Spaces:
Running
on
T4
Running
on
T4
Update auditqa/process_chunks.py
Browse files
auditqa/process_chunks.py
CHANGED
|
@@ -69,12 +69,13 @@ def load_chunks():
|
|
| 69 |
# add metadata information
|
| 70 |
chunks_list = []
|
| 71 |
for doc in doc_processed:
|
| 72 |
-
chunks_list.append(Document(page_content=doc['content'],
|
| 73 |
metadata={"source": category,
|
| 74 |
"subtype":subtype,
|
| 75 |
"year":file[-4:],
|
| 76 |
"filename":file,
|
| 77 |
-
"page":doc['metadata']['page']
|
|
|
|
| 78 |
|
| 79 |
all_documents[category].append(chunks_list)
|
| 80 |
|
|
|
|
| 69 |
# add metadata information
|
| 70 |
chunks_list = []
|
| 71 |
for doc in doc_processed:
|
| 72 |
+
chunks_list.append(Document(page_content= doc['content'],
|
| 73 |
metadata={"source": category,
|
| 74 |
"subtype":subtype,
|
| 75 |
"year":file[-4:],
|
| 76 |
"filename":file,
|
| 77 |
+
"page":doc['metadata']['page'],
|
| 78 |
+
"headings":doc['metadata']['headings']}))
|
| 79 |
|
| 80 |
all_documents[category].append(chunks_list)
|
| 81 |
|