Upload ocr_utils.py
Browse files- ocr_utils.py +8 -2
ocr_utils.py
CHANGED
|
@@ -5,9 +5,13 @@ import os
|
|
| 5 |
# 'en' is for English language
|
| 6 |
# In ocr_utils.py
|
| 7 |
# Set model_storage_directory to a folder inside your writable app directory (/app)
|
| 8 |
-
|
|
|
|
| 9 |
os.makedirs(MODELS_DIR, exist_ok=True) # Ensure the directory exists
|
| 10 |
|
|
|
|
|
|
|
|
|
|
| 11 |
reader = easyocr.Reader(['en'], model_storage_directory=MODELS_DIR)
|
| 12 |
|
| 13 |
def extract_keywords_from_report(file_path):
|
|
@@ -62,11 +66,13 @@ def score_text_for_risk(text):
|
|
| 62 |
|
| 63 |
# Cap the score at 1.0
|
| 64 |
return min(score, 1.0), keywords_found
|
|
|
|
| 65 |
# Example usage (Optional, can be removed once integrated into app.py)
|
| 66 |
if __name__ == '__main__':
|
| 67 |
# You would test this with a sample image report
|
| 68 |
# print(extract_keywords_from_report('path/to/sample_report.jpg'))
|
| 69 |
pass
|
|
|
|
| 70 |
if __name__ == '__main__':
|
| 71 |
# ⚠️ Replace 'test_report.png' with the actual file name
|
| 72 |
test_file_path = 'test_report.png'
|
|
@@ -79,4 +85,4 @@ if __name__ == '__main__':
|
|
| 79 |
print(f"Extracted Text: {extracted_text}")
|
| 80 |
print(f"Calculated Risk Score: {risk_score}")
|
| 81 |
else:
|
| 82 |
-
print("Error: test_report.png not found. Cannot run direct test.")
|
|
|
|
| 5 |
# 'en' is for English language
|
| 6 |
# In ocr_utils.py
|
| 7 |
# Set model_storage_directory to a folder inside your writable app directory (/app)
|
| 8 |
+
# ⚠️ Use absolute path to prevent EasyOCR from defaulting to /.EasyOCR
|
| 9 |
+
MODELS_DIR = "/app/easyocr_models" # <-- Changed to absolute path
|
| 10 |
os.makedirs(MODELS_DIR, exist_ok=True) # Ensure the directory exists
|
| 11 |
|
| 12 |
+
# DEBUG: Verify the directory
|
| 13 |
+
print(f"[DEBUG] EasyOCR models directory: {MODELS_DIR}")
|
| 14 |
+
|
| 15 |
reader = easyocr.Reader(['en'], model_storage_directory=MODELS_DIR)
|
| 16 |
|
| 17 |
def extract_keywords_from_report(file_path):
|
|
|
|
| 66 |
|
| 67 |
# Cap the score at 1.0
|
| 68 |
return min(score, 1.0), keywords_found
|
| 69 |
+
|
| 70 |
# Example usage (Optional, can be removed once integrated into app.py)
|
| 71 |
if __name__ == '__main__':
|
| 72 |
# You would test this with a sample image report
|
| 73 |
# print(extract_keywords_from_report('path/to/sample_report.jpg'))
|
| 74 |
pass
|
| 75 |
+
|
| 76 |
if __name__ == '__main__':
|
| 77 |
# ⚠️ Replace 'test_report.png' with the actual file name
|
| 78 |
test_file_path = 'test_report.png'
|
|
|
|
| 85 |
print(f"Extracted Text: {extracted_text}")
|
| 86 |
print(f"Calculated Risk Score: {risk_score}")
|
| 87 |
else:
|
| 88 |
+
print("Error: test_report.png not found. Cannot run direct test.")
|