Spaces:
Paused
Paused
Update json_handling.py
Browse files- json_handling.py +32 -18
json_handling.py
CHANGED
|
@@ -34,28 +34,42 @@ import json
|
|
| 34 |
|
| 35 |
def process_whole_doc(file_path):
|
| 36 |
images = load_input(file_path)
|
| 37 |
-
|
| 38 |
-
complete_json = ""
|
| 39 |
-
|
| 40 |
|
| 41 |
-
|
| 42 |
-
|
| 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[] = [{}]
|
| 52 |
|
| 53 |
-
|
| 54 |
-
|
| 55 |
|
| 56 |
-
|
| 57 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
|
| 59 |
return complete_json
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
|
| 61 |
|
|
|
|
| 34 |
|
| 35 |
def process_whole_doc(file_path):
|
| 36 |
images = load_input(file_path)
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
+
complete_json = "["
|
| 39 |
+
first = True
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
+
for image in images:
|
| 42 |
+
partial_json = process_document(image) # returns string
|
| 43 |
|
| 44 |
+
if not first:
|
| 45 |
+
complete_json += ","
|
| 46 |
+
complete_json += partial_json
|
| 47 |
+
first = False
|
| 48 |
+
|
| 49 |
+
complete_json += "]"
|
| 50 |
|
| 51 |
return complete_json
|
| 52 |
+
# complete_json = ""
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
# # PART_NAME = "PART-1 - BILL OF ENTRY SUMMARY" # you can change per doc type
|
| 56 |
+
# for i, image in enumerate(images, start=1):
|
| 57 |
+
# partial_json = process_document(image)
|
| 58 |
+
# # print(partial_json)
|
| 59 |
+
# complete_json = complete_json+partial_json
|
| 60 |
+
# # # 🔹 Extract only "result"
|
| 61 |
+
# # result_data = partial_json.get("result", {})
|
| 62 |
+
# # page_key = f"Page {i}"
|
| 63 |
+
# # Ensure structure exists
|
| 64 |
+
# # # if PART_NAME not in complete_json:
|
| 65 |
+
# # complete_json[] = [{}]
|
| 66 |
+
|
| 67 |
+
# # if not complete_json[PART_NAME]:
|
| 68 |
+
# # complete_json[PART_NAME].append({})
|
| 69 |
+
|
| 70 |
+
# # ✅ Merge directly
|
| 71 |
+
# # complete_json[page_key]=result_data
|
| 72 |
+
|
| 73 |
+
# return complete_json
|
| 74 |
|
| 75 |
|