Spaces:
Running
Running
Create utils/document_utils.py
Browse files- utils/document_utils.py +56 -0
utils/document_utils.py
ADDED
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import logging
|
2 |
+
from typing import List
|
3 |
+
|
4 |
+
logs = []
|
5 |
+
class Document:
|
6 |
+
def __init__(self, metadata, page_content):
|
7 |
+
self.metadata = metadata
|
8 |
+
self.page_content = page_content
|
9 |
+
|
10 |
+
def apply_sentence_keys_documents(relevant_docs: List[Document]):
|
11 |
+
result = []
|
12 |
+
'''for i, doc in enumerate(relevant_docs):
|
13 |
+
doc_id = str(i)
|
14 |
+
title_passage = doc.page_content.split('\nPassage: ')
|
15 |
+
title = title_passage[0]
|
16 |
+
passages = title_passage[1].split('. ')
|
17 |
+
|
18 |
+
doc_result = []
|
19 |
+
doc_result.append([f"{doc_id}a", title])
|
20 |
+
|
21 |
+
for j, passage in enumerate(passages):
|
22 |
+
doc_result.append([f"{doc_id}{chr(98 + j)}", passage])
|
23 |
+
|
24 |
+
result.append(doc_result)'''
|
25 |
+
|
26 |
+
for relevant_doc_index, relevant_doc in enumerate(relevant_docs):
|
27 |
+
sentences = []
|
28 |
+
for sentence_index, sentence in enumerate(relevant_doc.page_content.split(".")):
|
29 |
+
sentences.append([str(relevant_doc_index)+chr(97 + sentence_index), sentence])
|
30 |
+
result.append(sentences)
|
31 |
+
|
32 |
+
return result
|
33 |
+
|
34 |
+
def apply_sentence_keys_response(input_string):
|
35 |
+
sentences = input_string.split('. ')
|
36 |
+
result = [[chr(97 + i), sentence] for i, sentence in enumerate(sentences)]
|
37 |
+
return result
|
38 |
+
|
39 |
+
def initialize_logging():
|
40 |
+
logger = logging.getLogger()
|
41 |
+
logger.setLevel(logging.INFO)
|
42 |
+
|
43 |
+
# Custom log handler to capture logs and add them to the logs list
|
44 |
+
class LogHandler(logging.Handler):
|
45 |
+
def emit(self, record):
|
46 |
+
log_entry = self.format(record)
|
47 |
+
logs.append(log_entry)
|
48 |
+
|
49 |
+
# Add custom log handler to the logger
|
50 |
+
log_handler = LogHandler()
|
51 |
+
log_handler.setFormatter(logging.Formatter('%(asctime)s - %(message)s'))
|
52 |
+
logger.addHandler(log_handler)
|
53 |
+
|
54 |
+
def get_logs():
|
55 |
+
"""Retrieve logs for display."""
|
56 |
+
return "\n".join(logs[-100:]) # Only show the last 50 logs for example
|