cjber commited on
Commit
2d24223
·
1 Parent(s): 48506a4

fix: don't exclude some urls

Browse files

Former-commit-id: 64f3c605b5eecfa488294dc5aea3bb39b3dad5d4 [formerly 7945dcf4157c02cad61135d2a48d5bbd8b6e610d]
Former-commit-id: ab341efe17edb92f118e460b125067ceb7c1285b

Files changed (1) hide show
  1. planning_ai/preprocessing/gcpt3.py +1 -4
planning_ai/preprocessing/gcpt3.py CHANGED
@@ -77,10 +77,7 @@ def download_attachments():
77
  .rows(named=True)
78
  ):
79
  if (
80
- row["attachments_url"].startswith(
81
- ("https://egov.scambs.gov.uk", "http://egov.scambs.gov.uk")
82
- )
83
- or str(row["attachments_id"]) in existing_files
84
  or str(row["attachments_id"]) in failed_files
85
  ):
86
  failed_files.add(row["attachments_id"])
 
77
  .rows(named=True)
78
  ):
79
  if (
80
+ str(row["attachments_id"]) in existing_files
 
 
 
81
  or str(row["attachments_id"]) in failed_files
82
  ):
83
  failed_files.add(row["attachments_id"])