Spaces:
Paused
Paused
Update json_handling.py
Browse files- json_handling.py +9 -9
json_handling.py
CHANGED
|
@@ -37,23 +37,23 @@ def process_whole_doc(file_path):
|
|
| 37 |
|
| 38 |
complete_json = {}
|
| 39 |
|
| 40 |
-
PART_NAME = "PART-1 - BILL OF ENTRY SUMMARY" # you can change per doc type
|
| 41 |
|
| 42 |
-
for image in images:
|
| 43 |
partial_json = process_document(image)
|
| 44 |
-
|
| 45 |
# 🔹 Extract only "result"
|
| 46 |
result_data = partial_json.get("result", {})
|
| 47 |
-
|
| 48 |
# Ensure structure exists
|
| 49 |
-
if PART_NAME not in complete_json:
|
| 50 |
-
|
| 51 |
|
| 52 |
-
if not complete_json[PART_NAME]:
|
| 53 |
-
|
| 54 |
|
| 55 |
# ✅ Merge directly
|
| 56 |
-
complete_json[
|
| 57 |
|
| 58 |
return complete_json
|
| 59 |
|
|
|
|
| 37 |
|
| 38 |
complete_json = {}
|
| 39 |
|
| 40 |
+
# PART_NAME = "PART-1 - BILL OF ENTRY SUMMARY" # you can change per doc type
|
| 41 |
|
| 42 |
+
for i,image in images:
|
| 43 |
partial_json = process_document(image)
|
| 44 |
+
print(partial_json)
|
| 45 |
# 🔹 Extract only "result"
|
| 46 |
result_data = partial_json.get("result", {})
|
| 47 |
+
page_key = f"Page {i}"
|
| 48 |
# Ensure structure exists
|
| 49 |
+
# # if PART_NAME not in complete_json:
|
| 50 |
+
# complete_json[] = [{}]
|
| 51 |
|
| 52 |
+
# if not complete_json[PART_NAME]:
|
| 53 |
+
# complete_json[PART_NAME].append({})
|
| 54 |
|
| 55 |
# ✅ Merge directly
|
| 56 |
+
complete_json[page_key]=result_data
|
| 57 |
|
| 58 |
return complete_json
|
| 59 |
|