |
return jsonable_encoder(obj)
|
| 38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
# --- END OF MODIFICATIONS ---
|
| 40 |
|
| 41 |
|
|
@@ -220,11 +260,14 @@ Output JSON:
|
|
| 220 |
# This converts numbers like 0.0000009 into a high-precision Decimal
|
| 221 |
# object inside Python, instead of a standard float.
|
| 222 |
structured_data = json.loads(json_str, parse_float=Decimal)
|
|
|
|
|
|
|
|
|
|
| 223 |
# --- END OF MODIFICATIONS ---
|
| 224 |
|
| 225 |
structured_data_cache[text_hash] = structured_data
|
| 226 |
logger.info(f"Gemini processing for {filename}, took {time.time() - start_time:.2f} seconds, {log_memory_usage()}")
|
| 227 |
-
# This will now print Decimal('0.0000009') in your console
|
| 228 |
logger.info(f"Structured Data: {structured_data}")
|
| 229 |
return structured_data
|
| 230 |
except Exception as e:
|
|
|
|
| 36 |
# For any other type, fall back to the default encoder
|
| 37 |
return jsonable_encoder(obj)
|
| 38 |
|
| 39 |
+
def convert_scientific_decimals(obj):
|
| 40 |
+
"""
|
| 41 |
+
Recursively convert any Decimal objects in scientific notation to decimal notation.
|
| 42 |
+
This handles cases where json.loads(parse_float=Decimal) creates Decimal('9E-7')
|
| 43 |
+
instead of the desired Decimal('0.0000009').
|
| 44 |
+
"""
|
| 45 |
+
if isinstance(obj, dict):
|
| 46 |
+
return {k: convert_scientific_decimals(v) for k, v in obj.items()}
|
| 47 |
+
elif isinstance(obj, list):
|
| 48 |
+
return [convert_scientific_decimals(item) for item in obj]
|
| 49 |
+
elif isinstance(obj, Decimal):
|
| 50 |
+
# Check if the decimal is in scientific notation by converting to string
|
| 51 |
+
decimal_str = str(obj)
|
| 52 |
+
if 'E' in decimal_str or 'e' in decimal_str:
|
| 53 |
+
# Convert to float first to get the actual numeric value, then back to Decimal
|
| 54 |
+
# This forces the decimal representation
|
| 55 |
+
float_val = float(obj)
|
| 56 |
+
# Use high precision string formatting to avoid float precision issues
|
| 57 |
+
if abs(float_val) < 1e-10: # Very small numbers
|
| 58 |
+
formatted = f"{float_val:.15f}"
|
| 59 |
+
elif abs(float_val) < 1e-6: # Small numbers
|
| 60 |
+
formatted = f"{float_val:.12f}"
|
| 61 |
+
elif abs(float_val) < 1: # Numbers less than 1
|
| 62 |
+
formatted = f"{float_val:.10f}"
|
| 63 |
+
else: # Larger numbers
|
| 64 |
+
formatted = f"{float_val:.8f}"
|
| 65 |
+
|
| 66 |
+
# Remove trailing zeros and decimal point if not needed
|
| 67 |
+
formatted = formatted.rstrip('0').rstrip('.')
|
| 68 |
+
if '.' not in formatted and float_val != int(float_val):
|
| 69 |
+
# If we removed all decimal places but it's not actually an integer
|
| 70 |
+
# This handles edge cases
|
| 71 |
+
formatted = f"{float_val:.1f}".rstrip('0').rstrip('.')
|
| 72 |
+
|
| 73 |
+
return Decimal(formatted)
|
| 74 |
+
else:
|
| 75 |
+
return obj
|
| 76 |
+
else:
|
| 77 |
+
return obj
|
| 78 |
+
|
| 79 |
# --- END OF MODIFICATIONS ---
|
| 80 |
|
| 81 |
|
|
|
|
| 260 |
# This converts numbers like 0.0000009 into a high-precision Decimal
|
| 261 |
# object inside Python, instead of a standard float.
|
| 262 |
structured_data = json.loads(json_str, parse_float=Decimal)
|
| 263 |
+
|
| 264 |
+
# CRITICAL: Convert any scientific notation Decimals to decimal format
|
| 265 |
+
structured_data = convert_scientific_decimals(structured_data)
|
| 266 |
# --- END OF MODIFICATIONS ---
|
| 267 |
|
| 268 |
structured_data_cache[text_hash] = structured_data
|
| 269 |
logger.info(f"Gemini processing for {filename}, took {time.time() - start_time:.2f} seconds, {log_memory_usage()}")
|
| 270 |
+
# This will now print proper decimal format like Decimal('0.0000009') in your console
|
| 271 |
logger.info(f"Structured Data: {structured_data}")
|
| 272 |
return structured_data
|
| 273 |
except Exception as e:
|