Spaces:
Paused
Paused
Update json_handling.py
Browse files- json_handling.py +8 -6
json_handling.py
CHANGED
|
@@ -35,15 +35,17 @@ import json
|
|
| 35 |
def process_whole_doc(file_path):
|
| 36 |
images = load_input(file_path)
|
| 37 |
|
| 38 |
-
complete_json =
|
|
|
|
| 39 |
|
| 40 |
# PART_NAME = "PART-1 - BILL OF ENTRY SUMMARY" # you can change per doc type
|
| 41 |
for i, image in enumerate(images, start=1):
|
| 42 |
partial_json = process_document(image)
|
| 43 |
-
print(partial_json)
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
|
|
|
| 47 |
# Ensure structure exists
|
| 48 |
# # if PART_NAME not in complete_json:
|
| 49 |
# complete_json[] = [{}]
|
|
@@ -52,7 +54,7 @@ def process_whole_doc(file_path):
|
|
| 52 |
# complete_json[PART_NAME].append({})
|
| 53 |
|
| 54 |
# ✅ Merge directly
|
| 55 |
-
complete_json[page_key]=result_data
|
| 56 |
|
| 57 |
return complete_json
|
| 58 |
|
|
|
|
| 35 |
def process_whole_doc(file_path):
|
| 36 |
images = load_input(file_path)
|
| 37 |
|
| 38 |
+
complete_json = ""
|
| 39 |
+
|
| 40 |
|
| 41 |
# PART_NAME = "PART-1 - BILL OF ENTRY SUMMARY" # you can change per doc type
|
| 42 |
for i, image in enumerate(images, start=1):
|
| 43 |
partial_json = process_document(image)
|
| 44 |
+
# print(partial_json)
|
| 45 |
+
complete_json = complete_json+partial_json
|
| 46 |
+
# # 🔹 Extract only "result"
|
| 47 |
+
# result_data = partial_json.get("result", {})
|
| 48 |
+
# page_key = f"Page {i}"
|
| 49 |
# Ensure structure exists
|
| 50 |
# # if PART_NAME not in complete_json:
|
| 51 |
# complete_json[] = [{}]
|
|
|
|
| 54 |
# complete_json[PART_NAME].append({})
|
| 55 |
|
| 56 |
# ✅ Merge directly
|
| 57 |
+
# complete_json[page_key]=result_data
|
| 58 |
|
| 59 |
return complete_json
|
| 60 |
|