cakiki commited on
Commit
d9c46e5
β€’
1 Parent(s): 55b51a2

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +2 -1
app.py CHANGED
@@ -5,6 +5,7 @@ import re
5
  import string
6
 
7
  import gradio as gr
 
8
  from elasticsearch import Elasticsearch
9
  from elasticsearch_dsl import Search, Q
10
 
@@ -61,7 +62,7 @@ def search(query, num_results=10):
61
  response = phrase_query(query[1:-1], num_results=num_results)
62
  else:
63
  response = match_query(query, num_results=num_results)
64
- results = [{"text": hit.content, "repository": f"{hit.repository}/{hit.path}", "license": hit.license[0], "language": hit.language} for hit in response]
65
  return process_results(results)
66
 
67
  description = """# <p style="text-align: center;"><span style='color: #e6b800;'>StarCoder:</span> Dataset Search πŸ” </p>
 
5
  import string
6
 
7
  import gradio as gr
8
+ import nh3
9
  from elasticsearch import Elasticsearch
10
  from elasticsearch_dsl import Search, Q
11
 
 
62
  response = phrase_query(query[1:-1], num_results=num_results)
63
  else:
64
  response = match_query(query, num_results=num_results)
65
+ results = [{"text": nh3.clean(hit.content), "repository": f"{hit.repository}/{hit.path}", "license": hit.license[0], "language": hit.language} for hit in response]
66
  return process_results(results)
67
 
68
  description = """# <p style="text-align: center;"><span style='color: #e6b800;'>StarCoder:</span> Dataset Search πŸ” </p>